Feasibility study on using the diffusion-weighted imaging liver pulsation artifact as marker of liver fibrosis

Read the full article See related articles

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

Background

Liver diffusion-weighted imaging (DWI) often shows signal loss in the left lobe due to cardiac pulsation, known as the pulsation artifact.

Purpose

This study was conducted to turn this bug into a feature. We hypothesized that the pulsation artifact, caused by mechanically transmitted waves, is influenced by tissue stiffness and, thereby, indicative of cirrhosis in the left liver lobe. Thus, we evaluated the potential of DWI pulsation artifacts to detect liver cirrhosis.

Study type

This study included both retrospective patients and prospective healthy volunteers.

Study population

Thirteen healthy volunteers, 16 patients with cirrhosis (Ishak 5 or 6), and three patients without cirrhosis (Ishak 0, 1, or 2) were included.

Field Strength/Sequence

Imaging was performed at 1.5 T and 3 T using echoplanar DWI of the liver with two b-values, b50 = 50 s/mm² and b800 = 800 s/mm².

Assessment

After manual segmentation of the left and right liver lobes, 18 first-order features were extracted from b50 and b800 images. We assumed the pulsation artifact was predominantly present in the left liver lobe in the b800 images. Thus, we computed the ratios of the intensity-normalized b800 features obtained from the left and right liver lobes. Additionally, we performed an intensity normalization to address the sometimes different signal intensities of left and right liver lobe. Then, we conducted a receiver operating characteristic (ROC) analysis to differentiate cirrhotic from non-cirrhotic livers.

Statistical tests

ROC analysis and leave-one-out bootstrapping were applied, both yielding area under the curve (AUC) values.

Results

Left liver pulsation artifacts were more pronounced in non-cirrhotic livers and largely absent in cirrhotic livers. The first quartile feature, i.e. the first quartile of the normalized signal, best distinguished cirrhotic from non-cirrhotic livers (AUC = 0.95; 95% confidence interval 0.87 - 1.00).

Data conclusion

The left liver DWI pulsation artifact reflects liver stiffness and could serve as a simple, intrinsic marker of liver cirrhosis. Analysis of the signal magnitude can differentiate cirrhotic from non-cirrhotic livers.

Article activity feed