Magnetic resonance image segmentation for knee osteoarthritis using active shape models
Knee osteoarthritis is a chronic joint inflammation disease that affects the aged population nowadays. The disease leads to gradual degradation of cartilage and thus deteriorates the function of the knee joint. Magnetic Resonance Imaging (MRI) provides promising results for the early detection of kn...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Conference or Workshop Item |
| Published: |
2015
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/59361/ http://dx.doi.org/10.1109/BMEiCON.2014.7017365 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Universiti Teknologi Malaysia |
| Summary: | Knee osteoarthritis is a chronic joint inflammation disease that affects the aged population nowadays. The disease leads to gradual degradation of cartilage and thus deteriorates the function of the knee joint. Magnetic Resonance Imaging (MRI) provides promising results for the early detection of knee osteoarthritis. Conventionally, the MR image segmentation for knee osteoarthritis is manually done by clinicians. Limitations of this process include being laborious, time-consuming and prone to subjective diagnosis error. Therefore, the development of an automated cartilage segmentation method is crucial to assist the medical research in knee osteoarthritis. This project applied the Active Shape Models (ASM) approach to create semi-automated cartilage segmentation software. A shape model was constructed from a training set consisting of 10 knee MR images which includes major variations of the knee cartilage shape. Principle component analysis (PCA) was utilized to identify the main axes of variations used to build the shape model. This shape model was finally used to segment the knee articular cartilage. Outcomes of the ASM segmentation were compared with the outcome of manual segmentation. Experimental results showed that the sensitivity of developed ASM approach increased averagely from 73.78% to 80.75%, proportional to the increasing of the number of iteration in the segmentation as well as landmark of the shape model. This technique is reliable to contribute to medical research in knee osteoarthritis by providing an efficient and high accuracy segmentation method for knee articular cartilage, to further assist in the detection of knee osteoarthritis via MRI technique. |
|---|