What type of study is this?

September 10, 2025PLoS ONEOpen Access

Liver MRI proton density fat fraction inference from contrast enhanced CT images using deep learning: A proof-of-concept study

View Full Paper

Ask AI

Bookmark

Authors

Proof-of-concept study demonstrates deep learning can predict liver fat fraction from CT, suggesting potential for improved diagnosis.

The deep learning model showed an MAE of 0.56 and an RMSE of 0.77 when comparing predicted MRI-PDFF to ground truth values.
Kappa analysis indicated substantial agreement (kappa = 0.75) between DNN-predicted categorical steatosis grades and clinical ground truth.
Despite limitations in inferring exact PDFF values, robust classification of fat fraction grades was achieved, outperforming previous methods.
The study cohort comprised 94 patients, with a mean PDFF of 3.8% and a range of 0.2-22.3%.

Nasir et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark