The Future of Surgical Diagnostics: Artificial Intelligence-Enhanced Detection of Ganglion Cells for Hirschsprung Disease

Abstract

Hirschsprung disease, a congenital disease characterized by the absence of ganglion cells, presents significant surgical challenges. Addressing a critical gap in intraoperative diagnostics, we introduce transformative artificial intelligence approach that significantly enhances the detection of ganglion cells in frozen sections. The data set comprises 366 frozen and 302 formalin-fixed-paraffin-embedded hematoxylin and eosin–stained slides obtained from 164 patients from 3 centers. The ganglion cells were annotated on the whole-slide images (WSIs) using bounding boxes. Tissue regions within WSIs were segmented and split into patches of 2000 × 2000 pixels. A deep learning pipeline utilizing ResNet-50 model for feature extraction and gradient-weighted class activation mapping algorithm to generate heatmaps for ganglion cell localization was employed. The binary classification performance of the model was evaluated on independent test cohorts. In the multireader study, 10 pathologists assessed 50 frozen WSIs, with 25 slides containing ganglion cells, and 25 slides without. In the first phase of the study, pathologists evaluated the slides as a routine practice. After a 2-week washout period, pathologists re-evaluated the same WSIs along with the 4 patches with the highest probability of containing ganglion cells. The proposed deep learning approach achieved an accuracy of 91.3%, 92.8%, and 90.1% in detecting ganglion cells within WSIs in the test data set obtained from centers. In the reader study, on average, the pathologists' diagnostic accuracy increased from 77% to 85.8% with the model’s heatmap support, whereas the diagnosis time decreased from an average of 139.7 to 70.5 seconds. Notably, when applied in real-world settings with a group of pathologists, our model’s integration brought about substantial improvement in diagnosis precision and reduced the time required for diagnoses by half. This notable advance in artificial intelligence–driven diagnostics not only sets a new standard for surgical decision making in Hirschsprung disease but also creates opportunities for its wider implementation in various clinical settings, highlighting its pivotal role in enhancing the efficacy and accuracy of frozen sections analyses.