Key result
An automated groupwise registration method for pancreas subsegmentation achieved high accuracy, with no significant difference in Dice Similarity Coefficient compared to manual expert annotations.
Why the study?
Existing automated whole-organ pancreas segmentation provides overall quantification but fails to address spatial heterogeneity in pancreatic disease.
Population
100 subjects for template creation and 50 independent subjects for validation from the UK Biobank imaging substudy
Comparison
Automated subsegmentation method vs expert manual annotations
Design
Retrospective validation study
Authors
Loading...
May aid efficient pancreas imaging analysis; leaves open prospective clinical validation before adoption.
Cross-Sectional (n=150)
Yes
Absolute Event Rate: 0.965% vs 0.968%
p-value: p=0.4358
An automated groupwise registration method successfully segments the pancreas into head, body, and tail on MRI with performance comparable to expert manual annotation, enabling regional quantitative analysis of pancreatic disease.
Bagur et al. (2021) conducted a cross-sectional in Healthy (nominally healthy cohort) (n=150). Automated groupwise registration method for pancreas subsegmentation vs. Manual expert annotation was evaluated on Dice Similarity Coefficient (DSC) for the pancreas head comparing automated groupwise registration to manual annotation (p=0.4358). An automated groupwise registration method for pancreas subsegmentation achieved high accuracy, with no significant difference in Dice Similarity Coefficient compared to manual expert annotations.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: