Revolutionizing breast cancer diagnosis
Diagnosing breast cancer from histopathological images remains a complex task due to the presence of intricate tissue structures and imaging artifacts. This study presents a robust artificial intelligence-based (AI) framework for enhancing breast cancer diagnosis through a multi-stage computer-aided...
محفوظ في:
| المؤلف الرئيسي: | |
|---|---|
| مؤلفون آخرون: | , , , , |
| التنسيق: | article |
| منشور في: |
2025
|
| الوصول للمادة أونلاين: | http://hdl.handle.net/11675/14452 https: |
| الوسوم: |
إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
|
| الملخص: | Diagnosing breast cancer from histopathological images remains a complex task due to the presence of intricate tissue structures and imaging artifacts. This study presents a robust artificial intelligence-based (AI) framework for enhancing breast cancer diagnosis through a multi-stage computer-aided diagnosis (CAD) system. Utilizing Vision Transformers (ViT), the proposed framework performs hierarchical classification, beginning with benign versus malignant categorization, followed by a finer-grained subclassification within each category. The publicly available Breast Cancer Histopathological Image Classification dataset (BreakHis), comprising 9109 microscopic tumor images collected from 82 patients, was employed for training and evaluation. The CAD system utilizes two ViT models (ViT/P16/224 and ViT/P32/384) demonstrating high diagnostic performance across all image magnifications. Evaluation metrics include accuracy, precision, recall, specificity, F1 score, intersection over union (IoU), balanced accuracy (BAC), Matthews correlation coefficient (MCC), and weighted sum metric (WSM). The ViT/P16/224 model consistently outperformed the alternative, particularly in the second classification stage, achieving an accuracy of 98.99%, precision of 98.13%, and recall of 98% at 40X magnification, culminating in an F1 score of 98.04%. These results underscore the effectiveness of AI-driven methods for improving histopathological image interpretation and advancing diagnostic precision in breast cancer detection. |
|---|