Automated diagnosis of 7 canine skin tumors using machine learning on H&E-stained whole slide images

Fragoso-Garcia M, Wilm F, Klopfleisch R, et al. Automated diagnosis of 7 canine skin tumors using machine learning on H&E-stained whole slide images. Vet Pathol. 2023;60(6). doi:10.1177/03009858231189205

This study evaluates a machine-learning-based algorithm for diagnosing seven common canine skin tumors using hematoxylin and eosin-stained slides. A convolutional neural network was trained on 350 digitized and annotated whole-slide images (WSIs) from seven tumor types and six normal skin structures. The algorithm achieved a slide-level classification accuracy of 95%, closely matching the 98% accuracy of six pathologists. The findings support the use of artificial intelligence as a diagnostic aid in veterinary oncology.