BREAST CANCER CLASSIFICATION USING HYBRID DEEP LEARNING MODELS
International Research Journal of Modernization in Engineering Technology and Science · 2024
Over the years, breast cancer has become the most deadly disease affecting women worldwide.Early detection of breast cancer lowers mortality and raises the likelihood of a full recovery.Medical imaging-based breast cancer screening techniques are being developed by researchers worldwide.The rapid progress of deep learning algorithms has attracted the interest of many in the medical imaging sector.Ultrasound images were used in this study to identify breast cancer.Using different deep CNN (ResNet-50, DenseNet-121, and a simple CNN) models as base classifiers, we have employed four Ultrasound imaging datasets with 8000 images for training and 1000 images for validation that contain photos of normal, benign, and malignant conditions.The recommended approach makes use of an ensemble made up of the ResNet-50 models, CNN, as well as DenseNet-121.Although our aggregation of their data improves the accuracy to over 95%, it is still not very accurate.We are sure that the suggested approach will be very beneficial to doctors in identifying cases of breast cancer early on, possibly leading to a timely diagnosis.