E2271. Application of AI in the Assessment, Diagnosis and Management of Non-Alcoholic Fatty Liver Disease/Non-Alcoholic Steatohepatitis
Authors
Edward Florez;
University of Mississippi
Johnny Yang;
University of Mississippi
John Overton;
University of Mississippi
Quinn Cottone;
University of Mississippi
Zainab Ahmad;
University of Mississippi
Elizabeth Kerby;
University of Mississippi
Candace Howard;
University of Mississippi
Background
Non-alcoholic fatty liver disease/non-alcoholic steatohepatitis (NAFLD/NASH) is the leading cause of liver fibrosis and cirrhosis, affecting one in three people, and is especially common among patients with type diabetes 2 and/or obesity. Approximately 25% of people with NAFLD will develop a more serious liver inflammation known as NASH, which can lead to cirrhosis or liver cancer. Fatty liver disease will soon be the leading cause for liver transplantation. Accurate assessment of liver fibrosis is important for predicting disease outcomes and assessing therapeutic responses in clinical practice and clinical trials. Disadvantages and difficulties of invasive and non-invasive metrics used in the diagnosis and treatment of fatty liver disease creates a perfect scenario for the use of artificial intelligence (AI).
Educational Goals / Teaching Points
To understand the principles and different techniques for the detection of liver fibrosis. To compare the performance of ultrasound-based transient elastography (TE) vs magnetic resonance elastography (MRE) for the diagnosis of liver stiffness in patients with fatty liver disease. To review the most prominent applications of AI in the spectrum of fatty liver disease from NAFLD to NASH to fibrosis and cirrhosis from diagnosis, to staging and disease severity determination, and monitoring drug treatment efficacy. To explore the pros and cons of the application of AI in patients with fatty liver disease and/or fibrosis and possible future directions in this field.
Key Anatomic/Physiologic Issues and Imaging Findings/Techniques
Although non-invasive approaches such as TE and MRE are preferably used when possible, histological evaluation of liver fibrosis through liver biopsy is the gold standard for determining the degree of infiltrative processes such as steatosis and fibrosis.
Conclusion
Large input data sets, such as demographic information, laboratory results, and different imaging modalities, showed excellent performance in the training, improvement, and validation of diverse AI-assisted methods used in NAFLD, NASH, and fibrosis detection among patients with NAFLD. The results reported in previous studies prove that the benefits and robustness of using AI-assisted compared head-to-head with conventional techniques are notably superior in accuracy, performance, and precision.
