Invasive Ductal Carcinoma (IDC) nuclei Classification using Mask RCNN | ||||
| Menoufia Journal of Electronic Engineering Research | ||||
| Volume 34, Issue 2, July 2025, Page 20-30 PDF (969.25 K) | ||||
| Document Type: Original Article | ||||
| DOI: 10.21608/mjeer.2025.351724.1103 | ||||
 View on SCiNiTO | ||||
| Authors | ||||
Amany Ibrahim  1; Hanaa` Torkey2; Ayman EL-SAYED 3
| ||||
| 1Computer Science and Engineering department, faculty of electrical engineering, Menoufia university | ||||
| 2Department of Computer Science and Engineering, Faculty of Electronic Engineering, Menoufia University, Egypt | ||||
| 3Computer science and Engineering, Faculty of Electronic Engineering, Menoufia University, Menouf 32952, Egypt. | ||||
| Abstract | ||||
| Breast cancer is the second most prevalent cancer globally and remains one of the leading causes of cancer-related mortality. Over the past few decades, the incidence of breast cancer has increased significantly, highlighting the critical need for early detection to improve survival rates. In response, researchers have been actively developing computer-aided diagnostic systems to assist in rapid and accurate diagnosis. Various datasets have been utilized in these efforts, leveraging the power of Artificial Intelligence (AI) to support radiologists in medical image analysis, ultimately enhancing patient diagnosis and treatment. Among the available diagnostic techniques, histopathology imaging remains the gold standard for detecting breast cancer with high accuracy. In this study, we employed ResNet- based architectures to implement a Mask Region-based Convolutional Neural Network (Mask R-CNN) for the automated detection of nuclei in histopathological breast cancer images. Following detection, the system classifies the cancer type, extracting multi-scale features using a combination of Feature Pyramid Networks (FPN) modules. To further enhance recognition accuracy, we utilized Region of Interest Align(RoIAlign), ensuring precise feature extraction. Experimental results demonstrate that our proposed approach not only delivers superior visual interpretability but also outperforms existing models in key performance metrics, achieving 97.7% accuracy, 97% recall, and a 96.7% F1 score. | ||||
| Keywords | ||||
| Breast cancer classification; Histological images; Mask regional convolutional network (Mask RCNN); Nuclei Segmentation; DC Classification | ||||
|
Statistics Article View: 115 PDF Download: 66 |
||||
