Segmentation and Classification of Heart Angiographic Images Using Machine Learning Techniques.

Area of Science:

• Medical Imaging
• Cardiology
• Computer-Assisted Diagnosis

Background:

• Heart angiography is crucial for diagnosing heart vessel abnormalities but is time-consuming and prone to errors.
• Accurate segmentation and classification of heart blood vessels are vital for timely disease detection.

Purpose of the Study:

• To develop a computer-assisted system for localizing human heart blood vessels in angiographic images.
• To enhance the accuracy and efficiency of diagnosing heart-related diseases through automated analysis.

Main Methods:

• Proposed a multiclass ensemble classification mechanism for heart blood vessel analysis.
• Implemented automated segmentation of heart blood vessels followed by feature extraction (texture, statistical, geometrical).
• Classified vessels into four categories: normal, block, narrow, and blood flow-reduced.

Main Results:

• The proposed computer-assisted approach successfully segmented and classified heart blood vessels.
• Extracted low-level features like texture, statistical, and geometrical properties for accurate analysis.
• Achieved high accuracy in categorizing vessels into normal, block, narrow, and blood flow-reduced states.

Conclusions:

• The developed system offers a useful, accurate, and time-saving environment for cardiologists.
• Automated segmentation and classification significantly improve the truthfulness and speed of diagnosing heart illnesses.
• This computer-assisted diagnosis system aids in the early detection and management of cardiovascular diseases.