Estimation of Lymph Node Metastasis in Cervical Cancer Using Deep Learning
Ryunosuke Kori, Kento Morita, Kenta K. Yoshida, Eiji Kondo, Tetushi Wakabayashi · 2024
Cervical cancer can metastasize to lymph nodes as it progresses. The presence or absence of lymph node metastasis (LNM) is crucial for determining the cancer stage and selecting appropriate treatment methods. However, the diagnostic accuracy of non-invasive imaging for LNM does not match that of biopsy, potentially leading to over- or under-treatment. Therefore, this study aims to develop a method to accurately predict the presence of LNM in cervical cancer, aiding in the determination of appropriate treatment strategies. In this research, we utilize artificial neural network (ANN) and Transformers to predict LNM. Experiments were conducted using 47 metastasis-positive and 95 metastasis-negative cases. The results demonstrated that both the ANN and Transformer models outperformed the 3D CNN (accuracy: 0.817, precision: 0.746, recall: 0.705, F1 score: 0.714), with the Transformer achieving the highest F1 score of 0.867.