The group led by Dr. David Podhaizer from Boston University conducted a retrospective review of 29 patients with chronic liver disease who had undergone nontargeted ultrasound-guided liver biopsies. They identified regions of interest on two to three ultrasound images per patient; these areas were chosen from the right lobe of the liver and did not include vessels or bile ducts.
The researchers then analyzed these sections using a programming language that identified 45 texture features, and they compared the results with Ishak fibrosis scores (on a scale of 0 to 6, with 0 equal to no fibrosis and 6 equal to cirrhosis).
Based on liver biopsy, each Ishak fibrosis stage from 0 to 5 was seen in four patients; five patients demonstrated Ishak stage 6. The identified texture characteristics showed strong correlation between Ishak fibrosis stage, histogram data, and two particular ultrasound features: gray-level run length (GLRL) and gray-level gradient matrix (GLGM).
Texture analysis could be used with ultrasound as a noninvasive method to diagnose and monitor progression of liver fibrosis, Podhaizer's team concluded.
