Optimization and Benchmarking of Image Segmentation for Improved Landmark Detection in Lower Limb X-Rays and Accurate Coronal Plane Alignment of the Knee Classification

Recent studies have explored image segmentation for landmark detection in computer vision and medical imaging of the lower limb, showing promising results. However, the proposed methodologies vary significantly, and a comparison with existing methods is lacking. In the present study, we investigated...

Full description

Saved in:
Bibliographic Details
Main Authors: Sebastian Amador Sanchez, Ashkan Zarghami, Philippe van Overschelde, Jef Vandemeulebroucke
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/11008572/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850230024246919168
author Sebastian Amador Sanchez
Ashkan Zarghami
Philippe van Overschelde
Jef Vandemeulebroucke
author_facet Sebastian Amador Sanchez
Ashkan Zarghami
Philippe van Overschelde
Jef Vandemeulebroucke
author_sort Sebastian Amador Sanchez
collection DOAJ
description Recent studies have explored image segmentation for landmark detection in computer vision and medical imaging of the lower limb, showing promising results. However, the proposed methodologies vary significantly, and a comparison with existing methods is lacking. In the present study, we investigated image segmentation for landmark detection on full lower-limb X-rays in detail and benchmark it against conventional landmark detection approaches. We detected eight landmarks in full lower limb X-rays and investigated methodological aspects to optimize image segmentation performance: network architecture (U-Net vs. Swin-UNETR), mask size centered at the landmark position to segment, and coordinate computation technique from the segmentation map. We contrasted image segmentation against optimized heatmap, coordinate, and segmentation-guided coordinate regression methods. The evaluation assessed the landmark detection error and phenotype classification accuracy based on lower limb alignment. The optimal segmentation approach employed a U-Net to segment circular masks (radius = 15 pixels), using probability thresholding before the centroid computation. Regarding landmark detection accuracy, image segmentation (median Euclidean distance (interquartile range) = 1.16 mm (1.50 mm)) was more accurate than heatmap (1.19 mm (1.61 mm)), coordinate (3.11 mm (2.87 mm)), and segmentation-guided coordinate regression (1.47 mm (1.67 mm)). Image segmentation outperformed heatmap, coordinate, and segmentation-guided coordinate regression in phenotype classification accuracy, achieving an average F1-score of 0.79, versus 0.72, 0.47, and 0.77, respectively. Our study led to an optimized approach for landmark detection using image segmentation, outperforming alternative detection approaches tuned and tested on the same data, highlighting image segmentation’s potential for broader medical imaging research applications.
format Article
id doaj-art-cbb66076973f477e9fe762c5d1bf0f8b
institution OA Journals
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-cbb66076973f477e9fe762c5d1bf0f8b2025-08-20T02:04:00ZengIEEEIEEE Access2169-35362025-01-0113923509236410.1109/ACCESS.2025.357234211008572Optimization and Benchmarking of Image Segmentation for Improved Landmark Detection in Lower Limb X-Rays and Accurate Coronal Plane Alignment of the Knee ClassificationSebastian Amador Sanchez0https://orcid.org/0000-0001-9831-7011Ashkan Zarghami1https://orcid.org/0009-0006-8892-3361Philippe van Overschelde2Jef Vandemeulebroucke3Department of Electronics and Informatics, Vrije Universiteit Brussel, Brussels, BelgiumDepartment of Electronics and Informatics, Vrije Universiteit Brussel, Brussels, BelgiummoveUP, Ghent, BelgiumDepartment of Electronics and Informatics, Vrije Universiteit Brussel, Brussels, BelgiumRecent studies have explored image segmentation for landmark detection in computer vision and medical imaging of the lower limb, showing promising results. However, the proposed methodologies vary significantly, and a comparison with existing methods is lacking. In the present study, we investigated image segmentation for landmark detection on full lower-limb X-rays in detail and benchmark it against conventional landmark detection approaches. We detected eight landmarks in full lower limb X-rays and investigated methodological aspects to optimize image segmentation performance: network architecture (U-Net vs. Swin-UNETR), mask size centered at the landmark position to segment, and coordinate computation technique from the segmentation map. We contrasted image segmentation against optimized heatmap, coordinate, and segmentation-guided coordinate regression methods. The evaluation assessed the landmark detection error and phenotype classification accuracy based on lower limb alignment. The optimal segmentation approach employed a U-Net to segment circular masks (radius = 15 pixels), using probability thresholding before the centroid computation. Regarding landmark detection accuracy, image segmentation (median Euclidean distance (interquartile range) = 1.16 mm (1.50 mm)) was more accurate than heatmap (1.19 mm (1.61 mm)), coordinate (3.11 mm (2.87 mm)), and segmentation-guided coordinate regression (1.47 mm (1.67 mm)). Image segmentation outperformed heatmap, coordinate, and segmentation-guided coordinate regression in phenotype classification accuracy, achieving an average F1-score of 0.79, versus 0.72, 0.47, and 0.77, respectively. Our study led to an optimized approach for landmark detection using image segmentation, outperforming alternative detection approaches tuned and tested on the same data, highlighting image segmentation’s potential for broader medical imaging research applications.https://ieeexplore.ieee.org/document/11008572/CPAK classificationdeep learningimage segmentationlandmark detectionlower limb X-rays
spellingShingle Sebastian Amador Sanchez
Ashkan Zarghami
Philippe van Overschelde
Jef Vandemeulebroucke
Optimization and Benchmarking of Image Segmentation for Improved Landmark Detection in Lower Limb X-Rays and Accurate Coronal Plane Alignment of the Knee Classification
IEEE Access
CPAK classification
deep learning
image segmentation
landmark detection
lower limb X-rays
title Optimization and Benchmarking of Image Segmentation for Improved Landmark Detection in Lower Limb X-Rays and Accurate Coronal Plane Alignment of the Knee Classification
title_full Optimization and Benchmarking of Image Segmentation for Improved Landmark Detection in Lower Limb X-Rays and Accurate Coronal Plane Alignment of the Knee Classification
title_fullStr Optimization and Benchmarking of Image Segmentation for Improved Landmark Detection in Lower Limb X-Rays and Accurate Coronal Plane Alignment of the Knee Classification
title_full_unstemmed Optimization and Benchmarking of Image Segmentation for Improved Landmark Detection in Lower Limb X-Rays and Accurate Coronal Plane Alignment of the Knee Classification
title_short Optimization and Benchmarking of Image Segmentation for Improved Landmark Detection in Lower Limb X-Rays and Accurate Coronal Plane Alignment of the Knee Classification
title_sort optimization and benchmarking of image segmentation for improved landmark detection in lower limb x rays and accurate coronal plane alignment of the knee classification
topic CPAK classification
deep learning
image segmentation
landmark detection
lower limb X-rays
url https://ieeexplore.ieee.org/document/11008572/
work_keys_str_mv AT sebastianamadorsanchez optimizationandbenchmarkingofimagesegmentationforimprovedlandmarkdetectioninlowerlimbxraysandaccuratecoronalplanealignmentofthekneeclassification
AT ashkanzarghami optimizationandbenchmarkingofimagesegmentationforimprovedlandmarkdetectioninlowerlimbxraysandaccuratecoronalplanealignmentofthekneeclassification
AT philippevanoverschelde optimizationandbenchmarkingofimagesegmentationforimprovedlandmarkdetectioninlowerlimbxraysandaccuratecoronalplanealignmentofthekneeclassification
AT jefvandemeulebroucke optimizationandbenchmarkingofimagesegmentationforimprovedlandmarkdetectioninlowerlimbxraysandaccuratecoronalplanealignmentofthekneeclassification