Authors:
Nicoletta Cera^1,2, Joana Pinto^1,3, Minghao Dong^4,5, Steven Durning^6,7, Janniko R. Georgiadis^8,9

Affiliations:
1. Faculty of Psychology and Education Sciences, University of Porto, 4200-135 Porto, Portugal
2. Research Unit in Medical Imaging and Radiotherapy, Cross I&D Lisbon Research Centre, Escola Superior de Saúde da Cruz Vermelha Portuguesa, 1300-125 Lisbon, Portugal
3. Faculty of Medicine, University of Porto, 4200-319 Porto, Portugal
4. Engineering Research Center of Molecular and Neuro Imaging of the Ministry of Education, School of Life Science and Technology, Xidian University, No. 2 South Taibai Road, Xi’an 710071, China
5. Xian Key Laboratory of Intelligent Sensing and Regulation of trans-Scale Life Information, Xidian University, 710071 Xi’an, China
6. Department of Health Professions Education, Uniformed Services University of the Health Sciences, Bethesda, MD, USA
7. Department of Internal Medicine, Uniformed Services University of the Health Sciences, Bethesda, MD, USA
8. Department of Biomedical Sciences, University of Groningen, University Medical Centre Groningen (UMCG), Antonius Deusinglaan 1, 9713 AV, Groningen, The Netherlands
9. School of Medicine and Health Sciences, Carl von Ossietzky University Oldenburg, Oldenburg, Germany

Correspondence:
Nicoletta Cera — cera.nicoletta@gmail.com
Janniko R. Georgiadis — j.r.georgiadis@umcg.nl

Abstract

Healthcare systems require the efficient development of expert performance. Several studies have explored the cognitive foundations of medical expert performance, especially in radiology. Studying at the brain level could provide further insight into specific mechanisms mediating medical expert performance. Researchers have recently begun to systematically employ neuroimaging in this field. Most studies focus on specific specializations rather than identifying shared neural substrates across disciplines. This systematic review and activation likelihood estimation (ALE) meta-analysis followed the PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) guidelines. A total of 297 studies examining neural correlates were identified by comparing expert and novice medical performance. After screening, 22 studies were included in the final analysis. For studies reporting three-dimensional coordinates, ALE meta-analysis revealed consistent involvement of the medial frontal lobe, including the superior frontal gyrus, dorsomedial and ventromedial prefrontal cortex, and inferior frontal and fusiform gyri. Radiology-specific analyses highlighted activation in the ventromedial prefrontal cortex, the left pre-supplementary motor area (pre-SMA), along with the fusiform and opercular inferior frontal gyri. Internal medicine-based studies highlighted involvement of the SMA, inferior frontal gyrus, and dorsomedial prefrontal cortex. Our results revealed involvement, at different levels, of the medial frontal cortex, including the SMA and superior and inferior frontal gyri, which is part of the network relevant for inhibitory control and decision-making. The development of decision-making during the diagnostic process is relevant for the training of future professionals.

Keywords: Medical expert performance, radiology, learning, fMRI, SMA

Introduction

Physicians currently undergo over a decade of rigorous education, moving from the classroom to clinical clerkships and specialization, gradually building expertise through patient interactions and hands-on experience (Silvennoinen et al., 2009). Given the expense, time, and labor involved in health profession education, optimizing its duration and content is critical. What physicians need to know or need to do really well, i.e. what is their expertise, is, however, becoming increasingly difficult to predict. For instance, precision medicine (e.g. in oncology) demands deep, specialized knowledge of biomedical, technological, and clinical concepts within a discipline to tailor treatments to a specific patient (Wang & Wang, 2023; Grauman et al., 2023). On the other hand, multimorbidity challenges require a holistic approach in which experts integrate knowledge across overlapping clinical disciplines to meet patient needs (Pearson-Stuttard et al., 2019). At the same time, artificial intelligence (AI) applications impact the way medical students interact with knowledge (Moskovich and Rozani, 2025) and, in fact, may already be reshaping the concept of medical expertise, shifting it toward the ability to leverage this technology in daily clinical practice (see, e.g., Khosravi et al., 2024). In the dimension of longitudinal development, there is the question of when medical performance is sufficient, i.e. what level of performance is required in what situation to successfully execute the profession. One can argue that all of these dimensionalities live within the same expert domain that is of a predominantly cognitive nature. Medical experts should adequately address their patients’ needs, and this involves at least some logical assessment of their condition or situation, which could represent a cross-cutting cognitive mechanism that is cardinal to the development of performance.

Medical expert performance looks smooth, effortless, and almost automatic (Ericsson, 2006). For instance, an expert radiologist examining a thoracic X-ray may spot a case of pneumonia in a split second (Nakashima et al., 2016; Bertram et al., 2016), and an expert neurologist may provide a fast and accurate diagnosis of herpes simplex encephalitis when presented only with a written patient description (van den Berg et al., 2020). Making a diagnosis implies several cognitive functions, such as working memory, focused attention, memory recall, and decision-making ( Wimmers et al., 2005; Janelle & Hatfield, 2008; Hruska et al., 2016a; ten Cate & Durning, 2018); nevertheless, this complex procedure seems to be automated by expert clinicians. We assume that this is the result of many years of experience and training, but what are the key mechanisms involved in achieving this level of performance? While training duration, early-age training onset, and experience in general are considered relevant variables for assessing the level of expert performance (Ericsson, 2004), extensive experience does not invariably lead to high levels of performance in professionals (Ericsson, 2004; 2006). Moreover, expert performance in different domains—perceptual, cognitive, and motor (the so-called primary expertise domains) (Bilalić, 2017)—may require different developmental pathways. Yet, there is an overlap in how different medical professionals are trained. Implicit knowledge, “know-how,” is usually acquired with practical experience and learning from others (Ritchie et al., 2020). The predominant method in surgical education, “see one-do one-teach one” (Silvennoinen et al., 2009), also applies to developing radiological skills, where visual and attentional strategies for recognizing abnormalities in X-ray films are learned without explicit instruction. Practices such as this may support our assertion that medical expert performance at large may build on the development of generic cognitive mechanisms (Harris & Bacon, 2019).

Neuroimaging techniques can reveal brain activity patterns associated with specific cognitive processes, thereby advancing our understanding of the underlying mechanisms of, e.g., decision-making and memory (e.g. Logothetis, 2008; Poldrack et al., 2012). Until quite recently, neural-level information about medical expert performance had to be inferred from brain imaging studies in non-medical domains, such as chess, sport, and music (Amunts et al., 1997; Leff et al ., 2008; Righi et al., 2013; Kim et al., 2014; Wu et al., 2024). Only in the last decade have researchers started to use brain imaging techniques to study expert performance in medical disciplines systematically. Such studies generally investigate medical expert performance within a specific medical specialization, i.e. related to a specific content and context. As a result, it remains unclear whether there is a common neural substrate for medical expert performance across disciplines. Our study seeks to explore this question.

We used activation likelihood estimation (ALE) meta-analysis, which is a widely used analytical technique based on random effects inference that accounts for sample size and is designed to identify statistically significant spatial convergence among activation foci reported in previously published neuroimaging studies (Turkeltaub et al., 2012). ALE specifically models the likelihood of activation across studies to determine where consistent patterns emerge in brain activation (Eickhoff et al., 2012). While it shares with other meta-analyses the advantage of mitigating study-specific noise, introduced by variations in methodology, design, or sampling, ALE’s unique contribution lies in its ability to localize convergence at the voxel level. This makes it particularly suitable for identifying brain regions that are consistently involved in medical expert performance, potentially revealing neural mechanisms that are stable across different experimental contexts. In turn, this can inform more targeted hypotheses about the development and nature of medical expertise.

Methods

To perform the present systematic review, we followed the approach recommended by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Moher et al., 2009). Moreover, the PICO strategy was followed to perform the electronic search in the principal databases (Scopus, Web of Science, and PubMed), combining terms related to the purpose of this systematic review, such as “expertise,” “clinical,” “fMRI,” “MRI,” and “medicine.”

We limited the search to items that investigated the effect of expert performance (O) on the structural and functional (C) brain differences between medical expert and novice performance in professionals or students (P), or after the effect of specific internship or training, using brain imaging techniques, such as functional magnetic resonance imaging (fMRI) (I) (Eriksen & Frandsen, 2018). The search terms have been combined using the Boolean operators “AND” and “OR” as follows: medical: “medic”[All Fields] OR “medical”[All Fields] OR “medicalization”[MeSH Terms] OR “medicalization”[All Fields] OR “medicalizations”[All Fields] OR “medicalize”[All Fields] OR “medicalized”[All Fields] OR “medicalizes”[All Fields] OR “medicalizing”[All Fields] OR “medically”[All Fields] OR “medicals”[All Fields] OR “medicated”[All Fields] OR “medication’s”[All Fields] OR “medics”[All Fields] OR “pharmaceutical preparations”[MeSH Terms]; expertise: “expertise”[All Fields] OR “expertises”[All Fields] OR “expertising”[All Fields]; fMRI: “magnetic resonance imaging”[MeSH Terms] OR (“magnetic”[All Fields] AND “resonance”[All Fields] AND “imaging”[All Fields]) OR “magnetic resonance imaging”[All Fields] OR “fmri”[All Fields]; brain: “brain”[MeSH Terms] OR “brain”[All Fields] OR “brains”[All Fields] OR “brain’s”[All Fields].

The inclusion and exclusion criteria were based on the topic, study design, intervention, and population (Table 1). Therefore, all the selected studies were published in English, and the search was performed without time limits.

Table 1:

Inclusion and exclusion criteria.

Inclusion criteria	Exclusion criteria
Design
Experimental studies	Other designs
Longitudinal studies	Systematic reviews
	Meta-analyses
Population
Adults	Children
Medical professionals with specific specialties	Adolescents
Medical students	Elderly
Medical residents	Patients with neuropsychiatric disorders
Intervention
Magnetic resonance imaging	Other neuroimaging techniques
Functional magnetic resonance imaging	Psychophysiological techniques
Tasks to assess expertise with different modalities
Resting state
Topic
Brain activations	Brain activations, functional connectivity, gray and white matter assessment in different populations
Functional connectivity
Gray and white matter assessment

Initially, all the titles, abstracts, and keywords were screened to find a match with the inclusion criteria (Table 1). Then, all potential hits of interest were retrieved and screened to identify relevant references in the reference list. Moreover, narrative and systematic reviews and meta-analyses were also retrieved, and the reference lists were screened. After we removed duplicates, we checked the titles and abstracts to determine whether they matched the inclusion or exclusion criteria. We subsequently retrieved and read the full texts to confirm their eligibility. The authors independently performed the literature search, screening, and methodological assessment. When a consensus on the results was reached, the articles were included. The quality of the studies was assessed by applying the Newcastle–Ottawa scale (NOS—Wells et al., 2000). The NOS scale is used to assess the quality and risk of bias in case-control, cross-sectional, and longitudinal studies, assessing selection, comparability, and exposure. The NOS uses a “star” system, and a score of 7–9 stars indicates a low risk of bias, a score of 4–6 stars indicates an unclear risk of bias, and a score of 3 or fewer stars indicates a high risk of bias. Following the guidelines mentioned above, information about the characteristics was extracted from each included study (Table S1).

Technical information, such as magnetic field power, MRI acquisition technique (structural or functional, i.e. “fMRI”), and analytical technique used, was extracted from each included article, together with the 3D MRI coordinates reported in the full texts (Supplementary Materials—Excel file).

In the first stage, information about the total sample size, size of each group (i.e. expert and novice groups), age (mean and standard deviation), number of women included, and performance on the different tasks, when reported, was extracted. Then, a series of meta-analyses were calculated using the package MAJOR implemented in Jamovi (v.2.3.28; The Jamovi project, https://www.jamovi.org). All the data analyses were carried out using the Q-test and I² statistic, where a value of >50% was considered to represent statistical heterogeneity, Z, P-value, 95% confidence interval (CI), and tau-square (Higgins et al., 2003). The effect size (Cohen’s d) of the results of the behavioral tasks was extracted from each study when present or calculated from the statistical results reported.

GingerAle 3.02 was used to perform an ALE (https://www.brainmap.org/ale/) (Turkeltaub et al., 2012) on the 3D coordinates reported in the included studies. Then, all the coordinates reported in the Talairach stereotaxic space were converted into Montreal Neurological Institute (MNI) space using GingerALE. We decided to report the coordinates from the contrast analyses or those for each stimulus category or group. ALE is an analytical technique widely used to calculate fMRI coordinate-based meta-analyses by determining foci convergence, as reported in previous studies. To create modeled activation maps for each experiment, the ALE algorithm was implemented to calculate the maximum probability of activation. Following this, the union of these maps was computed on a voxel-by-voxel basis, taking into account the sample size of each study and comparing the findings with a null distribution map.

Permutation was applied to compute the reliability of an ALE map, allowing for the determination of the difference between the true activation foci convergence and random clustering (Eickhoff et al., 2012).

To assess the brain correlates consistently associated with expert medical performance across medical domains, an ALE map with all the coordinates in MNI space from the selected studies was calculated and tested against the null map. The resulting ALE images were converted to Z scores to simplify the interpretation and show their significance.

ALE maps were overlaid and displayed using Mango software (http://ric.uthscsa.edu/mango/) on an MNI template. To extract anatomical labels and their corresponding Brodmann areas, we utilized the AAL (Anatomical Automatic Labeling) atlas and Brodmann atlases, which are available in the MRIcron software (https://www.nitrc.org/projects/mricron) for Windows. All the anatomical labels were verified by an expert neuroanatomist (J.G.).

Results

Characteristics of the included studies and the random effects meta-analysis results

After applying the inclusion and exclusion criteria, 22 studies were included in the present systematic review (Fig. 1– Haller & Radue, 2005; Harley et al., 2009; Melo et al., 2011; Dong et al., 2013, 2014, 2015, 2022; Durning et al., 2012, 2015, 2016; Morris et al., 2015; Bilalić et al., 2016; Hruska et al., 2016a, 2016b; Kok et al., 2018, 2021; Ouellette et al., 2020; van den Berg et al., 2020; Su et al., 2022; ; Wang et al., 2021, 2023; Zhang et al., 2022). Table 2 shows the principal characteristics of the included studies. Since expert performance was the principal variable, 336 experts, 20 intermediates, and 284 non-experts were recruited as reported in the included studies. The studies included 640 (241 women) adult participants (age range 21.2–51.6 years), with mean ages of 33.1 and 27.1 years, respectively, for participants showing expert performance and those who did not.

Figure 1:

Flowchart of the selection process.

Table 2:

Principal results displayed in the included studies.

Source	Participants	Magnetic field/task/analytical technique	Results
Haller & Radue, 2005	N = 24 K = 2 (1) 12 radiologists (10 males), age = 35.8 ± 3.6 years (2) 12 non-radiologists (8 males), age = 33.0 ± 6.9 years	1.5T fMRI Task: radiologic images vs electron microscope images (each condition can be manipulated vs non-manipulated) Decision: original vs manipulated GLM	Radiologists > non-radiologist: strongest activations in the bilateral, left-dominant, and posterior superior and inferior parietal lobule
Harley et al., 2009	N = 21 K = 3 (1) 7 practicing thoracic radiologists (4 males), age = 51.6 years (10 years’ experience post-residency; mean = 18.9 years) (2) 7 fourth-year radiology residents (4 males), age = 30.9 years 3) 7 first-year radiology residents (5 males), age = 28.6 years	3T fMRI Task = normal vs abnormal chest radiographs (scrambled vs non-scrambled) GLM; correlation analysis	Activity in the right fusiform face area (FFA) correlated with visual expertise. In contrast, activity in the left lateral occipital (LO) correlated negatively with expertise, and the amount of LO that responded to radiographs was smaller in experts than in novices. Activity in the FFA and LO correlated negatively in the experts
Melo et al., 2011	N = 25 K = 1 (1) 25 radiologists (16 males), age = 35.9 years (experience = 11.6 years)	3T fMRI Task: radiographs containing lesions, animals, and letters. GLM	Lesion > animal: activation was significantly greater in the left inferior frontal sulcus and posterior cingulate cortex
Durning et al., 2012	N = 17 K = 1 Board-certified internists in internal medicine (15 males), age = 39.5 ± 7 years	3T fMRI Task: think aloud Multiple-choice questions from ABIM and National Board of Medical Examiners (NBME) (16 easy and 16 hard questions) Guessing vs no guessing (press a button vs voice response) GLM	Guessing vs no-guessing during answering: bilateral precuneus and left mid-temporal gyrus (P < 0.01)
Dong et al., 2013	N = 44 K = 2 (1) 22 Professional acupuncturists (12 males), age = 28.7 ± 1.4 years (5 years of training) (2) 22 NC (12 males), age = 28.2 ± 2.3 years	3T MRI Behavioral measure: (1) test on tactile discrimination ability; (2) test on fine motor skills; (3) test on emotion regulation proficiency Voxel-based morphometry Regression analysis	Larger gray matter volumes in acupuncturists in the hand representation of the contralateral primary somatosensory cortex (SI), right lobule V/VI, and bilateral ventral anterior cingulate cortex/ventral medial prefrontal cortex. Gray matter volumes of the SI and lobule V/VI positively correlated with the duration of acupuncture practice
Morris et al., 2015	N = 9 K = 3 (1) Novices (n = 3) (2) Intermediates (n = 3) (3) Experts (n = 3) Based on the surgical experience. Age = NA	3T fMRI 10 times repetition: finger tapping (control), rest period (null event), “pick” up shoelace (null event), knot tying (action), rest period, imagining tying a knot (perception) GLM	No difference between “intermediates vs experts”. Experts > novices: right postcentral gyrus for finger tapping task Experts > novices: left middle occipital for the imagining task
Dong et al., 2014	N = 32 K = 2 (1) 16 professional acupuncturists (8 males), age = 28 ± 1.6 years (2) 16 NC (8 males), age = matched with (1); NA	3T fMRI Behavioral measure: (1) test on tactile discrimination ability; (2) test on fine motor skills; (3) test on emotion regulation proficiency Resting-state ReHo	Acupuncturists > NC in tactile-motor and emotional regulation domain and showed increased coherence in local BOLD signal fluctuations in the left primary motor cortex (MI), left primary somatosensory cortex (SI), and left ventral medial prefrontal cortex/orbitofrontal cortex (VMPFC/OFC)
Dong et al., 2015	N = 46 K = 2 (1) 23 acupuncturists (10 males), age = 27.5 ± 1.7 years Experience = 65.5 ± 12.7 months (2) 23 Controls (10 males), age = 27.1 ± 1.5 years No experience	3T fMRI Behavioral measures: test on tactile discrimination ability; test on emotion regulation proficiency; pre-fMRI task with pictures of body parts [painful (acupuncture needle) and neutral condition)] Resting-state ALFF	Behavioral results: acupuncturists have better tactile discrimination; no between-group differences for emotion regulation; unpleasantness ratings were significantly lower in the acupuncturist group. Brain results: significantly higher ALFF values for acupuncturists in vmPFC and somatosensory cortex I (SI) in correspondence with the hand representation
Durning et al., 2015	N = 27 K = 2 (1) 17 experts: Board-certified internists in internal medicine (15 males), age = 39.5 ± 7 years (2) 10 novices: internal medicine interns with faculty appointments at the Uniformed Services University (7 males), age = 29.6 ± 2 years	3T fMRI Task: think aloud Multiple-choice questions from ABIM and National Board of Medical Examiners (NBME) (16 easy and 16 hard questions) Guessing vs no guessing (press a button vs voice response) GLM Conjunction analysis	Expert vs novices: experts: fewer activations in the right postcentral gyrus, bilateral DLPFC, DMPFC, bilateral ventromedial prefrontal cortex, bilateral lateral and medial OFC, ventral ACC, and dorsal ACC. Significantly greater activation in experts > novices: rostrolateral prefrontal cortex and cuneus.
Durning et al., 2016	N = 17 K = 1 (1) Board-certified internists in internal medicine (15 males), age = 39.5 ± 7 years	3T fMRI Behavioral measures: diagnostic thinking inventory questionnaire (DTI—41 items), 2 subscales: (1) flexibility in thinking; (2) structure in memory Task: multiple-choice questions Covariate analysis	Flexibility in thinking: bilateral vmPFC and parahippocampal gyrus Structure in memory: left dlPFC, vmPFC, and IPL DTI total score: left dlPFC, vmPFC, and dmPFC All the covariates were applied to answer > reading contrast in the multiple-choice questions task
Hruska et al., 2016b	N = 20 K = 2 (1) 10 gastroenterologists (5 males), age = 39.5 ± 4.5 years (2) 10 novices: second-year medical students (8 males), age = 26.5 ± 5.3 years	3T fMRI Task: written clinical cases classified as easy or hard (4 answer choices) GLM	Increased activation in the prefrontal cortex in novices for both easy and hard clinical cases suggests novices utilize WM more so than experts during clinical reasoning. We found that clinicians’ level of expertise elicited differential activation of regions of the human prefrontal cortex associated with WM during clinical reasoning
Bilalić et al., 2016	N = 31 K = 2 (1) 16 radiologists (10 males), age = 35.2 ± 4.3 years 2) 15 medical students (9 males), age = 28.1 ± 4.9 years	3T fMRI Task: 1-back with 5 classes of stimuli (faces, tools, rooms, upright X-ray, and inverted X-ray)—8 blocks for each condition GLM and MVPA	FFA: distinguish between X-rays and other stimuli by MPVA Radiologist > students in the sensitivity to X-rays Radiologists: FFA activations obtained on faces to differentiate X-ray stimuli from other stimuli The overlap in the FFA activation is not based on the visual similarity of faces and X-rays but rather on the processes necessary for expertise
Hruska et al., 2016a	N = 20 K = 2 (1) 10 Novices: 2nd year medical students (8 males), age = 26.5 ± 5.3 years (2) 10 experts: practicing gastroenterologists (5 males), age = 39.5 ± 4.5 years	3T fMRI Task: 16 written clinical cases classified as easy or hard (4 answer choices) GLM	Expert clinicians had greater activations than novice clinicians in the right dorsal lateral, right ventral lateral, and right parietal cortex
Kok et al., 2018	N = 30 K = 3 (1) 10 novices: 3rd year medical students (2 males), age = 21.2 ± 0.9 years (2) 10 intermediates: orthopedic-surgery residents (7 males), age = 29 ± 2.5 years (3) 10 experts: orthopedic surgery specialists (9 male), age = 46.9 ± 10.5 years	3T fMRI Task: action view (surgical procedure vs daily life activities). Participants were asked to reproduce the hand movement sequence GLM	Parietal and post-central regions of the mirror neuron system were found to be related to expertise
Ouellette et al., 2020	N = 29 K = 2 (1) 12 experts: attending radiologists, radiology fellows, and 4th year radiology residents (11 males), age = 38.2 ± 10.3 years (2) 17 non-radiologists: non-radiologist physicians and non-imaging researchers with a PhD or MD (15 males), age = 30.6 ± 5.5years	3T fMRI, DTI, cortical thickness Task: radiological imaging assessment with a diagnostic solution. GLM	Radiologists showed overall lower task-related brain activation than non-radiologists, in the left lateral occipital cortex, left superior parietal lobule, occipital pole, right superior frontal and precentral gyri, lingual gyrus, and left intraparietal sulcus. No structural differences
van den Berg et al., 2020	N = 16 K = 1 (1) Neurologists with 10 years of experience	3T fMRI Task: prototypical clinical cases vs ambiguous clinical cases (written sentences). The task was based on reading of the sentences, and reasoning, Type 1 and Type 2 cases FC (ROI to ROI)	Compared with reading control sentences, diagnosing cases resulted in increased activation in the caudate nucleus and frontal and parietal regions FC: ambiguous cases showed stronger connectivity between regions in the frontal, parietal, and temporal cortex, in addition to the cerebellum
Kok et al., 2021	N = 28 K = 2 (1) 17 radiologists (7 males), age = 29.6 ± 3.5 years (2) 11 non-radiologists (2 males), age = 28.4 ± 6.2 years	3T fMRI Task: faces vs objects to localize FFA N = 66 radiographs with at least 1 abnormality. Fast presentation vs long presentation. Diagnostic decision Y/N. ROI-based analysis (FFA)	The right FFA was identified. Right FFA shows more activation for radiologists in training versus non-radiologists, during fast trials, and less activation during long trials.
Wang et al., 2021	N = 40 K = 2 (1) 20 IR: undergraduate students (19 males), 1 month training supervised by an expert radiologist, age = 23.30 ± 1.19 years (2) NC: students with no clinical expertise in radiography (10 males), age = 23.55 ± 1.12 years	3T fMRI Behavioral measure: 4 s assessment of tumor vs non-tumor X-rays Resting-state FC	FC: significant intergroup differences: (1) between cingulate gyri and frontal areas, precuneus, and temporal areas; (2) between the lingual gyrus and occipital and frontal areas; (3) between the fusiform gyrus, cingulate gyri, and frontal areas.
Dong et al., 2022	N = 44 K = 2 (1) 22 RIG: 4-week training students in the X-ray department (11 males), age = 23 ± 0.7 years (2) 22 NC (11 males), age = 23 ± 0.5 years	3T fMRI Behavioral measure: Radiological Expertise Task (RET) and Cambridge Face Memory Test (CFMT) Resting-state ReHo	ReHo between-group difference showed the involvement of the right hippocampus and ventral anterior temporal lobe.
Su et al., 2022	N = 32 K = 1 Medical students who trained for 4 weeks in the X-ray department (15 males), age = 22.47 ± 1.02 years	3T fMRI Longitudinal study Behavioral measure: radiological expertise task; Resting-state ALFF and SVM	After training, the highest discriminative power was shown in the following brain regions: bilateral cingulate gyrus, left superior frontal gyrus, bilateral precentral gyrus, bilateral superior parietal lobule, and bilateral precuneus
Zhang et al., 2022	N = 44 K = 2 (1) 22 RIG: radiology internists, students who trained for 4 weeks in the X-ray department (11 males), age = 23 ± 0.7 years (2) 22 NC: non-expert (11 males), age = 23 ± 0.5 years	3T fMRI Behavioral measure: radiological expertise task; Resting-state ALFF	Higher ALFF in the right fusiform gyrus and left orbitofrontal cortex were observed, and the ALFF in the fusiform gyrus was correlated with the intern radiologists’ behavioral expertise
Wang et al., 2023	N = 44 K = 2 (1) 22 RI: radiology internists, students who trained for 4 weeks in the X-ray department (11 males), age = 23 ± 0.7 years (2) 22 NC: non-expert (11 males), age = 23 ± 0.5 years	3T fMRI Behavioral measure: The Cambridge Face Memory Test (CFMT) and Radiological Expertise Task (RET); Resting-state DC	Significant differences in DC between the RI and control group in the brain regions associated with visual processing, decision-making, memory, attention, and working memory

Abbreviations: N = number of participants; K = number of groups; NC = normal controls; fMRI = functional magnetic resonance imaging; GLM = general linear model; ALFF = amplitude of low-frequency fluctuations; DC = degree centrality; ReHo = regional homogeneity; SVM = support vector machine; FC = functional connectivity; Y = yes; N = no; FFA = fusiform face area; ROI = region of interest; DTI = diffusion tensor imaging; MVPA = multi-voxel pattern analysis.

Neuroplasticity was investigated by radiologists (11 studies), specialists in internal medicine (three studies), and acupuncturists (three studies), followed by surgeons and gastroenterologists (two studies), and neurologists (one study). Eighteen studies took the approach of comparing expert performance and non-expert performance. Four studies took a within-group approach, with only one study employing a longitudinal paradigm, investigating the neuroplasticity changes induced in the students before and after training (Su et al., 2022).

Regarding the magnetic field, all studies except one used a 3T MRI scanner to collect data. Among the selected studies, all studies acquired blood oxygen level-dependent (BOLD) fMRI data, except one study that used structural MRI data (Dong et al., 2013). Fifteen studies (68%) had participants perform some specific task; in the remaining studies, a resting state was used. Among the tasks employed, some are specific to assessing expert performance, such as the think-aloud task (Durning et al., 2015) or the assessment of clinical cases (Durning et al., 2016; Hruska et al., 2016a, 2016b; van den Berg et al., 2020—Table 1). The studies that investigated expert performance in radiologists applied tasks to assess specific cognitive abilities, such as visual object recognition and visual attention (Table 2; Fig. 2).

Figure 2:

Forest and funnel plots of the demographic and behavioral results reported in the included studies.
The analyses were carried out using the proportion of number of experts, novices, and women; standardized mean difference for age; and the effect size for the behavioral results at the different tasks as the outcome measure. Table 3 shows the results of the studies included in the analysis (k). A random-effects model was fitted to the data. The amount of heterogeneity (Tau²) was estimated using the Hunter–Schmidt estimator (Hunter & Schmidt, 1990; Viechtbauer, 2005).

Table 3:

Results for the random effects meta-analysis calculated for the different outcomes as displayed in the included studies.

Outcomes	Z score	P value	CI	Tau² (SE)	I 2	Q	d.f.	P value
No. of experts (k = 21)	10.8	<0.001	0.479–0.693	0.0551 (SE = 0.0236)	93.05%	323.369	20	<0.001
No. of novices (k = 17)	22	<0.001	0.429–0.512	0 (SE = 0.0026)	0%	9.817	16	0.876
No. of women (k = 21)	7.97	<0.001	0.268–0.443	0.0348 (SE = 0.0129)	84.98%	140.727	20	<0.001
Age (k = 16)	3.80	<0.001	0.394–1.232	0.5352 (SE = 0.2469)	78.51%	75.487	15	<0.001
Task performance (k = 16)	2.406	<0.001	2.406–5.310	7.0705 (SE = 3.0416)	82.75%	93.542	15	<0.001

Abbreviations: CI = confidence interval; d.f. = degrees of freedom; SE = standard error.

In addition to the estimate of tau², the Q-test for heterogeneity (Cochran, 1954) and the I² statistic are reported (Table 3), indicating heterogeneity in the number of experts (21 studies), novices (17 studies), age (16 studies), and behavioral task results (16 studies—Fig. 2) (I² > 50% and tau² > 0). Heterogeneity in age can be a logical consequence of the approach used in most studies to infer expert performance from years of clinical practice, which can be found in older physicians.

Figure 2 shows the asymmetries in the funnel plots for the above-mentioned results. Homogeneous results were observed for the number of women included in the studies (21 studies).

fMRI and ALE meta-analysis results

As shown in Table 4, the studies identified a wide range of brain regions associated with the performance of medical experts. However, only 15 (Haller & Radue, 2005; Melo et al., 2011; Durning et al., 2012, 2015, 2016; Morris et al., 2015; Dong et al., 2014, 2015; Hruska et al., 2016b; Bilalić et al., 2016; Ouellette et al., 2020; van den Berg et al., 2020; Wang et al., 2021; Zhang et al., 2022; Dong et al., 2022) out of the 22 included studies provided 3D coordinates. We generated an ALE map that incorporated all the reported 3D coordinates (Fig. 3a). Due to the limited number of studies that reported coordinates (n = 16), we were unable to apply corrections for the false discovery rate (FDR) or familywise error (FWE). Consequently, we followed the recommendations of Eickhoff et al. (2012) and set the threshold for significance at an uncorrected P-value of <0.0005. The resulting map highlights the involvement of cortical areas in the frontal and occipital lobes. In the frontal lobe, we observed engagement mostly in the medial parts. This occurred both dorsally, with notable clusters in the superior frontal gyrus [including the supplementary motor cortex (SMC) and dorsomedial prefrontal cortex (dmPFC)] and inferior frontal gyrus (IFG), and also ventrally, on the border of the middle orbital gyrus and anterior cingulate cortex [corresponding to the ventromedial prefrontal cortex (vmPFC)]. Similarly, the occipital lobe showed involvement in both its dorsal and ventral aspects, with clusters in the superior occipital gyrus (SOG) and fusiform gyrus.

Table 4:

Results for all medical specialties.

			MNI coordinates
Cluster	Hemisphere	BA	x	y	z	Z	ALE	P	Size (mm3)
Inferior frontal gyrus (IFG)	L	44	−46	8	34	4.369	0.006	0.00 001	840
Dorsomedial prefrontal cortex (dmPFC)	L	9	−4	46	42	3.939	0.006	0.00 004	800
Ventromedial prefrontal cortex (vmPFC)	R	10	14	56	4	4.039	0.006	0.00 003	776
Superior occipital gyrus (SOG)	R	18	16	−90	22	3.633	0.005	0.00 014	224
Fusiform gyrus	L	37	−36	−56	−16	3.436	0.005	0.00 030	56

Abbreviations: L = left; R = right; BA = Brodmann area.

Figure 3:

(a) Pooled ALE meta-analysis maps. ALE map of the resulting clusters for studies on medical expertise. Maps are superimposed on a 2 × 2 × 2 mm MNI template according to neurological convention. The colored bar denotes the corresponding ALE value ranges indicated on the maps (P < 0.0005). (b) Radiology results. ALE maps of the resulting clusters for studies on radiological expertise. The map is superimposed on a 2 × 2 × 2 mm MNI template according to neurological convention. The colored bar denotes the corresponding ALE value ranges indicated on the maps (P < 0.0005). (c) Internal medicine results. ALE maps of the resulting clusters for studies on internal medicine expertise. The map is superimposed on a 2 × 2 × 2 mm MNI template according to neurological convention. The colored bar denotes the corresponding ALE value ranges indicated on the maps (P < 0.0005). Abbreviations: SMA = somatomotor area; SOG = superior occipital gyrus; IFG = inferior frontal gyrus; vmPFC = ventromedial prefrontal cortex; dmPFC = dorsomedial prefrontal cortex.

As previously mentioned, there is significant heterogeneity in the medical specialties of the participants, the experimental designs, and the tasks employed in these studies. Due to the variety of studies, we were unable to perform a conjunction analysis for groups with homogeneous tasks, designs, and specialties. To address this, we created two separate maps for the medical specialties that were more common among the studies, specifically radiology and internal medicine (Fig. 3b and c; Table 4).

We created one map for the resting-state studies (Fig. 4b; Table 5) and another for the task-based studies (Fig. 4a). Despite the lack of a quantitative comparison, the maps showed overlapping results. The radiology-expertise map showed clusters corresponding to the IFG, vmPFC, SOG, and fusiform gyrus, which were observed in the pooled map (Fig. 3a). However, the internal medicine map showed a prevalent involvement of the frontal lobes, with the dmPFC, supplementary motor area (SMA), and IFG.

Figure 4:

(a) ALE meta-analysis maps for task-based and resting-state studies. ALE map of the resulting clusters for the task-based studies. Maps are superimposed on a 2 × 2 × 2 mm MNI template according to neurological convention. The colored bar denotes the corresponding ALE value ranges indicated on the maps (P < 0.0005). (b) Resting-state studies. ALE maps of the resulting clusters for resting-state studies. The map is superimposed on a 2 × 2 × 2 mm MNI template according to neurological convention. The colored bar denotes the corresponding ALE value ranges indicated on the maps (P < 0.0005).

Table 5:

Results for radiology, internal medicine, task-based and testing-state ALE meta-analyses.

Cluster			MNI coordinates
Radiology	Side	BA	x	y	z	Z	ALE	P	Size (mm3)
Fusiform gyrus	L	37	−34	−56	−16	3.8160	0.0040	0.0001	376
Superior occipital gyrus (SOG)	R	18	14	−94	22	3.8020	0.0040	0.0001	312
Pre-SMA	L	8	−8	26	56	3.8090	0.0040	0.0001	288
Ventromedial prefrontal cortex (vmPFC)	R	10	16	56	6	3.6900	0.0040	0.0001	256
Inferior frontal gyrus (IFG)	L	44	−44	10	34	3.5300	0.0040	0.0002	96
Internal medicine
Inferior frontal gyrus	L	45	−42	34	10	4.4030	0.0035	0.0000	912
SMA	R	6	0	4	50	4.1370	0.0030	0.0000	656
Dorsomedial prefrontal cortex (dmPFC)	L	9	−2	46	46	3.8150	0.0030	0.0001	648
Task-based studies
Inferior frontal gyrus (IFG)	L	44	−46	8	34	4.7138	0.0063	0.0000	1128
Dorsomedial prefrontal cortex (dmPFC)	L	9	−2	44	44	3.9819	0.0052	0.0000	632
Fusiform gyrus	L	37	−36	−56	−16	3.7164	0.0048	0.0001	272
Precentral gyrus	R	4	40	−28	56	3.4956	0.0045	0.0002	352
Middle occipital gyrus	L	19	−32	−74	30	3.3892	0.0043	0.0004	232
Resting state studies
Ventromedial prefrontal cortex (vmPFC)	L	10	−6	52	−4	4.3617	0.0033	0.0000	1452
Ventromedial prefrontal cortex (vmPFC)	R	10	10	52	8	3.7295	0.0025	0.0000	824
Anterior cingulate cortex	L	32	−4	40	16	3.4178	0.0021	0.0003	40
Anterior cingulate cortex	R	32	10	36	16	3.4108	0.0020	0.0003	24

Abbreviations: L = left; R = right; BA = Brodmann area; SMA = somatomotor area.

All the task-based studies involved frontal regions, resulting in clusters in the inferior frontal gyrus (IFG), dorsomedial prefrontal cortex (dmPFC), and precentral gyrus, as well as occipital regions, including a cluster located in the left fusiform gyrus. The only result obtained from the resting-state studies was the ventromedial prefrontal cortex, which extends to the anterior cingulate cortex bilaterally (BA10/32).

Discussion

Understanding how medical expert performance is achieved and how to characterize it is becoming increasingly relevant in a rapidly changing healthcare landscape where training and education are under growing pressure. In the present study, we sought to investigate such characterization at the brain level. Until quite recently, neural-level information about medical expert performance had to be inferred from studies in non-medical domains, such as chess, sports, and music (Amunts et al., 1997; Leff et al., 2008; Righi et al., 2013; Kim et al., 2014). Only in the last decade have researchers started to use brain imaging techniques to study expert performance in medical disciplines systematically. Medical expert performance probably relies on heterogeneous cognitive mechanisms (e.g. visual recognition of a lesion on a radiograph versus the sensorimotor coupling required for surgical skill); at the same time, there may be neurocognitive mechanisms that are shared. Most situations involve patients and some logical assessment of their condition, and the ability to perform this assessment is the result of continuous exposure to and experience with the relevant features of such situations, probably resulting in neuroplastic changes in the brain. Knowing whether there is such a common substrate at the neurocognitive level and, if so, what are its characteristics, can bring the focus in basic medical education and training back from how disciplines differ to a focus on what they share, advocating more generalistic approaches to medical education such as broad analytical and clinical reasoning abilities. In this study, we present a comprehensive and quantitative meta-analysis attempting to identify consistent neural correlates of medical expert performance. We found that there was considerable heterogeneity across studies, including differences in participant demographics (e.g. age, gender, level of expertise) and experimental tasks (e.g. visual vs. cognitive focus, complexity, and context). The ALE meta-analysis conducted across all included studies suggested that medical expert performance is associated mainly with specific areas in the frontal cortex (superior frontal gyrus, ventromedial prefrontal cortex, and inferior frontal gyrus), with the addition of fewer specific occipital areas. Some areas seemed sensitive to heterogeneity across studies. Below, we discuss these findings.

The apparent heterogeneity across studies in this research field has important implications for how we interpret the findings because the observed patterns may partly reflect biases in the included studies. For instance, age-related differences—often conflated with experience due to reliance on years of clinical practice to define expertise—might contribute to frontal cortex activity levels (Rajah & D’Esposito, 2005). Likewise, task-specific features, such as the heavy reliance on visual tasks in radiology, could drive activation in occipital regions, potentially limiting the generalizability of these effects to other domains of medical expertise. This prompted us to perform separate analyses for study groupings to learn more about what parts of the brain might be especially sensitive to the heterogeneity that is currently present in this research field. The strongest effect appeared to occur for basic fMRI design, i.e. whether participants needed to perform a task or not (the so-called resting state). This analysis showed that the resting-state approach, which is used to study, e.g., how metrics of functional network connectivity might be characteristic of medical expert performance or of participants labeled with medical experts, was strongly associated with the ventromedial prefrontal cortex (vmPFC). On the other hand, effects in the dorsal and lateral parts of the frontal cortex, as well as in the fusiform gyrus, seemed to be elicited primarily when the study design involved a task. This distinction is well aligned with the literature, showing the vmPFC as a core element of the default mode network (a network primarily active during rest) (Rajah & D’Esposito, 2005; Lopez-Persem et al., 2019), areas in the dorsal part of the frontal cortex that are generally related to cognitive tasks (Friedman & Robbins, 2022), and the fusiform gyrus tightly connected to (primarily) visual recognition tasks (Ayzenberg & Behrmann, 2022). The vmPFC is part of the ventral prefrontal cortex, which is considered the neural-level interface of the many different contexts that occur simultaneously in a person (e.g. body state, positive and negative experiences, abstract thinking, and social interactions), making it not only a highly interconnected area of the brain but also a key area for goal selection and decision-making (Badre et al., 2012; Fine & Hayden, 2022). The ventromedial prefrontal cortex is involved in the inhibitory control needed for the decision-making process, a relevant part of the diagnostic process (Yu et al., 2015; de Kloet et al., 2021). It is reasonable to assume that metrics of vmPFC connectivity reflect the capacity for effective decision-making (Hiser & Koenigs, 2018), something that is critical for any physician (Elstein & Schwarz, 2002). So, while the observed brain pattern certainly reflects the type of fMRI design, there is good reason to assume that it also reflects, at least to some extent, medical expert performance.

A recent ALE meta-analysis of over 80 brain imaging studies comparing musicians and non-musicians showed that the brain pattern discerning musical expert performance was distributed instead of localized and involved several areas, mainly in the frontal, temporal, and parietal lobes (Criscuolo et al., 2022). Our meta-analysis also suggested that the brain patterns associated with medical expertise are distributed rather than localized. Musical expert performance often involves some form of auditory processing and/or sensorimotor task, which explains the consistent temporal and parietal lobe involvement, whereas medical expert performance is often tested with tasks reliant on visual search or reading, which favor occipital involvement. Frontal involvement was a main finding in both meta-analyses, supporting the concept that, compared with novices or non-experts, experts do not just show performance that is quantitatively superior (e.g. faster); rather, they also use qualitatively different cognitive strategies (i.e. approaching situations or problems differently) (Bilalić et al., 2016). In the case of medical expert performance, this may become apparent in the way that experts can quickly discern what is relevant and what can safely be ignored, and how they select the best action or plan and execute it with minimal error (Hruska et al., 2016b; ten Cate & Durning, 2018). Interestingly, weight was measured in the lateral frontal cortex for musical expertise and in the medial frontal cortex, especially the superior frontal gyrus (SFG), for medical expertise. What might the SFG contribute that is particularly important for medical expert performance?

In the present study, we identified three SFG clusters that seemed to be associated with medical expert performance in slightly different ways. Two clusters were located in a part of the SFG corresponding to the supplementary motor cortex (SMC) and were associated with internal medicine (caudal SMC) and radiology (rostral SMC) expert performance. These SMC subregions correspond to the locations of the supplementary motor area (SMA) and pre-SMA, respectively (Nachev et al., 2008). It is currently unclear whether this different involvement of the SMC could highlight different cognitive aspects of diagnostic processing prevalent in radiology and internal medicine. It has been well established that the SMA and pre-SMA are involved in processing rules or knowledge structures—mappings of conditions and the best reactions to them—whereas the pre-SMA would be involved in more abstract or complex mappings (Nachev et al., 2008; Bonini et al., 2014). Indeed, experts often act with very limited external cues, drawing on internal models, rules, and sequences of motor actions or thought processes (Carrigan et al., 2022), but it is difficult to see why this should be different, as both radiology and internal medicine experts need much less information to reach a more accurate diagnosis faster compared with non-experts (Manning et al., 2006). On the other hand, there were considerable task differences between studies on radiology and those on internal medicine expertise. Most internal medicine studies involve so-called thinking-aloud protocols, whereas radiology studies involve visual recognition. This is relevant because the SMA is also strongly implicated in speech production, imagery, and familiarity. In a recent study of dancers, the SMA was found to increase its activity during imagined dances with familiar music (Olshansky et al., 2015). Perhaps SMA involvement in tasks that elicit internal medicine expert performance reflects the fact that experts are more familiar with articulating a diagnosis, i.e. using language to translate the symptoms of patients into a diagnosis, using diagnostic rules (i.e. condition‒action mappings). Novices experience difficulty in acquiring familiarity with this process. According to Norman et al. (2007), during the diagnostic process, the impact of familiarity with a specific language occurs at the level of holistic, non-analytic processing and during feature interpretation. Indeed, novices were found to be more confident in making a diagnosis when they were supported by familiar feature descriptions (Young et al., 2007). A third SFG cluster located rostral to the SMC in the dorsomedial prefrontal cortex was associated with medical expert performance across all medical disciplines included in the meta-analysis, but especially if the study involved a task. The involvement of this rostral SFG area across several different tasks could signify an inclination toward more abstract cognitive processes, such as evaluating or reasoning (Parson & Osherson, 2001; Mansi et al., 2022), which are also typical of the clinical diagnostic process.

The predominant involvement of SFG could fit well with cognitive strategies that medical experts have shown to use, such as fast initial global search and subsequent focal analysis (zooming in on what is relevant) in radiologists detecting lesions on a radiograph (Gandomkar & Mello-Thoms, 2019). Evidence from non-medical fields supports such a role for SFG: experts process relevant information as single units, lowering the cost of attention and working memory (Akyürek et al., 2017), which effectively renders the situation less complex. In addition, the performance of mathematicians, who develop high levels of automaticity (Jackson & Coney, 2005; Stickney et al., 2012; Baker & Cuevas, 2018) , is positively associated with SFG gray matter density (Popescu et al., 2019). The SFG and SMC, in particular, have also been implicated in inhibitory control (Floden & Stuss, 2006; Hsu et al., 2011), which is very important for counteracting impulsive behavior and selecting the right action.

Finally, there is strong connectivity between areas of the SFG and vmPFC, which could serve to integrate such automated cognitive processes with the vmPFC, an area of the frontal cortex that, as we have already mentioned, has access to many different contexts and information critical to decision-making (de Kloet et al., 2021). These two regions were found to be part of the “stopping network,” which plays a crucial role in inhibitory control (Aron & Poldrack, 2006; Zandbelt et al., 2013; Yu et al., 2015). One could therefore speculate that the predominant involvement of the medial frontal cortex is in support of the view that medical expert performance, like other types of expertise, relies on complex knowledge structures that can be effectively linked to the selection of the correct goals, policies, or actions. The net effect would be that a complex situation, such as making a clinical diagnosis, is reduced to something seemingly simple and automatic (Bilalić, 2017).

The heterogeneity across studies in this field, as well as the still limited mechanistic insight, means that the view on frontal cortex involvement in medical expert performance presented here should not be read as a strong conclusion. In fact, there are interesting findings in individual studies that may point to other types of mechanisms involving the frontal cortex. For instance, in one study (Durning et al., 2015), expert physicians recruited more dorsal prefrontal areas, whereas non-experts recruited more ventral prefrontal areas, including the orbitofrontal and anterior cingulate cortex, when diagnosing clinical cases. The specific task that participants were required to perform (thinking aloud during clinical reasoning) was probably less automatized in non-experts, possibly leading to more heavy taxing of the vmPFC. However, under different circumstances, this could also apply to medical experts. For instance, medical experts may decide to deviate from the protocol if the situation calls for prioritizing other clinical goals, which may involve choosing from several more (uncertain) policies necessitating the engagement of ventral prefrontal areas (Badre et al., 2012; Morriss et al., 2021). This may also be expressed in the way the PFC communicates with other parts of the brain. A study in expert neurologists (van den Berg et al., 2020) found that relative to solving typical cases, solving ambiguous clinical cases changed the connectivity of the prefrontal cortex with regions outside of the prefrontal cortex (parietal regions, cerebellum), which may point to additional strategies that experts can employ to address uncertainty. Likewise, enhanced functional connectivity was frequently observed in trained experts, particularly in studies involving acupuncturists and medical trainees. Regions such as the medial prefrontal cortex (Dong et al., 2014; Zhang et al., 2022) and somatosensory cortex showed stronger local coherence and functional connectivity, which was correlated with training and expertise. Unfortunately, the data available lack sufficient non-expert or novice data, which precludes a more in-depth analysis of medical expert performance.

The left inferior frontal gyrus has also been implicated in medical expert performance. The location in the frontal operculum positions it in Broca’s area, which is, of course, renowned for its role in speech and semantic tasks more broadly (DeWit & Rauschecker, 2016). Most studies included in this meta-analysis did not require verbal responses, with a few important exceptions (Durning et al., 2012, 2013, 2015, 2016); thus, it seems unlikely that this effect is associated with overt speech due to thinking aloud during the hypothesis-driven decisional process. The opercular part of the left inferior frontal gyrus corresponds to the caudal part of Broca’s area. This area has been implicated in several other functions, including hand movement (Rizolatti et al., 1996) and the phenomenon of inner speech or internal monologue (Geva et al., 2011). Unlike verbal responses, hand movements were required in most studies to deliver responses, but control subjects such as novices and non-experts were also required to produce these responses, making it an unlikely explanation for the stronger activation in experts. The different uses of inner speech by medical experts are more likely explanations. It has been proposed to be involved in various cognitive functions, including working memory (Alderson-Day & Fernyhough, 2015; Briedis, 2020) and problem-solving, both of which are instrumental to clinical reasoning (Wallace et al., 2017). Moreover, a specific form of inner speech, self-talk, has been proposed to play a significant role in behavioral control and motivation during competition and high-performance sports (see Hardy, 2006, for a review), which may resemble the way medical experts approach their performance.

Outside of the frontal lobe, there was homogeneous involvement of the FFA across different forms of medical expertise. The FFA is particularly famous for its role in face perception (Kanwisher & Yovel, 2006; McGugin et al., 2016); however, visual experts seem to activate it also for non-facial stimuli. For instance, chess players activate the FFA when asked to assess a chess position and consider the best next move (Bilalić et al., 2010, 2012). These findings support the idea that the FFA allows the individuation of stimuli within the same visual category in an experience-dependent manner (Gauthier et al., 2000), which means that experts would activate the FFA more when exposed to stimuli within their visual expertise (Sigurdardottir & Gauthier, 2015).

In medical expertise brain imaging studies, finding FFA is not uncommon, especially in studies on radiology expert performance (Harley et al., 2009; Bilalić et al., 2016; Kok et al., 2021). Notably, radiological expertise has captivated, more than other medical expertise, the scientific community’s interest, and in the present meta-analysis, about half of the studies meeting the inclusion criteria were about radiology expert performance. The analysis of study groupings suggested strongly that the FFA, mainly on the left side, was more strongly involved in radiology expert performance than in other types of medical expert performance. In an interesting study, Kok et al. (2021) directly tested how this FFA involvement would relate to the two-phase diagnostic process that is typical of radiology expert performance: a fast initial phase known as the “holistic mode” followed by a slower “checking (search-to-find) mode” of image processing (Kundel et al., 2007; Sheridan & Rheingold, 2017). Their study showed that the FFA was only involved in the former (Kok et al., 2021). This is in line with brain imaging studies of radiology expertise, which generally employ short stimulus durations, which are most likely to elicit holistic mode image processing in experts and, indeed, FFA activity. Some of the studies included in this meta-analysis analyzed face and object (i.e. rooms or tools) recognition skills using standardized tests (Dennett et al., 2012) to discover possible a priori general visual memory skills that could distinguish experts from non-experts. Such differences were generally not observed (Bilalić et al., 2016). Immediate and accurate recognition of key features and subsequent decisions, as seen in radiology experts, would probably require coupling with knowledge structures stored elsewhere in the brain. The separate analysis performed in radiology expertise brain imaging studies indeed pointed to the involvement of several areas of the frontal lobe in radiology expertise, such as the pre-SMA and the opercular part of the inferior frontal gyrus. Whether the more evaluative checking mode, which probably involves different, more deliberate cognitive processing, would be associated, e.g. with more rostral prefrontal areas, is a matter of speculation. Future studies focused on brain connectivity during different image processing modes are necessary to provide further evidence of the underlying anatomical and functional networks.

Limitations

Nearly all studies that were included employed fMRI, and most required participants to perform tasks relevant to their medical practice. In addition, most studies included groups of socially recognized medical experts and non-experts. Experts are typically defined on time-based criteria, as expert performance is usually acquired after several years of practice. In most of the studies included, clinical practice in years was indeed taken as the principal indicator of medical expertise. However, other criteria, such as professional achievements or official or governmental assessments of professionals, are often not reported. In addition, studies do not always include a direct measure to verify medical expert performance. During the experiments, performance by participants labeled as experts was on average superior to those labeled as non-experts, but it is important to emphasize that several factors can influence professional achievements, and extensive clinical practice, after formal education, does not always lead one to exhibit reproducibly superior performance on representative, authentic tasks in their field (Ericsson, 2006). We call upon our research community to go to greater lengths to verify that medical experts included in expertise studies not only show superior performance but also that it is reproducible.

The medical disciplines, as well as the clinical competencies, that have been investigated with brain imaging methods are still quite limited. As already mentioned, approximately half of the studies that met the inclusion criteria were in the field of radiology, with a focus on the fast assessment of lesions. This inevitably means that radiology and visual expertise have a relatively strong say in the current pattern of areas reflecting putative medical expertise. It will be important for future studies to involve other medical disciplines and study other competencies. For instance, Goldman (2018) stresses that expert performance has a strict prosocial connotation because an expert can help others solve a variety of problems. This definition extends the construct of expertise beyond the classic expertise domains to imply idiosyncratic attitudes and social skills as well. Social corollaries of medical expertise may be instrumental in patient care (e.g. communication) as well as supervision (e.g. teaching, training) interactions, but we are currently largely unaware of the underlying mechanisms.

Several potential influencing factors, such as experimental design, imaging modality, and expertise domains, contribute to the findings. However, no analysis was conducted to assess heterogeneity and impact on the results, which is a limitation.

The functional and structural connectivity patterns characterizing medical expert performance have only been investigated to a very limited extent (few studies in radiologists and acupuncturists), which means that here, we can only attempt to interpret the involvement of the identified pattern at large and the individual brain areas that are part of it. Having brain areas that consistently activate in studies that involve some form of expert performance is very promising, as it provides important information above and beyond individual studies. At the same time, we need to acknowledge that the concept of medical expert performance at the brain level that we present here is still quite limited. It does not capture the highly dynamic nature of brain function that, in reality, relies on many constellations of areas working together in ever-changing, context-dependent networks that adapt to cognitive demands, sensory inputs, and internal states (e.g. Kringelbach & Deco, 2020).

Conclusion

We can only predict what will be asked of healthcare professionals in the future, but it seems almost certain that there will be major shifts driven, e.g., by workforce shortages, changes in the magnitude and composition of patient populations, and far-reaching implications of technological advances. As expertise is at least also a product of the human brain, it makes sense to look for a characterization of medical expert performance at the neural level. Although the included studies showed heterogeneity, we observed a consistent pattern across various medical specialties and tasks, particularly within the medial PFC. The involvement of the medial PFC and SMC highlights the complex role played by executive functions, specifically, by inhibitory control, which is fundamental to any clinical scenario or decision-making process. Therefore, it is important to improve training programs and curricula to stimulate the decision-making process and develop critical thinking in future healthcare practitioners with the help of new technologies, such as virtual or augmented reality and AI. Future studies are necessary to test specific hypotheses that our current meta-analysis suggests could be relevant, e.g. deliberately manipulating abstraction, difficulty, or flexibility; studying how prefrontal cortical areas respond to those challenges; or investigating the context-specific functional connectivity of such areas.

Disclaimer

The views expressed herein are those of the authors and not necessarily those of the Department of Defense, the Uniformed Services University, or other Federal entities in the United States.

Supplementary Material

kkaf019_Supplemental_File

kkaf019_supplemental_file.docx ^{(13KB, docx)}

Acknowledgements and funding

J.P. was supported by a grant from the Portuguese Foundation for Science and Technology (FCT-2024.01355.BD). M.H.D. was supported by the National Key R&D Program of China (Grant No. 2022YFF1202400)

Contributor Information

Nicoletta Cera, Faculty of Psychology and Education Sciences, University of Porto, 4200-135 Porto, Portugal; Research Unit in Medical Imaging and Radiotherapy, Cross I&D Lisbon Research Centre, Escola Superior de Saúde da Cruz Vermelha Portuguesa, 1300-125 Lisbon, Portugal.

Joana Pinto, Faculty of Psychology and Education Sciences, University of Porto, 4200-135 Porto, Portugal; Faculty of Medicine, University of Porto, 4200-319 Porto, Portugal.

Minghao Dong, Engineering Research Center of Molecular and Neuro Imaging of the Ministry of Education, School of Life Science and Technology, Xidian University, No. 2 South Taibai Road, Xi’an 710071, China; Xian Key Laboratory of Intelligent Sensing and Regulation of trans-Scale Life Information, Xidian University, 710071 Xi’an, China.

Steven Durning, Department of Health Professions Education, Uniformed Services University of the Health Sciences, Bethesda, MD, USA; Department of Internal Medicine, Uniformed Services University of the Health Sciences, Bethesda, MD, USA.

Janniko R Georgiadis, Department of Biomedical Sciences, University of Groningen, University Medical Centre Groningen (UMCG), Antonius Deusinglaan, Groningen, 1, 9713 AV, The Netherlands; School of Medicine and Health Sciences, Carl von Ossietzky University Oldenburg, Oldenburg, Germany.

Author contributions

Nicoletta Cera (Conceptualization, Data curation, Methodology, Writing—original draft), Joana Pinto (Data curation), Minghao Dong (Validation), and Janniko R. Georgiadis (Conceptualization, Writing—original draft, Writing—review & editing).

Conflict of interest statement

None declared.

References

1. Akyürek  EG, Kappelmann  N, Volkert  M  et al. (2017) What you see is what you remember: visual chunking by temporal integration enhances working memory. J Cogn Neurosci. 29:2025–36. [DOI] [PubMed] [Google Scholar]
2. Alderson-Day  B, Fernyhough  C (2015) Inner speech: development, cognitive functions, phenomenology, and neurobiology. Psychol Bull. 141:931. [DOI] [PMC free article] [PubMed] [Google Scholar]
3. Amunts  K, Schlaug  G, Jäncke  L  et al. (1997) Motor cortex and hand motor skills: structural compliance in the human brain. Hum Brain Mapp. 5:206–15. [DOI] [PubMed] [Google Scholar]
4. Aron  AR, Poldrack  RA (2006) Cortical and subcortical contributions to Stop signal response inhibition: role of the subthalamic nucleus. J Neurosci. 26:2424–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
5. Ayzenberg  V, Behrmann  M (2022) Does the brain’s ventral visual pathway compute object shape?. Trends Cogn Sci. 26: 1119–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
6. Badre  D, Doll  BB, Long  NM  et al. (2012) Rostrolateral prefrontal cortex and individual differences in uncertainty-driven exploration. Neuron. 73:595–607. [DOI] [PMC free article] [PubMed] [Google Scholar]
7. Baker  AT, Cuevas  J (2018) The importance of automaticity development in mathematics. Georgia Educational Researcher. 14:13–23. [Google Scholar]
8. Bertram  R, Kaakinen  J, Bensch  F  et al. (2016) Eye movements of radiologists reflect expertise in CT study interpretation: a potential tool to measure resident development. Radiology. 281:805–15. [DOI] [PubMed] [Google Scholar]
9. Bilalić  M (2017) The Neuroscience of Expertise. Cambridge University Press; Cambridge [Google Scholar]
10. Bilalić  M, Grottenthaler  T, Nägele  T  et al. (2016) The faces in radiological images: fusiform face area supports radiological expertise. Cereb Cortex. 26:1004–14. [DOI] [PubMed] [Google Scholar]
11. Bilalić  M, Langner  R, Erb  M  et al. (2010) Mechanisms and neural basis of object and pattern recognition: a study with chess experts. J Exp Psychol Gen. 139:728–42. [DOI] [PubMed] [Google Scholar]
12. Bilalić  M, Turella  L, Campitelli  G  et al. (2012) Expertise modulates the neural basis of context dependent recognition of objects and their relations. Hum Brain Mapp. 33:2728–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
13. Bonini  F, Burle  B, Liégeois-Chauvel  C  et al. (2014) Action monitoring and medial frontal cortex: leading role of supplementary motor area. Science. 343:888–91. [DOI] [PubMed] [Google Scholar]
14. Briedis  M (2020) Phenomenological ethnography of radiology: expert performance in enacting diagnostic cognition. Phenomenol Cognitive Sci. 19:373–404. [Google Scholar]
15. Carrigan  AJ, Charlton  A, Foucar  E  et al. (2022) The role of cue-based strategies in skilled diagnosis among pathologists. Hum Factors. 64:1154–67. [DOI] [PubMed] [Google Scholar]
16. Cochran  WG (1954) The combination of estimates from different experiments. Biometrics. 10:101–29. [Google Scholar]
17. Criscuolo  A, Pando-Naude  V, Bonetti  L  et al. (2022) An ALE meta-analytic review of musical expertise. Sci Rep. 12:11726. [DOI] [PMC free article] [PubMed] [Google Scholar]
18. de Kloet  SF, Bruinsma  B, Terra  H  et al. (2021) Bi-directional regulation of cognitive control by distinct prefrontal cortical output neurons to thalamus and striatum. Nat Commun. 12:1994. [DOI] [PMC free article] [PubMed] [Google Scholar]
19. Dennett  HW, McKone  E, Tavashmi  R  et al. (2012) The Cambridge Car Memory Test: a task matched in format to the Cambridge Face Memory Test, with norms, reliability, sex differences, dissociations from face memory, and expertise effects. Behav Res Methods. 44:587–605. [DOI] [PubMed] [Google Scholar]
20. DeWitt  I, Rauschecker  JP (2016) Convergent evidence for the causal involvement of anterior superior temporal gyrus in auditory single-word comprehension. Cortex. 77:164–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
21. Dong  M, Li  J, Shi  X  et al. (2015) Altered baseline brain activity in experts measured by amplitude of low frequency fluctuations (ALFF): a resting state fMRI study using expertise model of acupuncturists. Front Hum Neurosci. 9:99. [DOI] [PMC free article] [PubMed] [Google Scholar]
22. Dong  M, Qin  W, Zhao  L  et al. (2014) Expertise modulates local regional homogeneity of spontaneous brain activity in the resting brain: an fMRI study using the model of skilled acupuncturists. Hum Brain Mapp. 35:1074–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
23. Dong  M, Zhang  P, Chai  W  et al. (2022) Early stage of radiological expertise modulates resting-state local coherence in the inferior temporal lobe. Psychoradiology. 2:199–206. [DOI] [PMC free article] [PubMed] [Google Scholar]
24. Dong  M, Zhao  L, Yuan  K  et al. (2013) Length of acupuncture training and structural plastic brain changes in professional acupuncturists. PLoS One. 8:e66591. [DOI] [PMC free article] [PubMed] [Google Scholar]
25. Durning  SJ, Costanzo  ME, Artino  AR  et al. (2015) Neural basis of nonanalytical reasoning expertise during clinical evaluation. Brain Behav. 5:e00309. [DOI] [PMC free article] [PubMed] [Google Scholar]
26. Durning  SJ, Costanzo  ME, Beckman  TJ  et al. (2016) Functional neuroimaging correlates of thinking flexibility and knowledge structure in memory: exploring the relationships between clinical reasoning and diagnostic thinking. Med Teach. 38:570–7. [DOI] [PubMed] [Google Scholar]
27. Durning  SJ, Graner  J, Artino  AR  Jr  et al. (2012) Using functional neuroimaging combined with a think-aloud protocol to explore clinical reasoning expertise in internal medicine. Mil Med. 177:72–8. [DOI] [PubMed] [Google Scholar]
28. Durning  SJ, Ratcliffe  T, Artino  AR  Jr  et al. (2013) How is clinical reasoning developed, maintained, and objectively assessed? Views from expert internists and internal medicine interns. J Contin Educ Health Prof. 33:215–23. [DOI] [PubMed] [Google Scholar]
29. Eickhoff  SB, Bzdok  D, Laird  AR  et al. (2012) Activation likelihood estimation meta-analysis revisited. Neuroimage. 59:2349–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
30. Elstein  AS, Schwarz  A (2002) Clinical problem solving and diagnostic decision making: selective review of the cognitive literature. BMJ. 324:729–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
31. Ericsson  KA (2004) Deliberate practice and the acquisition and maintenance of expert performance in medicine and related domains. Acad Med. 79:S70–81. [DOI] [PubMed] [Google Scholar]
32. Ericsson  KA (2006) The influence of experience and deliberate practice on the development of superior expert performance. The Cambridge Handbook of Expertise and Expert Performance. 38:2–2. [Google Scholar]
33. Eriksen  MB, Frandsen  TF (2018) The impact of patient, intervention, comparison, outcome (PICO) as a search strategy tool on literature search quality: a systematic review. J Med Libr Assoc. 106:420. [DOI] [PMC free article] [PubMed] [Google Scholar]
34. Fine  JM, Hayden  BY (2022) The whole prefrontal cortex is premotor cortex. Philos Trans R Soc Lond B Biol Sci. 377:20200524. [DOI] [PMC free article] [PubMed] [Google Scholar]
35. Floden  D, Stuss  DT (2006) Inhibitory control is slowed in patients with right superior medial frontal damage. J Cogn Neurosci. 18:1843–9. [DOI] [PubMed] [Google Scholar]
36. Friedman  NP, Robbins  TW (2022) The role of prefrontal cortex in cognitive control and executive function. Neuropsychopharmacology. 47:72–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
37. Gandomkar  Z, Mello-Thoms  C  (2019);  Visual search in breast imaging. Br J Radiol 92:20190057. [DOI] [PMC free article] [PubMed] [Google Scholar]
38. Gauthier  I, Skudlarski  P, Gore  JC  et al. (2000) Expertise for cars and birds recruits brain areas involved in face recognition. Nat Neurosci. 3:191–7. [DOI] [PubMed] [Google Scholar]
39. Geva  S, Jones  PS, Crinion  JT  et al. (2011) The neural correlates of inner speech defined by voxel-based lesion–symptom mapping. Brain. 134:3071–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
40. Goldman  AI (2018) Expertise. Topoi. 37:3–10. [Google Scholar]
41. Grauman  Å, Ancillotti  M, Veldwijk  J  et al. (2023) Precision cancer medicine and the doctor-patient relationship: a systematic review and narrative synthesis. BMC Med Inform Decis Mak. 23:286. [DOI] [PMC free article] [PubMed] [Google Scholar]
42. Haller  S, Radue  EW (2005) What is different about a radiologist’s brain?. Radiology. 236:983–9. [DOI] [PubMed] [Google Scholar]
43. Hardy  J (2006) Speaking clearly: a critical review of the self-talk literature. Psychol Sport Exercise. 7:81–97. [Google Scholar]
44. Harley  EM, Pope  WB, Villablanca  JP  et al. (2009) Engagement of fusiform cortex and disengagement of lateral occipital cortex in the acquisition of radiological expertise. Cereb Cortex. 19:2746–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
45. Harris  N, Bacon  CEW (2019) Developing cognitive skills through active learning: a systematic review of health care professions. Athletic Training Edu J. 14:135–48. [Google Scholar]
46. Higgins  JP, Thompson  SG, Deeks  JJ  et al. (2003) Measuring inconsistency in meta-analyses. BMJ. 327:557–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
47. Hiser  J, Koenigs  M (2018) The multifaceted role of the ventromedial prefrontal cortex in emotion, decision making, social cognition, and psychopathology. Biol Psychiatry. 83:638–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
48. Hruska  P, Hecker  KG, Coderre  S  et al. (2016a) Hemispheric activation differences in novice and expert clinicians during clinical decision making. Adv Health Sci Educ Theory Pract. 21:921–33. [DOI] [PubMed] [Google Scholar]
49. Hruska  P, Krigolson  O, Coderre  S  et al. (2016b) Working memory, reasoning, and expertise in medicine-insights into their relationship using functional neuroimaging. Adv Health Sci Educ Theory Pract. 21:935–52. [DOI] [PubMed] [Google Scholar]
50. Hsu  TY, Tseng  LY, Yu  JX  et al. (2011) Modulating inhibitory control with direct current stimulation of the superior medial frontal cortex. Neuroimage. 56:2249–57. [DOI] [PubMed] [Google Scholar]
51. Hunter  JE, Schmidt  FL (1990) Dichotomization of continuous variables: the implications for meta-analysis. J Appl Psychol. 75:334. [Google Scholar]
52. Jackson  N, Coney  J (2005) Simple arithmetic processing: the question of automaticity. Acta Psychol (Amst). 119:41–66. [DOI] [PubMed] [Google Scholar]
53. Janelle  CM, Hatfield  BD (2008) Visual attention and brain processes that underlie expert performance: implications for sport and military psychology. Military Psychol. 20:S39–69. [Google Scholar]
54. Kanwisher  N, Yovel  G (2006) The fusiform face area: a cortical region specialized for the perception of faces. Philos Trans R Soc Lond B Biol Sci. 361:2109–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
55. Khosravi  M, Zare  Z, Mojtabaeian  SM  et al. (2024) Artificial intelligence and decision-making in healthcare: a thematic analysis of a systematic review of reviews. Health Serv Res Manag Epidemiol. 11:23333928241234863. [DOI] [PMC free article] [PubMed] [Google Scholar]
56. Kim  W, Chang  Y, Kim  J  et al. (2014) An fMRI study of differences in brain activity among elite, expert, and novice archers at the moment of optimal aiming. Cogn Behav Neurol. 27:173–82. [DOI] [PubMed] [Google Scholar]
57. Kok  E, De Bruin  AB, van Geel  K  et al. (2018) The neural implementation of surgical expertise within the mirror-neuron system: an fMRI study. Front Hum Neurosci. 12:291. [DOI] [PMC free article] [PubMed] [Google Scholar]
58. Kok  EM, Sorger  B, van Geel  K  et al. (2021) Holistic processing only? The role of the right fusiform face area in radiological expertise. PLoS One. 16:e0256849. [DOI] [PMC free article] [PubMed] [Google Scholar]
59. Kringelbach  ML, Deco  G (2020) Brain states and transitions: insights from computational neuroscience. Cell Rep. 32:108128. [DOI] [PubMed] [Google Scholar]
60. Kundel  HL, Nodine  CF, Conant  EF  et al. (2007) Holistic component of image perception in mammogram interpretation: gaze-tracking study. Radiology. 242:396–402. [DOI] [PubMed] [Google Scholar]
61. Leff  DR, Leong  JJ, Aggarwal  R  et al. (2008) Could variations in technical skills acquisition in surgery be explained by differences in cortical plasticity?. Ann Surg. 247:540–3. [DOI] [PubMed] [Google Scholar]
62. Logothetis  NK (2008) What we can do and what we cannot do with fMRI. Nature. 453:869–78. [DOI] [PubMed] [Google Scholar]
63. Lopez-Persem  A, Verhagen  L, Amiez  C  et al. (2019) The human ventromedial prefrontal cortex: sulcal morphology and its influence on functional organization. J Neurosci. 39:3627–39. [DOI] [PMC free article] [PubMed] [Google Scholar]
64. Manning  D, Ethell  S, Donovan  T  et al. (2006) How do radiologists do it? The influence of experience and training on searching for chest nodules. Radiography. 12:134–42. [Google Scholar]
65. Mansi  SA, Teresa  MM, Seri  S  et al. (2022) Frontal intrinsic connectivity networks support contradiction identification during inductive and deductive reasoning. Cogn Comput. 14:677–92. [Google Scholar]
66. McGugin  RW, Van Gulick  AE, Gauthier  I (2016) Cortical thickness in fusiform face area predicts face and object recognition performance. J Cogn Neurosci. 28:282–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
67. Melo  M, Scarpin  DJ, Amaro  E  et al. (2011) How doctors generate diagnostic hypotheses: a study of radiological diagnosis with functional magnetic resonance imaging. PLoS One. 6:e28752. [DOI] [PMC free article] [PubMed] [Google Scholar]
68. Moher  D, Liberati  A, Tetzlaff  J  et al. (2009) Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Ann Intern Med. 151:264–9. [DOI] [PubMed] [Google Scholar]
69. Morris  MC, Frodl  T, D’Souza  A  et al. (2015) Assessment of competence in surgical skills using functional magnetic resonance imaging: a feasibility study. J Surg Educ. 72:198–204. [DOI] [PubMed] [Google Scholar]
70. Morriss  J, Bell  T, Biagi  N  et al. (2021) Intolerance of uncertainty is associated with heightened responding in the prefrontal cortex during cue-signalled uncertainty of threat. Cogn Affect Behav Neurosci. 22:88–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
71. Moskovich  L, Rozani  V (2025) Health profession students’ perceptions of ChatGPT in healthcare and education: insights from a mixed-methods study. BMC Med Educ. 25:98. [DOI] [PMC free article] [PubMed] [Google Scholar]
72. Nachev  P, Kennard  C, Husain  M (2008) Functional role of the supplementary and pre-supplementary motor areas. Nat Rev Neurosci. 9:856–69. [DOI] [PubMed] [Google Scholar]
73. Nakashima  R, Komori  Y, Maeda  E  et al. (2016) Temporal characteristics of radiologists’ and novices’ lesion detection in viewing medical images presented rapidly and sequentially. Front Psychol. 7:1553. [DOI] [PMC free article] [PubMed] [Google Scholar]
74. Norman  GR, Young  ME, Brooks  LR (2007) Non-analytical models of clinical reasoning: the role of experience. Med Educ. 41:1140–5. [DOI] [PubMed] [Google Scholar]
75. Olshansky  MP, Bar  RJ, Fogarty  M  et al. (2015) Supplementary motor area and primary auditory cortex activation in an expert break-dancer during the kinesthetic motor imagery of dance to music. Neurocase. 21:607–17. [DOI] [PubMed] [Google Scholar]
76. Ouellette  DJ, Van Staalduinen  E, Hussaini  SH  et al. (2020) Functional, anatomical and diffusion tensor MRI study of radiology expertise. PLoS One. 15:e0231900. [DOI] [PMC free article] [PubMed] [Google Scholar]
77. Parsons  LM, Osherson  D (2001) New evidence for distinct right and left brain systems for deductive versus probabilistic reasoning. Cereb Cortex. 11:954–65. [DOI] [PubMed] [Google Scholar]
78. Pearson-Stuttard  J, Ezzati  M, Gregg  EW (2019) Multimorbidity—a defining challenge for health systems. Lancet Public Health. 4:e599–600. [DOI] [PubMed] [Google Scholar]
79. Poldrack  RA, Mumford  JA, Schonberg  T  et al. (2012) Discovering relations between mind, brain, and mental disorders using topic mapping. PLoS Comput Biol. 8:e1002707. [DOI] [PMC free article] [PubMed] [Google Scholar]
80. Popescu  T, Sader  E, Schaer  M  et al. (2019) The brain-structural correlates of mathematical expertise. Cortex. 114:140–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
81. Rajah  MN, D’Esposito  M (2005) Region-specific changes in prefrontal function with age: a review of PET and fMRI studies on working and episodic memory. Brain. 128:1964–83. [DOI] [PubMed] [Google Scholar]
82. Righi  G, Tarr  MJ, Kingon  A (2013) Category-selective recruitment of the fusiform gyrus with chess expertise. In Staszewski  JJ (ed.) Expertise and Skill Acquisition. New York: Psychology Press, 261–80. [Google Scholar]
83. Ritchie  MJ, Parker  LE, Kirchner  JE (2020) From novice to expert: a qualitative study of implementation facilitation skills. Implement Sci Commun. 1:25. [DOI] [PMC free article] [PubMed] [Google Scholar]
84. Rizzolatti  G, Fadiga  L, Gallese  V  et al. (1996) Premotor cortex and the recognition of motor actions. Brain Res Cogn Brain Res. 3:131–41. [DOI] [PubMed] [Google Scholar]
85. Sheridan  H, Rheingold  EM (2017) The holistic processing account of visual expertise in medical image perception: a review. Front Psychol. 8:1620. [DOI] [PMC free article] [PubMed] [Google Scholar]
86. Sigurdardottir  HM, Gauthier  I (2015) Expertise and object recognition. Brain Mapp. 2:523–7. [Google Scholar]
87. Silvennoinen  M, Mecklin  JP, Saariluoma  P  et al. (2009) Expertise and skill in minimally invasive surgery. Scand J Surg. 98:209–13. [DOI] [PubMed] [Google Scholar]
88. Stickney  EM, Sharp  LB, Kenyon  AS (2012) Technology-enhanced assessment of math fact automaticity: patterns of performance for low-and typically achieving students. Assessment for Effective Intervention. 37:84–94. [Google Scholar]
89. Su  J, Zhang  X, Zhang  Z  et al. (2022) Real-world visual experience alters baseline brain activity in the resting state: a longitudinal study using expertise model of radiologists. Front Neurosci. 16:904623. [DOI] [PMC free article] [PubMed] [Google Scholar]
90. ten Cate  O, Durning  SJ (2018) Understanding clinical reasoning from multiple perspectives: a conceptual and theoretical overview. In: ten Cate  O, Custers  EJFM, Durning  SJ (eds.) Principles and Practice of Case-based Clinical Reasoning Education: a Method for Preclinical Students. Berlin: Springer, 35–46. [Google Scholar]
91. Turkeltaub  PE, Eickhoff  SB, Laird  AR  et al. (2012) Minimizing within-experiment and within-group effects in Activation Likelihood Estimation meta-analyses. Hum Brain Mapp. 33:1–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
92. van den Berg  B, de Bruin  ABH, Marsman  JC  et al. (2020) Thinking fast or slow? Functional magnetic resonance imaging reveals stronger connectivity when experienced neurologists diagnose ambiguous cases. Brain Commun. 2:fcaa023. [DOI] [PMC free article] [PubMed] [Google Scholar]
93. Viechtbauer  W (2005) Bias and efficiency of meta-analytic variance estimators in the random-effects model. J Edu Behav Stat. 30:261–93. [Google Scholar]
94. Wallace  GL, Peng  CS, Williams  D (2017) Interfering with inner speech selectively disrupts problem solving and is linked with real-world executive functioning. J Speech Lang Hear Res. 60:3456–60. [DOI] [PubMed] [Google Scholar]
95. Wang  H, Yao  R, Zhang  X  et al. (2023) Visual expertise modulates resting-state brain network dynamics in radiologists: a degree centrality analysis. Front Neurosci. 17:1152619. [DOI] [PMC free article] [PubMed] [Google Scholar]
96. Wang  RC, Wang  Z (2023) Precision medicine: disease subtyping and tailored treatment. Cancers. 15:3837. [DOI] [PMC free article] [PubMed] [Google Scholar]
97. Wang  Y, Jin  C, Yin  Z  et al. (2021) Visual experience modulates whole-brain connectivity dynamics: a resting-state fMRI study using the model of radiologists. Hum Brain Mapp. 42:4538–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
98. Wells  GA, Shea  B, O’Connell  D  et al. (2000) The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses.
99. Wimmers  PF, Schmidt  HG, Verkoeijen  PP  et al. (2005) Inducing expertise effects in clinical case recall. Med Educ. 39:949–57. [DOI] [PubMed] [Google Scholar]
100. Wu  J, Wang  J, Georgiadis  JR  et al. (2024) Expertise, brain plasticity, and resting state. Psychoradiology. 4:kkae020. [DOI] [PMC free article] [PubMed] [Google Scholar]
101. Young  M, Brooks  L, Norman  G (2007) Found in translation: the impact of familiar symptom descriptions on diagnosis in novices. Med Educ. 41:1146–51. [DOI] [PubMed] [Google Scholar]
102. Yu  J, Tseng  P, Hung  DL  et al. (2015) Brain stimulation improves cognitive control by modulating medial-frontal activity and preSMA-vmPFC functional connectivity. Hum Brain Mapp. 36::  4004–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
103. Zandbelt  BB, Bloemendaal  M, Hoogendam  JM  et al. (2013) Transcranial magnetic stimulation and functional MRI reveal cortical and subcortical interactions during stop-signal response inhibition. J Cogn Neurosci. 25:157–74. [DOI] [PubMed] [Google Scholar]
104. Zhang  T, Dong  M, Wang  H  et al. (2022) Visual expertise modulates baseline brain activity: a preliminary resting-state fMRI study using expertise model of radiologists. BMC Neurosci. 23:24. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

kkaf019_Supplemental_File

kkaf019_supplemental_file.docx ^{(13KB, docx)}