#6695 THE ROLE OF A DEEP LEARING MODEL FOR EVALUATING HISTOPATHOLOGY OF ACUTE RENAL TUBULAR INJURY

Abstract Background and Aims Acute tubular injury is the most common intrinsic cause in acute kidney injury. Histopathological diagnosis may help distinguish between different types of aid treatment. Currently, there have not been any studies that uses deep learning model for assisting histopathological The aim our study classification segmentation identified four structures renal based on models. Method We used to evaluate degree (grade 0, grade 1 2) (glomerulus, healthy tubules, tubules with cast, necrotic tubules) specifically classify specific after cisplatin trained a using ResNet-18 DeepLabV3 ResNet-50 backbone mouse PAS-stained histopathology Results algorithm achieved area under receiver operating characteristics curve 0.98, 0.77, 0.80 1, 2, respectively. In structures, highest Intersection over Union Dice coefficient were obtained “glomerulus” class, followed by “healthy tubules”, “tubules cast”, “necrotic tubules” classes. overall performance all classes test set 0.7256, 0.8185. Conclusion predictive could identify injury, contribute more effective evaluation pathology.