An advanced AI-driven deep learning framework for early detection and precise diagnosis of breast cancer from medical images.
Authors
Affiliations (8)
Affiliations (8)
- Department of Electronics and Communication Engineering, SRM Institute of Science and Technology, Ramapuram, Chennai, India. Electronic address: [email protected].
- Department of Computer Science and Engineering, Vel Tech Rangarajan Dr.Sagunthala R&D Institute of Science and Technology, Avadi, Chennai, Tamil Nadu, 600 062, India. Electronic address: [email protected].
- Department of Information Technology, Meenakshi Sundararajan Engineering College, Kodambakkam, Chennai, India. Electronic address: [email protected].
- Department of Electronics and Communication Engineering Kamaraj College of Engineering & Technology, 625701, India. Electronic address: [email protected].
- Department of Electronics and Communication Engineering, Kamaraj College of Engineering & Technology, Madurai, Tamil Nadu, India. Electronic address: [email protected].
- Department of Electronics and Communication Engineering, JB Institute of Engineering and Technology, Hyderabad, Telangana, India. Electronic address: [email protected].
- Department of Information Technology, Sri Muthukumaran Institute of Technology , Chennai -69. Electronic address: [email protected].
- Centre for Advanced Wireless Integrated Technology, Chennai Institute of Technology, Chennai, India. Electronic address: [email protected].
Abstract
A malignant disorder known as breast cancer (BC) arises when cells in breast tissue grow out of control, frequently due to genetic, hormonal, or environmental reasons. Reducing death rates and increasing treatment outcomes depend on early and precise identification. However, a significant challenge lies in the variability of breast tissue among patients, dependence on high-resolution imaging, and difficulty interpreting subtle abnormalities, which delays accurate diagnosis and timely treatment in clinical settings. In this paper, an advanced AI-driven deep learning pipeline for highly accurate medical image analysis, enabling early detection and precision diagnosis of breast cancer (PDBC-KARN), is proposed. Initially, the input image is collected from the breast cancer detection dataset. Then, images were pre-processed by utilizing the Bilinear Double-Order Filter (BDOF) for image resizing and normalizing the image. Then the pre-processed images are given to the Kolmogorov-Arnold Recurrent Network (KARN) to detect breast cancer and are classified as benign and malignant. The Stellar Oscillation Optimizer (SOO) is utilized to optimize the weight parameters of KARN. The proposed PDBC-KARN approach is implemented in Python and establishes substantial improvements in accuracy, precision, recall, specificity, loss, training and validation accuracy, training and validation recall, confusion matrix, and F1-score. The proposed PDBC-KARN approach achieves an accuracy of 98.5% for benign and 97.0% for malignant, an f1-score of 98.9% for benign and 97.5% for malignant cases with existing methods, like dual-view deep learning (DL) techniques for accurate breast cancer (BC) detection in mammograms (BCDM- DNN), RI-ViT: a multi-scale hybrid technique depending upon vision transformer for detection of breast cancer in histopathological images(BCDH- CNN) PSO-optimized fractional order CNNs for enhanced detection of breast cancer (BCD- FOCNN) respectively.