Bilateral Context Modeling for Residual Coding in Lossless 3D Medical Image Compression.

Area of Science:

• Medical Imaging
• Data Compression
• Computer Vision

Background:

• Residual coding is a prevalent technique in lossless compression, utilizing a lossy layer followed by lossless compression of residues.
• Context modeling and prior exploration are fundamental to residual coding principles.

Purpose of the Study:

• To propose a novel residual coding framework for efficient lossless compression of 3D medical images.
• To leverage an off-the-shelf video codec as the lossy layer and a Bilateral Context Modeling based Network (BCM-Net) as the residual layer.

Main Methods:

• The proposed framework employs a Bilateral Context Modeling based Network (BCM-Net) for efficient lossless compression of residues.
• BCM-Net explores intra-slice and inter-slice bilateral contexts using symmetry-based intra-slice context extraction (SICE) and bi-directional inter-slice context extraction (BICE) modules.
• SICE mines bilateral intra-slice correlations exploiting anatomical symmetry, while BICE explores bilateral inter-slice correlations from bi-directional references.

Main Results:

• The proposed method demonstrates superior performance compared to existing state-of-the-art techniques on popular 3D medical image datasets.
• Efficient redundancy reduction is achieved through the exploration of bilateral contexts.
• The framework enables high-efficiency lossless compression of 3D medical image residues.

Conclusions:

• The developed residual coding framework significantly improves lossless compression of 3D medical images.
• The BCM-Net, with its SICE and BICE modules, effectively captures essential bilateral contextual information for compression.
• The proposed method offers a promising solution for reducing storage and transmission burdens in medical imaging applications.