Multi-Plane Image Compression in Digital Pathology
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Digital pathology systems face significant challenges in storing and transmitting large multi-plane images due to excessive data size, which is several times greater than single-plane images, resulting in high time and cost burdens.
Innovation Solution
An image compression method that selects optimal blocks with focal points across multiple planes, forms a virtual optimal plane image, generates predictive and differential plane images, and compresses these images to reduce redundancy and data size effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multi-plane images are used to capture samples with non-uniform thickness, then image quality and diagnostic accuracy are improved, but data size increases several times compared to single-plane images
Solution Approach 1:
The patent segments the multi-plane image data into multiple individual plane images, then applies independent compression algorithms to each plane. This segmentation allows for more efficient compression by treating each plane as a separate entity while preserving the diagnostic information across all planes, thereby reducing the overall data size while maintaining image quality.
Solution Approach 2:
The patent changes the compression parameters and algorithms applied to different planes based on their specific characteristics. By adapting compression settings for each plane individually, the system achieves optimal compression ratios while preserving the essential diagnostic features in each plane, thus reducing total data size without sacrificing image quality.
2Loss of information
If multi-plane images are stored and transmitted, then complete diagnostic information is preserved, but time and cost for storage and transmission increase significantly
Solution Approach 1:
The patent divides the multi-plane image set into separate plane images that can be independently compressed, stored, and transmitted. This segmentation enables parallel processing and optimized transmission strategies for each plane, reducing the total time and computational resources required while ensuring all diagnostic information is preserved across the planes.
Solution Approach 2:
The patent applies different compression and transmission parameters to different planes based on their diagnostic importance and data characteristics. This allows critical diagnostic planes to be transmitted with higher fidelity while less critical planes use more aggressive compression, optimizing the balance between information preservation and transmission efficiency.
3Quantity of substance
If conventional compression algorithms are applied to multi-plane images, then some data reduction is achieved, but redundancy removal is insufficient and data size remains excessively large
Solution Approach 1:
The patent segments the compression process to operate independently on each plane image, allowing for more thorough redundancy removal within each plane. This per-plane compression approach identifies and eliminates redundancies that would be missed in a holistic multi-plane compression approach, achieving better overall compression ratios while preserving diagnostic information.
Data Source
AI summary
Disclosed is an image compression method of a digital pathology system. The image compression method is a method of compressing digital slide images having first to nth original plane images (n is a natural number greater than or equal to 2). The image compression method includes selecting a block having an optimal focal point as an optimal block from each set of blocks positioned at identical positions of the first to nth original plane images; forming one plane image as a virtual optimal plane image by combining only the optimal blocks; generating block descriptors for forming the first to nth original plane images based on the virtual optimal plane image; generating first to nth predictive plane images from the virtual optimal plane image such that the first to nth predictive plane images are the least out of focus by using location information for the blocks and the block descriptors; generating first to nth differential plane images, the first differential plane image corresponding to a difference between the first original plane image and the first predictive plane image and the nth differential plane image corresponding to a difference between the nth original plane image and the nth predictive plane image; and compressing the first to nth differential plane images.