Breast Cancer Detection using Hybrid Optimised Deep Learning Method
Sumalatha Lingamgunta, Loshma Gunisetti, Eliganti Ramalakshmi · International Journal of Ad Hoc and Ubiquitous Computing · 2024
Breast cancer (BC) is the abnormal growth of breast tissue, forming tumours. In existing research, the mammograms use X-rays to examine breast tissue. Although the radiation exposure per test was minimal, it accumulated over multiple screenings. This issue is overcome by developing the hybrid optimised deep learning method in this research. This research presents a novel hybrid approach known as Remora Krill Herd optimisation-based LeNet (RKHO-LeNet) for BC detection. Initially, the pre-processing is performed by using the Gaussian filter, and then the segmentation process is done by utilising the UNet++ model. Furthermore, the LeNet model is employed for BC detection, with the hyperparameters of the LeNet is tuned by RKHO. Furthermore, the RKHO is a combination of Remora optimisation (ROA) and Krill Herd optimisation (KHO). The proposed RKHO-LeNet model is validated using metrics such as accuracy, sensitivity, and specificity, which provides outcomes such as 0.929, 0.925 and 0.939.