Ultrasound Imaging in Liver Disease: Systematic Insights into Classification Technique

Noninvasive and inexpensive, ultrasound imaging serves an essential function in the diagnosis of liver diseases. However, problems such as image variability, operator dependency, and ability to separate initial stages of disease diminish its efficiency. These issues best illustrate the growing imperative for better healthcare approaches that are not only sustainable, but also affordable in the context of low-resource locations. This study therefore aims at systematically compiling and reports on the most recent works in classification techniques, that is deep learning (DL), machine learning (ML) and transfer learning (TL) which are considered to provide solutions to above stated challenges. These AI formulations make them compatible with Sustainable Development Goals (SDGs), particularly, Good Health and Well-Being since they offer the expensive and innovative methods of increasing the reliability of diagnostic outcomes and minimizing a human operator’s influence at a lowered price. These innovations enable the production of low-cost and easily scalable diagnostic solutions for liver disease, especially among the population that has no access to it. This review will demarcate and assess the different AI approaches used on the ultrasound imaging data to overcome the present barriers to ultrasound diagnosis in LRI while directing the future research towards efficient, inexpensive, and scalable resourceful diagnostic framework.