May 10, 2018
2 min read

New imaging model provides superior fibrosis stage diagnosis

You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

We were unable to process your request. Please try again later. If you continue to have this issue please contact

Deep learning radiomics of elastography demonstrated superior predictive value in assessing cirrhosis and advanced fibrosis compared with 2D shear wave elastography and biomarkers among patients with chronic hepatitis B, according to recently published data.

“Utilizing the data acquired from limited number of hospitals to train and establish the [deep learning radiomics of elastography (DLRE)] model is likely to be sufficient in applying it for assessing liver fibrosis stages in other hospitals with a consistent accuracy,” Kun Wang, MD, from the Third Affiliated Hospital of Sun Yat-sen University in China, and colleagues wrote.

To determine the accuracy of DLRE compared with 2D-SWE, aspartate-to-platelet ratio index and Fibrosis-4 index, the researchers randomly assigned 266 patients to a training cohort — designed to optimize the radiomic model’s parameters — and 132 patients into a validation cohort.

Baseline characteristics and distribution of fibrosis stages were similar between the two groups.

In the training cohort, DLRE demonstrated the highest diagnostic accuracy for fibrosis stage 2 (AUC = 0.99; 95% CI, 0.97-1), fibrosis stage 3 (AUC = 0.99; 95% CI, 0.97-1), and fibrosis stage 4 (AUC = 1; 95% CI, 0.99-1) compared with the other diagnostic methods, including 2D-SWE, which had the second highest accuracy for stage 2 (AUC = 0.74; 95% CI, 0.68-0.79), stage 3 (AUC = 0.81; 95% CI, 0.75-0.85) and stage 4 (AUC = 0.87; 95% CI, 0.83-0.91).

DLRE sensitivity and specificity analyses were also superior to 2D-SWE and other biomarkers for fibrosis stage 2, stage 3 and stage 4.

In the validation cohort, DLRE continued to demonstrate higher diagnostic accuracy for fibrosis stage 2 (AUC = 0.85; 95% CI, 0.81-0.89), stage 3 (AUC = 0.98; 95% CI, 0.96-1) and stage 4 (AUC = 0.97; 95% CI, 0.94-0.99), which were significantly higher than the other methods (P < .01).

In a subanalysis of all enrolled patients, the researchers observed no significant differences between DLRE and 2D-SWE at fibrosis stage 2, stage 3 or stage 4 regarding different alanine aminotransferase or BMI levels. The AUC for 2D-SWE in fibrosis stage 4 among patients without severe inflammation was significantly higher than patients with severe inflammation (P < .001), whereas DLRE showed no significant difference between inflammation subgroups.

“In the assessment of significant fibrosis, the performance of DLRE became worse in the validation cohort, even though it was very accurate in the training,” the researchers wrote. “One possible way to overcome this challenge is to integrate multiple strategies for fibrosis classification. If DLRE can be further optimized and integrated with other approaches, such as [liver stiffness measurement] by 2D-SWE and biomarkers, it might be possible to achieve a better performance in classifying [fibrosis stage 2 or higher].” – by Talitha Bennett

Disclosure: The authors report no relevant financial disclosures.