A Block Adaptive Near-Lossless Compression Algorithm for Medical Image Sequences and Diagnostic Quality Assessment

Area of Science:

• Medical Imaging
• Image Compression
• Signal Processing

Background:

• Traditional lossless compression methods offer perfect reconstruction but limited compression ratios.
• Lossy compression methods achieve higher ratios but can degrade image quality, potentially losing diagnostic information.
• Near-lossless compression offers a balance, providing high compression with a controlled maximum error per pixel.

Purpose of the Study:

• To develop and evaluate a novel near-lossless compression algorithm for medical images.
• To achieve a high compression ratio while strictly maintaining diagnostic fidelity.
• To demonstrate the algorithm's effectiveness on various medical image types and bit depths.

Main Methods:

• The algorithm utilizes a resolution and modality independent threshold-based predictor.
• It incorporates a resolution independent gradient edge detector (RIGED) for inter-pixel redundancy removal.
• Block adaptive arithmetic encoding (BAAE) and optimal quantization (q) levels are employed for coding redundancy removal and efficiency.

Main Results:

• The proposed method achieved a BPP of 1.37 and a PSNR of 51.35 dB for 8-bit images, outperforming other near-lossless techniques.
• For 16-bit standard and real-time medical datasets, average BPP values of 3.411 and 2.609 were obtained, respectively.
• High compression efficiency was maintained without compromising the diagnostic quality of the recovered images.

Conclusions:

• The developed near-lossless predictive coding technique effectively enhances compression ratios for medical images.
• The algorithm successfully preserves crucial diagnostic information, making it suitable for clinical applications.
• This approach offers a significant improvement over existing methods for efficient and reliable medical image compression.