Abstract
Comparative Experimental Investigation and Application of Five Classic Pre-Trained Deep Convolutional Neural Networks via Transfer Learning for Diagnosis of Breast Cancer
Highlights
Breast cancer is the most common cancer among women, and this type of cancer is recognized as one of the major causes of death (Cuzick et al, 2020; Lei et al, 2021) (Patnaik, Byers, DiGuiseppi, Dabelea, & Denberg, 2011) (Oh et al, 2020)
In order to make pre-trained deep convolutional neural network models (DCNN) models suitable for the purpose of this study, some layers were updated according to the new situation by using the transfer learning technique
The classification layer is the last layer connected after the softmax layer in a DCNN architecture, and the cross-entropy loss is calculated in this layer
Summary
Breast cancer is the most common cancer among women, and this type of cancer is recognized as one of the major causes of death (Cuzick et al, 2020; Lei et al, 2021) (Patnaik, Byers, DiGuiseppi, Dabelea, & Denberg, 2011) (Oh et al, 2020). 5 classical DCNN models (Alexnet, Googlenet, Resnet[18], Squeezenet, and Shufflenet) were used for breast cancer diagnosis and classification, using the transfer learning approach. Quadruple classification as cancerous, normal, actionable and benign, as well as a dual classification and diagnostic study as ‘actionable + cancer’ and ‘normal + benign’ were carried out For this purpose, five different DCNN models have been implemented by changing and redesigning some layers, owing to the transfer learning approach. The motivation of the study is to prove that classical deep learning methods can be used in the diagnosis and classification of breast cancer by using the transfer learning approach and to shed light on the future studies in this field
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
