Identification of Advanced Breast Cancer by Using Aggregate Deep Learning Method

Metastasis is the main factor contributing to death in breast cancer patients. Faster and more accurate deep learning algorithms are being researched as potential replacements for the current, labor-intensive techniques used to diagnose metastases from lymph nodes. For testing and training purposes, a total of 220025 whole-slide pictures representing the lymph nodes of two cohorts of patients were identified. In order to detect metastatic cancer, we used a hybrid convolutional neural network model. A total of 57458 unlabeled images were utilized in order to verify the accuracy, sensitivity, specificity, and P-value of our diagnostic method. The DL-based approach was created to automatically and selectively assess and identify metastasizing lymph nodes. In the quantification procedure, accuracy was 98.84%. Moreover, the accuracy rates for VGG16 and Recall were 92.42% and 91.25%, respectively. The differentiation levels of metastatic cancer may have an impact on recognition performance, according to later study. Our devised diagnostic complex showed a high degree of efficacy and accuracy for lymph node diagnosis. Patients with breast cancer may find it simpler to perform pathological screening for metastases thanks to our innovative DL-based method.