Resampling Medical Images for Deep Learning DFOV Robustness
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Deep learning neural networks trained on specific display field of view (DFOV) or spatial resolution struggle with accuracy when applied to medical images with different DFOV or spatial resolution, leading to decreased inferencing performance in tasks like image quality enhancement, denoising, and image kernel transformation.
Innovation Solution
A system that resamples medical images to match the trained DFOV or spatial resolution of the neural network, executes the network on the resampled image, and then resamples the output back to the original image's DFOV or spatial resolution, using techniques like up-sampling and down-sampling to maintain accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a deep learning neural network is trained on a specific display field of view (DFOV) or spatial resolution, then the network achieves high accuracy for that specific resolution, but the accuracy decreases when applied to medical images with different DFOV or spatial resolution
Solution Approach 1:
The system performs preliminary resampling of the input medical image to transform it into the target DFOV before executing the neural network. This preliminary action ensures the image is in the correct format for the trained network, preventing DFOV mismatch errors and maintaining high inferencing accuracy across different input resolutions.
Solution Approach 2:
The resampling operation acts as an intermediary between the input image and the neural network. It transforms the input image from its original DFOV to the target DFOV expected by the network, serving as a bridge that reconciles the mismatch between input and network expectations without requiring the network itself to be modified.
2Productivity
If the neural network is executed directly on images with mismatched DFOV, then the processing is faster and simpler, but inaccuracies and artefacts are introduced in the output
Solution Approach 1:
The system performs preliminary resampling of the input medical image to transform it into the target DFOV before executing the neural network. This preliminary action ensures the image is in the correct format for the trained network, preventing DFOV mismatch errors and maintaining high inferencing accuracy across different input resolutions.
Solution Approach 2:
The resampling operation acts as an intermediary between the input image and the neural network. It transforms the input image from its original DFOV to the target DFOV expected by the network, serving as a bridge that reconciles the mismatch between input and network expectations without requiring the network itself to be modified.
Data Source
AI summary
Systems/techniques that facilitate deep learning robustness against display field of view (DFOV) variations are provided. In various embodiments, a system can access a deep learning neural network and a medical image. In various aspects, a first DFOV, and thus a first spatial resolution, on which the deep learning neural network is trained can fail to match a second DFOV, and thus a second spatial resolution, exhibited by the medical image. In various instances, the system can execute the deep learning neural network on a resampled version of the medical image, where the resampled version of the medical image can exhibit the first DFOV and thus the first spatial resolution. In various cases, the system can generate the resampled version of the medical image by up-sampling or down-sampling the medical image until it exhibits the first DFOV and thus the first spatial resolution.