| Autor: |
Ashwini, P., Suguna, N., Vadivelan, N. |
| Zdroj: |
Multimedia Tools & Applications; Apr2024, Vol. 83 Issue 14, p41785-41803, 19p |
| Abstrakt: |
Early and accurate diagnosis of breast cancer (BC) using digital mammograms can improve disease detection accuracy. Medical images can be detected, segmented, and classified for the design of computer-aided diagnosis (CAD) models which assist radiologists in accurately diagnosing breast lesions. Therefore, this study proposes an Improved Bald Eagle Search Optimization with Entropy-based Deep Feature Fusion (IBESO-EDFFM) model for BC Diagnosis on Digital Mammograms. The goal of the IBESO-EDFFM technique lies in the proper detection and segmentation of BC using feature fusion and hyperparameter tuning concepts. For the feature extraction process, the IBESO-EDFFM technique employs an entropy-based feature fusion process, comprising three deep learning models namely Capsule Network (CapsNet), Inception v3, and EfficientNet. Besides, Improved Bald Eagle Search Optimization (IBESO) with Bidirectional-Quasi Recurrent Neural Network (BiQRNN) is utilized for the identification and classification of breast cancer. Finally, a fully convolutional network with RMSProp optimizer is exploited for the segmentation of abnormal regions from the classified images. The experimental result analysis of the IBESO-EDFFM technique is tested on the MIAS mammography dataset from the Kaggle repository and the comparative results show the better performance of the IBESO-EDFFM technique over recent approaches with maximum accuracy of 98.96%. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
Full text from SpringerLink