Biological Image Tile Processing for Neural Network Prediction Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional patient response prediction techniques in clinical trials are inefficient and inaccurate due to the large size of biological images, requiring billions of parameters, and fail to extract sufficient features from medical images, resulting in low accuracy ranging from 20%-45%.
Innovation Solution
Implementing image processing and machine learning techniques to process biological images by generating image tiles, preprocessing them to identify and color normalize tissue locations, and using artificial neural networks (ANNs) to predict patient responses by identifying discrete tissue components, thereby increasing computational efficiency and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional patient response prediction techniques are used on large biological images, then comprehensive feature extraction is attempted, but computational requirements become excessively high (billions of parameters) and accuracy remains low (20%-45%)
Solution Approach 1:
The patent divides large biological images into smaller image tiles, allowing the neural network to process manageable portions independently. This segmentation reduces the computational burden from billions of parameters to a fraction, while maintaining comprehensive feature extraction across the entire image through aggregation of tile-level predictions
Solution Approach 2:
The patent introduces spatial arrangement information as an additional dimension by organizing image tiles in their original spatial context. This allows the system to capture both local features within tiles and global spatial relationships, improving prediction accuracy without proportionally increasing computational complexity
2Productivity
If traditional techniques process entire biological images, then all regions are analyzed, but computational efficiency is low and processing time is excessive
Solution Approach 1:
By dividing the image into tiles that can be processed in parallel, the system achieves significant speedup. Each tile is independently processed by the neural network, allowing for parallel computation that dramatically reduces total processing time compared to sequential analysis of the entire image
Solution Approach 2:
The patent processes only the most relevant regions by identifying discrete biological tissue components within tiles rather than analyzing every pixel uniformly. This selective processing focuses computational resources on diagnostically important areas, improving efficiency without sacrificing diagnostic comprehensiveness
3Measurement precision
If traditional methods use simple feature extraction, then processing is faster, but prediction accuracy remains insufficient (20%-45%)
Solution Approach 1:
The patent performs preliminary processing by dividing images into tiles and pre-identifying potential tissue components before full neural network analysis. This preparation step organizes data in a way that enables the network to focus on discriminative features, achieving high accuracy without excessive processing time during the actual prediction phase
Solution Approach 2:
The patent replaces traditional mechanical feature extraction methods with an artificial neural network that automatically learns optimal features from image tiles. This substitution enables sophisticated pattern recognition and feature extraction that achieves 80%+ accuracy while maintaining computational efficiency through the tile-based approach
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In an aspect, a data processing system includes a computer-readable memory comprising computer-executable instructions, and at least one processor configured to execute executable logic including at least one artificial neural network trained to predict one or more responses to a chemical substance by identifying one or more discrete biological tissue components in a biological image. When the at least one processor is executing the computer-executable instructions, the at least one processor is configured to carry out operations including: receiving spatially arranged image data representing a biological image of a patient; generating spatially arranged image tile data representing a plurality of image tiles; processing the spatially arranged image tile data through one or more data structures storing one or more portions of executable logic included in the artificial neural network to predict one or more responses of a patient.