AI driven analysis of MRI to measure health and disease progression in FSHD
Level 4 - case-series / case-control
Algorithm development and imaging validation study without a prospective comparative clinical cohort.
OpenAlex W4400371052 · doi:10.1038/s41598-024-65802-x
What was done
Researchers developed and optimized an artificial intelligence-based image segmentation method to measure muscle volume, fat fraction, fat fraction distribution, and elevated short-tau inversion recovery (STIR) signal in the musculature of patients with facioscapulohumeral muscular dystrophy (FSHD). The approach was evaluated using intra-rater, inter-rater, and scan-rescan reliability analyses, alongside 3D pixel-mapping across full muscle lengths.
What was found
The abstract reports that intra-rater, inter-rater, and scan-rescan analyses demonstrated the AI segmentation method to be robust and precise, revealing distinct intramuscular patterns of disease. However, the abstract provides no specific numerical metrics (such as Dice similarity coefficients, error rates, or intraclass correlation coefficients).
Why it matters
Automating full-length 3D muscle segmentation on MRI could streamline objective tracking of disease progression and biomarker response in clinical trials for FSHD, overcoming the bottleneck of manual segmentation.
Limits
The abstract does not state the number of participant scans evaluated or disclose quantitative performance statistics. The tool's current scope excludes upper-body musculature, and its utility for detecting longitudinal progression remains to be formally tested in prospective cohorts.
Cited by
- partial Springbok imaging software uses a 20- to 30-minute full-body MRI scan to create a three-dimensional model and individual volume analysis of every muscle in the body.