Advancing Tumor Cell Classification and Segmentation in Ki-67 Images: A Systematic Review of Deep Learning Approaches

Breast cancer is one of the most diagnosed cancers, transforming it into a matter of great concern for public health. Early detection increases the probability of effective therapy and survival. However, it is a difficult and time-consuming procedure that relies on pathologists’ expertise. Over the past decade, Recent progress in deep learning methods has offered promising potential, compared to other machine learning approaches. Deep learning has shown greater performance in tackling several issues in many disciplines of medicine, Consequently, this has led to higher diagnostic precision and enhanced patient care. This comprehensive review extensively evaluates the most recent deep-learning algorithms for tumor cell classification and segmentation in Ki-67 images based on a systematic examination of the relevant literature. We outlined the procedures and findings of several studies, critically examine the approaches of various deep learning techniques, and indicate potential future research prospects. We also examined the potential consequences of using deep learning-based tumor cell detection in real-world clinical practice. Given the numerous benefits of deep learning techniques, deep learning will become increasingly more widely used in a wide range of diverse sectors of medicine in the future decades. This indicates the effectiveness of the strategy in offering a useful tool for breast cancer multi-classification in clinical settings.