A Comprehensive Review of Image-Based Breast Cancer Detection Techniques: Challenges and Perspectives

Breast cancer (BC) is the most common tumor present in women worldwide. An effective way to detect BC earlier is essential by using BC screening programs as it crucially reduces cancer rate and mortality rates. BC detection approaches rely on medical imaging, which is effective in identifying tumors. BC imaging data, such as mammograms, ultrasound, and magnetic resonance imaging (MRI), vary in format, quality, and modality, which poses a challenge in consistent and accurate detection. This survey analyzes many research papers, which are focused on widely used BC detection approaches based on images, and offers method-wise reviews, like deep learning, fuzzy-based, optimization-based, watershed-based, machine learning–based, and threshold-based methods. An analysis is engaged based on the classification of research methods, tools utilized, year of publication, datasets, and performance metrics for the detection of BC employing images. From the analysis performed, it is observed that machine learning methods are the most utilized methods, accuracy is the frequently used performance metric, and the deep learning approach is the most prevailing method for BC detection. This classification gives the strengths and weaknesses of the proposed method, thereby serving as a foundation for developing next-generation novel effectual BC detection methods by utilizing images.