A predictive classified vector quantizer and its subjective quality evaluation for X-ray CT images

Area of Science:

• Medical Imaging
• Image Compression
• Signal Processing

Background:

• Medical image compression is crucial for efficient storage and transmission.
• Existing methods like DCT coding may not fully preserve diagnostic quality.
• Classified Vector Quantization (CVQ) offers a potential alternative for high-fidelity compression.

Purpose of the Study:

• To introduce and evaluate a novel Classified Vector Quantizer (CVQ) for X-ray CT image compression.
• To compare the performance of the proposed CVQ against Discrete Cosine Transform (DCT) coding.
• To assess the impact of human visual perception on image compression quality.

Main Methods:

• Development of a new CVQ algorithm utilizing decomposition and prediction, eliminating the need for side information.
• Incorporation of human visual perception characteristics into classification and bit allocation for enhanced quality.
• Subjective evaluation of compressed X-ray CT head images (n=9) from three patients at 10:1 and 15:1 compression ratios.
• Statistical analysis of radiologist evaluations (n=13) using ANOVA and Tukey's multiple comparison.

Main Results:

• The proposed CVQ algorithm demonstrated significantly better image quality than DCT coding at a 0.05 significance level.
• Interframe CVQ achieved original image quality at a 10:1 compression ratio, statistically indistinguishable.
• While CVQ reproduced high-quality compressed images, its effect on diagnostic accuracy requires further investigation.

Conclusions:

• The novel CVQ algorithm provides superior image compression quality for X-ray CT scans compared to DCT.
• CVQ effectively leverages human visual perception to improve compression efficiency and quality.
• Further research is needed to determine the impact of CVQ on the diagnostic accuracy of medical images.