Breast Tumor Classification using Transfer Learning with Adaptive Crow Search Optimization
S. R. Sannasi Chakravarthy, Harikumar Rajaguru · 2023
Breast cancer is a prevalent form of cancer among women, with around 2,87,850 newer cases acknowledged in 2022. In addition, a substantial amount of female deaths, with 43,250 were accounted to this cancer type. Early detection plays a crucial role in reducing the higher mortality rate associated with breast cancer. However, precisely diagnosing this cancer type from mammography images involves the expertise of trained clinicians. For improving the diagnostic process, researchers have explored the potential of artificial intelligence techniques. The paper utilizes an Adaptive Crow-Search Optimization (AdCSO) algorithm to boost breast cancer diagnosis. The adaptability and premature convergence problems of conventional crow-search optimization algorithm is made improved using the AdCSO algorithm. The proposed framework consists of three main stages: data augmentation, feature extraction using EfficientNetV2 based on transfer learning, and optimized classification using a convolutional neural network (CNN). The mammograms from the INbreast dataset are employed for work evaluation. By leveraging transfer learning and an optimized CNN for classification, the proposed EfficientNetV2 with AdCSO-CNN provides a maximum classification performance of 97.05% accuracy as compared with other approaches. The proposed method is tested against the above problem with two output classes: benign and malignant.