Image processing in healthcare is an important field of research and a key factor in improving the accuracy, quality, and efficiency of medical diagnoses. The variety of medical imaging modalities, image quality variations, and the complexity of pathological feature identification across different conditions contribute to significant challenges for reliable and reproducible diagnosis. This chapter focused on the role of machine learning (ML) in real-life health applications such as tumor, lesion, or Region of Interest (RoI) detection, segmentation, and possible classification of several medical diseases such as lung and brain cancer. The chapter addressed the latest techniques in this area. Additionally, it examines key aspects including objective, preprocessing procedures, detection and segmentation approaches, registration methods, model evaluation, advantages, limitations, and prospective enhancements. It also addresses challenges in applying machine learning, including data scarcity, model interpretability, and generalization. Furthermore, it highlights the significance of machine learning in personalized medicine, accentuating its capacity to improve patient-specific diagnostic and therapeutic approaches. Furthermore, the chapter emphasizes some future directions pertinent to addressing several current challenges, improving model robustness, and, finally, increasing clinical application. This chapter serves as a valuable resource in recognizing the latest promising insights into key areas for future advancements in machine learning applications in image processing and healthcare.