A novel medical image quality index

Area of Science:

• Medical Imaging
• Image Processing
• Signal Processing

Background:

• Medical image quality assessment is crucial for accurate diagnosis.
• Existing objective metrics may have limitations in evaluating compressed medical images.
• Grey relational analysis offers a novel approach for quality evaluation.

Purpose of the Study:

• To propose and validate a novel medical image quality index based on grey relational coefficients.
• To evaluate the performance of this index across different medical imaging modalities (DR, CT, MRI).
• To compare the proposed index with existing objective quality metrics.

Main Methods:

• Utilized 120 images from DR, CT, and MRI modalities.
• Applied JJ2000 compression algorithm at ten different ratios (10-100).
• Evaluated reconstructed image quality using four grey relational coefficient methods, selecting Nagai's for its speed and sensitivity.

Main Results:

• The grey relational coefficient index demonstrated high consistency with popular objective quality metrics.
• Results showed different slopes but high consistency across various image areas for tested methods.
• Nagai's grey relational coefficient proved less sensitive to window size variations compared to other metrics.

Conclusions:

• The proposed grey relational coefficient-based index is a reliable tool for assessing medical image quality.
• This novel index offers advantages in calculation speed, sensitivity, and robustness to window size changes.
• The method shows potential for improved objective evaluation of compressed medical images.