Identification of Advanced Breast Cancer by Using Aggregate Deep Learning Method
V. J. Chakravarthy, Sangeetha Varadhan, Sridevi Sakhamuri, R. Vijayabharathi, Fathima. S. K, Saritha K · 2024
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.