Cell Image Labeling via Selective Neural Network Tile Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for identifying double positive cells in immunocytochemistry images are inefficient and difficult for pathologists to interpret, especially when dealing with invasive biopsy samples and computationally intensive neural networks.
Innovation Solution
A method involving the reception of digital images of stained immunocytochemistry samples, followed by computerized classification of cells based on color, shape, and texture, and subsequent labeling using a trained neural network applied only to candidate cell portions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a neural network is applied to the entire digital image to identify double positive cells, then the identification accuracy is improved, but the computational intensity increases significantly
Solution Approach 1:
The patent divides the digital image into multiple tiles or regions, and applies the neural network only to selected candidate tiles rather than the entire image. This segmentation approach reduces the computational load while maintaining identification accuracy for double positive cells in the selected regions.
Solution Approach 2:
The patent applies different processing strategies to different regions of the image. High-resolution neural network analysis is applied only to candidate tiles that are likely to contain double positive cells, while other regions receive less intensive processing, optimizing the balance between accuracy and computational resources.
2Extent of automation
If a neural network is used for cell classification, then the automation level is improved, but the interpretability for pathologists deteriorates due to the black box nature
Solution Approach 1:
The patent introduces an intermediary layer between the neural network and the pathologist by providing visual overlays that highlight predicted double positive cells directly on the digital image. This intermediary visualization maintains automation while restoring interpretability, allowing pathologists to verify and understand the AI's predictions.
Solution Approach 2:
The patent uses color-coded visual indicators to represent different cell classifications and prediction confidence levels. This visual encoding provides intuitive information to pathologists about the neural network's findings, improving interpretability without reducing automation.
3Reliability
If traditional biopsy methods are used to obtain tissue samples, then the diagnostic reliability is improved, but the invasiveness to the patient increases
Solution Approach 1:
The patent uses digital copies (images) of tissue samples obtained through less invasive methods such as liquid-based cytology or fine needle aspiration, replacing the need for traditional invasive biopsies. The digital image analysis maintains diagnostic reliability while significantly reducing patient invasiveness and discomfort.
Data Source
AI summary
The method of labeling an image of cells of a patient, in particular an immunocytochemistry image comprises the following steps. First, a digital image of a stained immunocytochemistry biological sample of the patient is received. Following by the step that a computerized classification of cells in the digital image based on color, shape or texture in the digital image, the digital image is labeled by application of a trained neural network on at least one portion of the digital image which comprises a digital image of one cell classified under a first category during the computerized classification.