Tumor-Infiltrating Lymphocyte Recognition in Primary Melanoma by Deep Learning Convolutional Neural Network

: The presence of tumor-infiltrating lymphocytes (TILs) has been associated with a favorable prognosis of primary melanoma (PM). The recent development of the artificial intelligence (AI)-based approach in digital pathology was proposed for the standardized assessment of TILs on hematoxylin and eosin-stained images [whole slide images (WSI)]. Here, we applied a new convolution neural network (CNN) analysis of PM WSI to automatically assess the infiltration of TILs and extract a TIL score. A CNN was trained and validated in a retrospective cohort of 307 PMs including a training set (237 WSI, 57,758 patches) and an independent testing set (70 WSI, 29,533 patches). After the classification of tumor patches by the presence or absence of TILs, we identified an AI-based TIL density index (AI-TIL). The proposed CNN showed high performance in recognizing TILs in PM WSI, showing specificity and sensitivity of 100% on the testing set. We showed that the AI-based TIL index correlated with conventional TIL evaluation and clinical outcome. The AI-TIL index was an independent prognostic marker associated directly with a favorable prognosis. A fully automated and standardized AI-TIL appears to be superior to conventional methods at differentiating the PM clinical outcome. Further studies are required to develop an easy-to-use tool to assist pathologists to assess TILs in the clinical evaluation of solid tumors.