Metabolic-associated fatty liver voxel-based quantification on CT images using a contrast adapted automatic tool

Area of Science:

• Radiology
• Medical Imaging
• Hepatology

Background:

• Metabolic-dysfunction associated fatty liver disease (MAFLD) is prevalent and requires accurate diagnosis.
• Current diagnostic methods like liver biopsy are invasive, and CT-based assessments have limitations.

Purpose of the Study:

• To develop an automated CT-based method for quantifying liver fat percentage.
• To overcome diagnostic deficiencies in contrast-enhanced (CE) and non-contrast-enhanced (NCE) CT assessments for MAFLD.

Main Methods:

• Automated segmentation of liver and spleen using nn-UNet on CE- and NCE-CT images.
• Calculation of radiodensity benchmarks: liver mean, liver-to-spleen ratio, and liver-spleen difference.
• Validation using vibration-controlled transient elastography (VCTE) and development of a patient classification method.

Main Results:

• The average of proposed benchmarks, particularly the liver-to-spleen ratio (CE) and liver-spleen difference (NCE), achieved the best accuracy.
• Automated whole-organ segmentation outperformed manual region-of-interest drawing for fat quantification.
• Biochemical data aided in stratifying atypical patients.

Conclusions:

• The developed method offers an automated, non-invasive alternative for MAFLD evaluation, addressing CT-related weaknesses.
• It enables precise fat quantification and early detection of abnormal CT patterns, avoiding unnecessary radiation exposure.
• This tool serves as a surrogate for assessing fatty liver in primary MAFLD evaluation, complementing elastography data.