Learning Data Generation Support Apparatus for Medical Image Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In the medical field, acquiring high-quality data necessary for deep learning is inefficient due to the large amount of data stored in medical image management systems, requiring manual discrimination of correct answer data, which is time-consuming and inaccurate.
Innovation Solution
A learning data generation support apparatus that automatically acquires and processes images from different modalities to extract anatomic regions, determining and registering comparison images as correct answer data by setting a reference image and comparison images based on the portion or disease, facilitating the generation of large amounts of necessary image data for deep learning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual discrimination of correct answer data is performed from large amount of medical image data, then data quality can be controlled, but time consumption increases and efficiency decreases
Solution Approach 1:
The system enables automatic self-service by having the image processing apparatus itself perform the discrimination of correct answer data without requiring manual intervention. The apparatus automatically compares images from different modalities, extracts anatomic regions, determines correctness based on pre-stored criteria, and registers suitable images as correct answer data, thereby eliminating the time-consuming manual discrimination process while maintaining data quality.
Solution Approach 2:
The patent replaces the mechanical manual discrimination process with an automated image processing system that uses computational algorithms to compare images, extract features, and determine correctness. This substitution of mechanical human judgment with automated computational mechanisms significantly increases productivity while maintaining measurement precision for data quality control.
2Measurement precision
If deep learning is performed with large amount of data, then recognition accuracy improves, but the complexity of data acquisition and processing increases
Solution Approach 1:
The system extracts only the necessary correct answer data from the large volume of stored medical images by automatically comparing images from different modalities and selecting those that meet predetermined criteria. This extraction process simplifies the data processing complexity by filtering out irrelevant images and retaining only high-quality data suitable for deep learning, thereby reducing the overall complexity while maintaining the ability to achieve high recognition accuracy.
Solution Approach 2:
The patent performs preliminary actions by pre-storing criteria for determining correct answer data and pre-processing images to extract anatomic regions before the actual discrimination process. This preliminary preparation simplifies the subsequent data acquisition and processing complexity by having the system ready with pre-established rules and pre-extracted features, enabling efficient automatic discrimination without requiring complex real-time processing.
Data Source
AI summary
Any one of acquired images is set as a reference image, and an image other than the reference image is set as a comparison image. According to a portion or a disease, a first image processing of extracting an anatomic region is executed with respect to the reference image, and a second image processing of extracting an anatomic region is executed with respect to the comparison image. Whether the comparison image is available as correct answer data is determined using the anatomic region of the reference image and the anatomic region of the comparison image. The comparison image determined to be available as the correct answer data is registered as the correct answer data.