A Novel Deep Learning Technique Inspired by Biomedicine for the Diagnosis of BL Cancer

Colorectal metastases are among the riskiest liver tumors. Prediction of breast cancer, liver cancer, and MRI is done using deep learning techniques. In two different ways, this tactic adds to the body of literature. Liver lesions are initially removed (Yousri and Elaziz in IEEE Access 8:25306–125330, 2020) from computed tomography images using a hybrid segmentation technique called SegNet-UNet-ABC (Cinar and Eroglu in Comput Biol Med 11, 2021), which combines the SegNet network, UNet network, and ABC. The ABC method is hybridized alongside each network's hyperparameters because they have a significant impact on segmentation performance. Second, a feature extractor and classifier for liver lesions is the LeNet-5/ABC technique, which combines the LeNet-5 model and the ABC algorithm. The LeNet-5/ABC technique uses the ABC to identify the ideal topology for constructing the LeNet-5 network since network structure influences how rapidly learning occurs for categorization. In this paper (Mohakud and Rajashree in Intell Cloud Comput 10:737–744, 2021), a deep neural network (DNN) model is developed. The suggested technique outperforms approaches looked at (Guttery and Satapathy in Inf Process Manag 10:1–36, 2021) by other researchers in terms of automatic detection speed, classification precision, and processing time required for automatic detection and classification of masses. These features are used to classify masses using Fisher's Linear Discriminate Analysis, Support Vector Machine, Multilayer Perception, and two training methods: Liebenberg-Marquardt (MLP-LM) and Bayesian Regularization (MLP-RBF).