The use of an automated sarcopenia measurement tool as a predictor of outcome in patients with abdominal aortic aneurysms (SPAR)
File(s)
Author(s)
Sudarsanam, Abhilash
Type
Thesis
Abstract
Introduction
Sarcopenia is associated with adverse outcomes in a variety of surgical specialties, including vascular surgery. Skeletal muscle index (SMI) and skeletal muscle density (SMD) have been shown to be a surrogate marker of sarcopenia and frailty. We explore the feasibility of using a fully automated deep-learning section-based muscle segmentation tool in calculating SMI and SMD, and predicting outcomes in patients undergoing abdominal aortic aneurysm (AAA) repair.
Methods
Pre-operative CT scans of patients undergoing AAA repair between 2016 and 2020 were run through and analysed by the automated sarcopenia segmentation tool, with 20% of scans undergoing manual segmentation to serve as ‘ground truth’. The primary outcome measure was the correlation between SMI at L3 (level of the 3rd lumbar vertebra) and 30-day composite outcome, which was a combination of death and defined post-operative complications.
Results
280 CT scans underwent successful automated segmentation, with excellent IRR noted between the manual and automated segmentations with an ICC 0.96 (95% CI 0.95 – 0.97, p < 0.05) for skeletal muscle area (SMA) and ICC 0.98 (95% CI 0.98 – 0.99, p < 0.05) for skeletal muscle density (SMD).
On regression modelling of SMI and 30-day composite outcome, good negative correlation was noted amongst women undergoing open AAA repair (r = -0.86 / R2 = 0.75, p<0.05).
Men undergoing open AAA repair in the lowest SMD quartile had poorer 30-day composite outcomes as compared to the highest quartile (t(34) = 2.741, p = 0.01).
Conclusion
This automated sarcopenia segmentation tool provides a rapid and accurate measure of SMI and SMD, and may help produce a sarcopenia score to guide clinicians on optimal treatment strategies and improve post-operative outcomes by identifying high-risk patients. Future research should look to validate the findings of this study using prospective, multicentre trials.
Sarcopenia is associated with adverse outcomes in a variety of surgical specialties, including vascular surgery. Skeletal muscle index (SMI) and skeletal muscle density (SMD) have been shown to be a surrogate marker of sarcopenia and frailty. We explore the feasibility of using a fully automated deep-learning section-based muscle segmentation tool in calculating SMI and SMD, and predicting outcomes in patients undergoing abdominal aortic aneurysm (AAA) repair.
Methods
Pre-operative CT scans of patients undergoing AAA repair between 2016 and 2020 were run through and analysed by the automated sarcopenia segmentation tool, with 20% of scans undergoing manual segmentation to serve as ‘ground truth’. The primary outcome measure was the correlation between SMI at L3 (level of the 3rd lumbar vertebra) and 30-day composite outcome, which was a combination of death and defined post-operative complications.
Results
280 CT scans underwent successful automated segmentation, with excellent IRR noted between the manual and automated segmentations with an ICC 0.96 (95% CI 0.95 – 0.97, p < 0.05) for skeletal muscle area (SMA) and ICC 0.98 (95% CI 0.98 – 0.99, p < 0.05) for skeletal muscle density (SMD).
On regression modelling of SMI and 30-day composite outcome, good negative correlation was noted amongst women undergoing open AAA repair (r = -0.86 / R2 = 0.75, p<0.05).
Men undergoing open AAA repair in the lowest SMD quartile had poorer 30-day composite outcomes as compared to the highest quartile (t(34) = 2.741, p = 0.01).
Conclusion
This automated sarcopenia segmentation tool provides a rapid and accurate measure of SMI and SMD, and may help produce a sarcopenia score to guide clinicians on optimal treatment strategies and improve post-operative outcomes by identifying high-risk patients. Future research should look to validate the findings of this study using prospective, multicentre trials.
Version
Open Access
Date Issued
2024-11-17
Date Awarded
2025-05-01
Copyright Statement
Attribution-NonCommercial 4.0 International Licence (CC BY-NC)
License URL
Advisor
Davies, Alun H
Rockall, Andrea
Publisher Department
Department of Surgery & Cancer
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Medicine (Research) MD (Res)