Related Experiment Video
Updated: Jan 7, 2026
Author Spotlight: Unraveling the Impact of Mechanical Ventilation on Diaphragm Function and Patient Outcomes
Published on: November 3, 2023
Development of an automated assessment of ultrasound measured diaphragm thickness
Alette A Koopman1, Jitka Burger1, Kay Laarman1
1Department of Paediatrics, Division of Paediatric Critical Care Medicine, Beatrix Children's Hospital, University Medical Center Groningen, University of Groningen, Huispostcode CA62 P.O. Box 30.001 9700 RB Groningen, the Netherlands.
Background And Objectives:
Inadequate titration of support during mechanical ventilation can cause ventilator-induced diaphragmatic dysfunction through changes in diaphragm structure, leading to prolonged weaning. Bedside diaphragm ultrasound (DUS) enables real-time, noninvasive assessment of diaphragm structure, but its interpretation is often hindered by interobserver variability. To improve the accuracy of DUS measurements, we developed an automated analysis framework and evaluated its inter-method agreement.
Methods:
The automated DUS analysis framework generates diaphragm thickness (Tdi)-time plots. We evaluated its performance using DUS clips from mechanically ventilated children < 5 years of age who could maintain spontaneous breathing. Tdi was measured at end-expiration (Tdi,ee) and peak-inspiration (Tdi,pi), manually by using a DICOM viewer and at the bedside, as well as automated by the software tool. Inter-method agreement was assessed by intra-class correlation coefficients (ICC).
Results:
In total 26 DUS clips of 10 patients were analyzed. Overall, median Tdi,pi was 1.58 mm [IQR 1.24-1.85] and Tdi,ee was 1.32 mm [IQR 0.99-1.46], resulting in a thickening fraction (TFdi) of 26.9 % [IQR 22.8-31.6]. Inter-method agreement between the software tool and bedside measurements was reflected by an ICC of 0.918 (95 % CI: 0.83-0.96). The ICC between software tool and the DICOM viewer was 0.991 95 % CI [0.98-0.99]. Bland-Altman analysis showed a mean difference of -0.0076 mm [95 % LOA -0.18 - 0.17] between DICOM- and tool-based Tdi(ee&pi) measurements.
Conclusions:
This proof-of-concept study demonstrates that automated analysis of diaphragm thickness and thickening fraction in mechanically ventilated children is feasible and provides reproducible quantification of Tdi, reflected by high intra-class correlation coefficients.
