MeQryEP: A Texture Based Descriptor for Biomedical Image Retrieval

Area of Science:

• Computer Vision
• Image Processing
• Medical Image Analysis

Background:

• Image texture analysis is crucial for various applications, including medical imaging and content-based retrieval.
• Existing methods like Local Binary Patterns (LBP) encode grayscale relationships but can be enhanced.
• Local Quinary Patterns (LQP) offer a non-binary approach to texture feature extraction.

Purpose of the Study:

• To introduce and evaluate Quinary Encoding on Mesh Patterns (MeQryEP) for biomedical image indexing and retrieval.
• To investigate the efficacy of Local Quinary Patterns (LQP) on mesh patterns in three orientations.
• To improve spatial structure information encoding for enhanced retrieval performance.

Main Methods:

• Developed MeQryEP utilizing local quinary patterns on mesh patterns in three orientations.
• Encoded grayscale relationships between pixels using mesh pattern directions.
• Applied the method to benchmark datasets: LIDC-IDRI-CT, VIA/I-ELCAP-CT (CT lung images), and OASIS-MRI (MRI brain images).

Main Results:

• MeQryEP demonstrated superior performance over state-of-the-art texture extraction methods.
• The method achieved higher average retrieval precision (ARP) and average retrieval rate (ARR).
• Analysis confirmed the viability and effectiveness of MeQryEP on diverse biomedical datasets.

Conclusions:

• MeQryEP offers an innovative approach to texture feature extraction for biomedical image retrieval.
• The method's use of mesh image structure enhances spatial information encoding, leading to improved retrieval results.
• MeQryEP represents a significant advancement in the field of medical image analysis and retrieval.