Clinical diagnostic systems based on machine learning and deep learning

Abstract Technology-enabled healthcare systems are becoming more intelligent, efficient, and effective with the advancement in machine learning (ML) and deep learning (DL) techniques. The proposed methodology is an application of the healthcare system that falls under the category of medical imaging, which helps in clinical diagnosis. As ultrasound imaging is safe, painless, and the patient is not exposed to ionizing radiation, it allows real-time imaging. Because of these features, ultrasound imaging is the most frequently preferred medical imaging in clinical practice for diagnostic purposes. Understanding and interpreting ultrasound images requires well-trained radiologists, and it requires more time in the diagnosis process. Hence an advisory system is needed that helps in identifying organs and any associated abnormalities in a short duration. The proposed automated healthcare system using ultrasound imaging facilitates an advisory system for the diagnosis of people suffering from abnormalities in the organs. In order to develop an automated healthcare system, exhaustive normal and abnormal intraabdominal ultrasound images are collected and preprocessed. Localization of the region of interest (ROI) is performed with an effective segmentation method. The final step is experimenting with ROI images to extract features to identify intraabdominal organs and their abnormalities. To conclude, in the chapter we have noted various challenges with ultrasound images and ML and DL techniques.