Octave CNN Feature Maps for Faster H&E Tissue Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods for Hematoxylin and Eosin (H&E) stained tissue samples are inefficient and time-consuming, particularly in determining feature maps using convolutional neural networks.
Innovation Solution
Implementing octave convolution in a convolutional neural network (CNN) to generate high-frequency and low-frequency feature maps, with the latter having half the spatial resolution of the former, allowing for reduced processing time and improved efficiency by facilitating easier pooling and upsampling operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional convolutional neural networks are used to generate feature maps from H&E stained tissue samples, then feature extraction can be performed, but the processing time is excessive and computational efficiency is low
Solution Approach 1:
The patent applies octave convolution to segment the feature extraction process into distinct high-frequency and low-frequency components. By separating the processing into these frequency bands, the system can handle different spatial resolutions simultaneously, improving overall processing efficiency and reducing total computation time while maintaining comprehensive feature extraction capability.
Solution Approach 2:
The patent implements different processing qualities for different spatial regions through octave convolution. High-frequency features are processed at full resolution where detailed information is critical, while low-frequency features are processed at reduced resolution where computational resources can be optimized. This local quality differentiation significantly reduces overall processing time without sacrificing diagnostic accuracy.
2Measurement precision
If high spatial resolution is maintained throughout the entire feature extraction process, then feature detail is preserved, but computational resources and processing time increase significantly
Solution Approach 1:
The patent applies different spatial resolution requirements to different frequency components through octave convolution. High-frequency feature maps maintain full spatial resolution to preserve fine cellular and subcellular details critical for diagnosis. Low-frequency feature maps use reduced spatial resolution since they capture broader structural information. This local quality differentiation maintains measurement precision where needed while significantly improving computational efficiency overall.
Data Source
AI summary
An apparatus and methods for determining features in an image of a Hematoxylin and Eosin (H&E) stained tissue sample. An apparatus can be configured to apply a machine learning model to the H&E stained tissue sample image to determine features in the image. Applying the machine learning model can include performing convolution operations on the H&E stained tissue sample image to generate a plurality of initial feature maps, applying octave-convolution-first-layer operations on the initial feature maps to generate initial high-frequency feature maps and low-frequency feature maps, applying octave-convolution operations on the high-frequency initial feature maps to generate refined high-frequency feature maps, applying octave-convolution operations on the low-frequency initial feature maps to generate refined low-frequency feature maps, and applying octave-convolution last layer operations on the refined high-frequency feature maps and the refined low-frequency feature maps to produce combined feature maps.