Whole Slide Tissue Segmentation for Rejoining Broken Specimen Pieces
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing computational pathology methods fail to adequately address the need for identifying and grouping tissue pieces on a slide, which are often broken or separated, and require manual annotation, limiting automation of tasks like core counting and anomaly detection.
Innovation Solution
A system utilizing artificial intelligence and synthetic data generation to train a panoptic segmentation model, enabling the identification of tissue fragments and their grouping, even when broken or separated, by introducing simulated tears and folds in tissue images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual annotation is used to identify and group tissue pieces, then accuracy of tissue identification can be maintained, but productivity and automation level are reduced
Solution Approach 1:
The system enables automated self-identification of tissue pieces through panoptic segmentation models that automatically detect, classify, and group tissue fragments without requiring manual annotation, thereby maintaining accuracy while significantly improving productivity
Solution Approach 2:
The patent replaces the mechanical manual annotation process with an AI-based panoptic segmentation system that uses deep learning models to automatically identify and group tissue pieces, eliminating the need for manual intervention while preserving identification accuracy
2Measurement precision
If manual annotation is used to identify and group tissue pieces, then accuracy of tissue identification can be maintained, but extent of automation is reduced
Solution Approach 1:
The system enables automated self-identification of tissue pieces through panoptic segmentation models that automatically detect, classify, and group tissue fragments without requiring manual annotation, thereby maintaining accuracy while significantly improving productivity
Solution Approach 2:
The patent replaces the mechanical manual annotation process with an AI-based panoptic segmentation system that uses deep learning models to automatically identify and group tissue pieces, eliminating the need for manual intervention while preserving identification accuracy
3Device complexity
If traditional image processing methods are used, then device complexity can be kept low, but ability to handle broken or separated tissue pieces is insufficient
Solution Approach 1:
The system employs a composite approach combining multiple deep learning models (panoptic segmentation, instance segmentation, and classification models) to handle the complexity of identifying and grouping broken tissue pieces, achieving high reliability through model ensembles rather than simple algorithms
Solution Approach 2:
The patent applies segmentation techniques at multiple levels: panoptic segmentation to identify different tissue types, instance segmentation to separate individual tissue pieces, and grouping algorithms to reconnect broken fragments, systematically breaking down the complex task of handling separated tissue pieces
Data Source
Figure 1A~1D
Figure 2A
Figure 2B
AI summary
Systems and methods are disclosed for identifying formerly conjoined pieces of tissue in a specimen, comprising receiving one or more digital images associated with a pathology specimen, identifying a plurality of pieces of tissue by applying an instance segmentation system to the one or more digital images, the instance segmentation system having been generated by processing a plurality of training images, determining, using the instance segmentation system, a prediction of whether any of the plurality of pieces of tissue were formerly conjoined, and outputting at least one instance segmentation to a digital storage device and/or display, the instance segmentation comprising an indication of whether any of the plurality of pieces of tissue were formerly conjoined.