Incorporating artificial intelligence in portable infrared thermal imaging for the diagnosis and staging of nonalcoholic fatty liver disease

Abstract

Metabolic Dysfunction-Associated Steatotic Liver Disease (MASLD) is one of the most prevalent chronic liver diseases worldwide. Thermal imaging combined with advanced image-processing and machine learning analysis accurately classified disease status in a study on mice; this study aimed to develop this tool for humans. This prospective study included 46 patients who underwent liver biopsy. Liver thermal imaging was performed on the same day as liver biopsy. We developed an image-processing algorithm that measured the relative spatial thermal variation across the skin covering the liver. The texture parameters obtained from the thermal images were input into the machine learning algorithm. Patients were diagnosed with MASLD and stratified according to nonalcoholic fatty liver disease activity score (NAS) and fibrosis stage using the METAVIR score. Twenty-one of 46 patients were diagnosed with MASLD. Using thermal imaging followed by processing, detection accuracy for patients with NAS >4 was 0.72.