Programmed for Automatic Bone Disorder Clustering Based on Cumulative Calcium Prediction for Feature Extraction.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Radiology

Background:

• Bone disorder prediction accuracy varies among physicians.
• A new method for estimating bone calcium levels from radiological images is proposed.
• This aims to create a precise bone disorder cataloging system.

Purpose of the Study:

• To develop and validate a precise bone disorder cataloging system.
• To improve the accuracy of calcium value estimation from bone radiological images.
• To enhance the classification of bone disorders such as osteoporosis and osteopenia.

Main Methods:

• A binning technique divides images into blocks for calcium volume estimation.
• Image preprocessing includes median filtering and Otsu-thresholding segmentation.
• Local Binary Pattern (LBP) and Multi-class Support Vector Machine (MSVM) are used for feature extraction and classification.

Main Results:

• The proposed binning technique achieved 97.4% accuracy and 98.3% sensitivity in bone disorder classification.
• Calcium values from two processing stages were summed for precise estimation.
• The method demonstrated superior performance compared to approaches without the binning technique.

Conclusions:

• The binning technique proves effective for bone disorder classification systems.
• Validation with physician-confirmed bone images showed an 80% success rate.
• The study highlights the potential for improved diagnostic accuracy in bone health.