Detection of turner syndrome using hand X-ray using anchor based links segmentation method.
Ramachandran R1, N Gobalakrishnan2, Arun Chokkalingam3
1Research Scholar, Anna University, Chennai, Tamilnadu, India.
Summary
This study introduces an Anchor Based Link (ABL) algorithm for segmenting hand X-rays to detect Turner Syndrome (TS) in children. The new method improves detection accuracy by analyzing bone dimensions from segmented metacarpal bones.
Area of Science:
- Medical Imaging
- Computer Vision
- Genetics
Background:
- Turner Syndrome (TS) is a chromosomal disorder affecting female growth, leading to issues like immature ovaries, short stature, and heart abnormalities.
- Early diagnosis of TS is crucial but can be challenging, especially in cases with subtle symptoms, often delaying identification until adolescence or adulthood.
- Accurate medical image segmentation is vital for diagnosing conditions like TS, yet existing algorithms struggle with common segmentation errors.
Purpose of the Study:
- To develop and present an improved algorithm for segmenting hand digital X-ray images specifically for detecting Turner Syndrome in children.
- To address limitations of current segmentation techniques, such as under-segmentation, over-segmentation, and edge inaccuracies.
- To validate a novel Anchor Based Link (ABL) segmentation approach for TS detection using the fourth metacarpal bone in hand X-rays.
Main Methods:
- An Anchor Based Link (ABL) segmentation algorithm was developed to analyze the fourth metacarpal bone in left-hand X-ray images.
- The proposed ABL method was compared against established watershed segmentation and Gaussian-Mixture-Model-based Hidden-Markov-Random-Field (GMM-HMRF) techniques.
- Segmentation accuracy was evaluated by analyzing the ratio of height to width of the left fourth finger, focusing on edge pixels of the segmented metacarpal bone.
Main Results:
- The ABL segmentation approach demonstrated superior performance compared to existing methods in segmenting the fourth metacarpal bone.
- Analysis of the segmented bone's dimensions provided a basis for differentiating between normal children and those with Turner Syndrome.
- The method achieved an average Dice coefficient of 0.60 ± 0.02 on a dataset of 50 sample X-ray hand images.
Conclusions:
- The proposed Anchor Based Link (ABL) segmentation method offers a promising tool for the early and accurate detection of Turner Syndrome from hand X-ray images.
- The algorithm's ability to accurately segment bone structures and analyze dimensional ratios contributes to improved diagnostic capabilities for TS.
- Further validation and application of this technique could enhance the clinical management of Turner Syndrome.
