Related Experiment Video
Updated: Aug 26, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.9K
Multi-structure bone segmentation in pediatric MR images with combined regularization from shape priors and
Arnaud Boutillon1, Bhushan Borotikar2, Valérie Burdin1
1IMT Atlantique, Brest, France; LaTIM UMR 1101, Inserm, Brest, France.
Artificial Intelligence in Medicine
|October 7, 2022
Summary
This study introduces a novel deep learning network for segmenting pediatric musculoskeletal magnetic resonance (MR) images, improving accuracy on limited data. The method enhances segmentation performance for children's bone imaging, aiding in disorder diagnosis.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Medicine
- Pediatric Orthopedics
Background:
- Accurate morphological and diagnostic evaluation of the pediatric musculoskeletal system is vital in clinical settings.
- Existing segmentation models struggle with the scarcity and heterogeneity of pediatric imaging data.
- Challenges in segmenting pediatric bone structures impact diagnosis and treatment planning.
Purpose of the Study:
- To develop a novel pre-trained regularized convolutional encoder-decoder network for segmenting heterogeneous pediatric magnetic resonance (MR) images.
- To address the limitations of current models in handling scarce pediatric imaging datasets.
- To improve the accuracy and consistency of multi-bone segmentation in pediatric musculoskeletal imaging.
Main Methods:
- Proposed a novel pre-trained regularized convolutional encoder-decoder network.
- Implemented a new optimization scheme with additional regularization terms in the loss function.
- Incorporated shape priors regularization using an auto-encoder and adversarial regularization via a discriminator for globally consistent and precise delineations.
Main Results:
- The proposed method achieved competitive or superior performance compared to existing approaches on metrics including Dice, sensitivity, specificity, and surface distance measures.
- Evaluated on scarce pediatric ankle and shoulder joint imaging datasets, including both healthy and pathological cases.
- Demonstrated improved prediction accuracy for models pre-trained on large non-medical image databases when integrated with the proposed approach.
Conclusions:
- The developed network offers a robust solution for segmenting challenging pediatric musculoskeletal MR images, even with limited data.
- The novel regularization techniques enhance segmentation accuracy and global consistency.
- This approach holds promise for advancing the management of pediatric musculoskeletal disorders through improved imaging analysis.
More Related Videos
Related Concept Videos
01:18Classification of Bones
6.2K
The bones of the human skeletal system are of varied shapes, sizes, and functions. They can be classified based on their shape and function into four major classes: long bones, short bones, flat bones, and irregular bones. Some classifications include a fifth type, the sesamoid bones, as a separate class, whereas others categorize them under short bones.
Long and Short Bones
The appendicular skeleton, particularly the upper and lower limbs, is primarily made of long and short bones. The...
Long and Short Bones
The appendicular skeleton, particularly the upper and lower limbs, is primarily made of long and short bones. The...
6.2K
01:55Bone Structure
49.0K
Within the skeletal system, the structure of a bone, or osseous tissue, can be exemplified in a long bone, like the femur, where there are two types of osseous tissue: cortical and cancellous.
49.0K
01:40Bone Remodeling
38.5K
Bone remodeling is a continuous and balanced process of bone resorption by osteoclasts and bone formation by osteoblasts. In adults, it helps maintain bone mass and calcium homeostasis. While mechanical stress can stimulate turnover as part of the normal maintenance and reparative process, several hormones also regulate bone remodeling.
38.5K