Related Experiment Video
Updated: Feb 25, 2026
11:27
A Modified Sonographic Algorithm for Image Acquisition in Life-Threatening Emergencies in the Critically Ill Newborn
Published on: April 7, 2023
7.6K
Automatic Torso Detection in Images of Preterm Infants
Meharmeet Kaur1, Andrew P Marshall1, Caillin Eastwood-Sutherland1
1School of Engineering and ICT, University of Tasmania, TAS, Hobart, 7001, Australia.
Journal of Medical Systems
|July 30, 2017
Summary
This study presents an automated algorithm for detecting preterm infant torsos in neonatal intensive care unit (NICU) images, improving non-invasive respiratory monitoring accuracy.
Area of Science:
- Medical Imaging
- Neonatal Care
- Computer Vision
Background:
- Patient respiratory monitoring systems are crucial but underutilized in neonatal intensive care units (NICUs).
- Accurate infant torso detection is essential for effective non-invasive respiratory monitoring.
Purpose of the Study:
- To develop and evaluate an automated algorithm for detecting the torso region in preterm infants.
- To enhance the application of imaging systems for respiratory monitoring in NICU settings.
Main Methods:
- Utilized normalized cut for image segmentation into clusters.
- Employed two fuzzy inference systems for detecting the nappy and torso.
- Tested the algorithm on overhead images of 16 preterm infants under varying NICU conditions.
Main Results:
- The algorithm successfully identified the torso in 15 out of 16 images.
- Demonstrated high agreement between algorithm-detected torsos and expert-identified regions.
- Algorithm robust to uncontrolled illumination, varied poses, and background clutter.
Conclusions:
- The proposed algorithm shows significant promise for automated torso detection in preterm infants.
- This technology can improve the accuracy and applicability of non-invasive respiratory monitoring in NICUs.
- Further development could enhance real-time monitoring capabilities for vulnerable neonates.
