DeepGleason: a System for Automated Gleason Grading of Prostate Cancer Using Deep Neural Networks

Advances in digital pathology and artificial intelligence (AI) offerpromising opportunities for clinical decision support and enhancing diagnosticworkflows. Previous studies already demonstrated AI's potential for automatedGleason grading, but lack state-of-the-art methodology and model reusability.To address this issue, we propose DeepGleason: an open-source deep neuralnetwork based image classification system for automated Gleason grading usingwhole-slide histopathology images from prostate tissue sections. Implementedwith the standardized AUCMEDI framework, our tool employs a tile-wiseclassification approach utilizing fine-tuned image preprocessing techniques incombination with a ConvNeXt architecture which was compared to variousstate-of-the-art architectures. The neural network model was trained andvalidated on an in-house dataset of 34,264 annotated tiles from 369 prostatecarcinoma slides. We demonstrated that DeepGleason is capable of highlyaccurate and reliable Gleason grading with a macro-averaged F1-score of 0.806,AUC of 0.991, and Accuracy of 0.974. The internal architecture comparisonrevealed that the ConvNeXt model was superior performance-wise on our datasetto established and other modern architectures like transformers. Furthermore,we were able to outperform the current state-of-the-art in tile-wisefine-classification with a sensitivity and specificity of 0.94 and 0.98 forbenign vs malignant detection as well as of 0.91 and 0.75 for Gleason 3 vsGleason 4 5 classification, respectively. Our tool contributes to the wideradoption of AI-based Gleason grading within the research community and pavesthe way for broader clinical application of deep learning models in digitalpathology. DeepGleason is open-source and publicly available for researchapplication in the following Git repository:https://github.com/frankkramer-lab/DeepGleason.