|
|
 |
| |
|
|
|
Optimizing Histopathological Imaging Analysis for Breast Cancer Detection using Enhanced Pelican Optimization Algorithm and Deep Feature Fusion |
|
|
|
PP: 805-818 |
|
|
|
doi:10.18576/amis/190407
|
|
|
|
Author(s) |
|
|
|
Arwa Darwish Alzughaibi,
Maryam Alsolami,
Mohammed Alshahrani,
Sultan Ahmed Almalki,
Mohammed Al-Jabbar,
Randa Alharbi,
|
|
|
|
Abstract |
|
|
| Breast cancer (BC) is a major common type of cancer in women. Earlier and accurate diagnoses of BC may enhance the treatment chances and reduce the mortality rate. Therefore, the emergence of reliable and accurate Computer-Aided Diagnosis (CAD) system BC images is a pressing concern for earlier diagnoses. Classical algorithms limit pathologists’ skills and are time- consuming. Automated histopathological image (HI) classification is an area of interest that might reduce the risk of mistakes and speed up BC diagnoses. Histopathology uses a biopsy to capture images of the diseased tissue. Lately, deep learning (DL) techniques have shown great efficiency in different medical imaging applications, such as the processing of HI. Therefore, this article develops an automated Histopathological Imaging Analysis for BC Detection using the Enhanced Pelican Optimization Algorithm and Deep Feature Fusion (BCD-EPOADFF) technique. The purpose of the BCD-EPOADFF technique is to examine the histopathological images for the detection and classification of BC. In the BCD-EPOADFF technique, the adaptive median filtering (AMF) technique can be applied to get rid of the noise and enhance the image quality. In the BCD-EPOADFF technique, a deep feature fusion process takes place comprising three DL models namely Residual Networks (ResNet), EfficientNet, and InceptionNet. Moreover, the hyperparameter tuning of the DL models takes place using EPOA which incorporates the traditional POA with oppositional-based learning (OBL) concept. Finally, the fuzzy neural network (FNN) model can be employed for automated detection and classification of BC. To validate the enriched performance of the BCD-EPOADFF method, wide-ranging simulations were involved. The experimental results stated the superior performance of the BCD-EPOADFF method over other models for distinct measures.
|
|
|
|
|
|
|
|
