Rios-Duarte, Jorge A

BACKGROUND There is a scarcity of artificial intelligence models trained on frozen pathology. One way to expand the clinical utility of models trained on permanent pathology is by applying them to frozen sections and fine-tune based on weaknesses. OBJECTIVE To qualitatively evaluate a deep learning model trained on permanent pathology to classify squamous cell carcinoma on Mohs surgery frozen sections to learn model shortcomings and inform retraining and fine-tuning. MATERIALS AND METHODS The authors trained a model for classification of tumor on 746 skin biopsy slides and tested it on 15 Mohs surgery frozen sections. The authors estimated performance metrics and compared the regions of interest generated by the model with the original H&E slides. RESULTS The model achieved an AUC-ROC of 0.985 and 0.796 for tumor classification in permanent pathology and in frozen sections, respectively. Regions of interest for frozen sections with scarce tumor areas were inaccurate, focusing on normal tissue for slides classified as false negative, or highlighting structures different from tumor (e.g., inflammation, muscle, and nerves) for slides classified as true positive. CONCLUSION Deep anatomical structures more commonly present in Mohs frozen pathology might represent data out-of-distribution for models trained on permanent pathology, potentially leading to inadequate model outputs.

ABSTRACTBackground/ObjectivesImmune‐mediated skin disorders, such as immunobullous dermatoses and leukocytoclastic vasculitis, rarely affect children. While direct immunofluorescence (DIF) biopsy is a standard diagnostic tool, limited data exist on pediatric DIF patterns, rates of positivity, pretest diagnostic concordance, and the relevance of biopsy site. This study sought to address these gaps.MethodsDIF data from all skin and mucosal specimens interpreted at Mayo Clinic's reference immunodermatology laboratory for patients aged 0–18 years (August 22, 2017 to November 30, 2023) were reviewed. DIF results were classified as positive if a characteristic pattern was seen and negative if the findings were negative or nondiagnostic.ResultsOf 986 pediatric DIF studies, 153 (15.5%) were positive and comparable to adult positivity rates (20.9%) during the same period. The most frequent DIF patterns were IgA‐predominant vasculitis (N = 85/153; 55.5%) and lichenoid tissue reaction (N = 21/153;13.7%). Concordance between pretest diagnosis and positive DIF results was highest for linear IgA bullous dermatosis (N = 7/7; 100%) and dermatitis herpetiformis (N = 6/6; 100%). Excluding these entities, DIF changed the pretest diagnosis in 16.7% (N = 19/114) of cases. While lower extremity biopsies were initially more likely to yield positive DIF results, this association disappeared when IgA vasculitis cases were excluded.ConclusionThe most frequent DIF pattern in children was that of IgA‐predominant vasculitis. Pediatric DIF positivity rates closely mirrored those of adults, supporting similar biopsy thresholds. DIF results differed from the pretest impression in a substantial percentage, supporting the value of DIF in select situations in the pediatric population. After controlling for IgA vasculitis, biopsy site was not associated with DIF positivity.