Breast Cancer Classification from Histology Images using Dense Residual Capsule Network

• Version
• Download 12
• File Size 0.00 KB
• File Count 1
• Create Date 13 December, 2025
• Last Updated 13 December, 2025

Breast Cancer Classification from Histology Images using Dense Residual Capsule Network

DOI : 10.46335/IJIES.2024.9.2.11

Authors: Pallavi B. Salunkhe, Dr. Pravin S. Patil

Abstract – Breast cancer stands as a prominent form of cancer predominantly affecting women globally, with a concerning trend of rising incidence in developing nations. The detection and diagnosis of breast cancer can be achieved through non-invasive methods and biopsy. Non-invasive methods primarily include imaging procedures such as diagnostic mammograms, Magnetic Resonance Imaging (MRI) of the breast, breast ultrasound, and thermography. While imaging procedures are commonly used for cancer screening, biopsy remains the most reliable method to confirm the presence of cancer. Histopathological analysis, vital for diagnosing cancer, requires specialized expertise and is time-consuming. It heavily relies on the experience of pathologists and can be affected by factors like fatigue and decreased attention. Recent advancements in image processing have significantly improved the accuracy of diagnosis. an image can be analysed for classification of Malignant and normal cells in a different datasets of breast cancer. Several machine learning/deep learning based approaches are being applied for analysis of microscopic images. Early identification holds significant importance in detection and treating breast cancer, ultimately reducing mortality rates. Hence, this study presents an automated technique for detecting breast cancer through the application of deep learning on histopathological images. Proposed method employs breast histology images, which were categorized into different classes using the Attention Residual Dense Capsule Network. This innovative approach was developed and evaluated using the Python platform and the BreakHis dataset. Performance assessment was conducted using various metrics including f-measure, recall, specificity, precision, and accuracy to gauge the efficacy of the proposed methodology.