Medical Image Processing Device for Contrast Phase Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing medical image processing technologies face challenges in accurately estimating temporal information, particularly when imaging time point data is missing or incorrect in DICOM tags, and require manual data preparation for learning-based algorithms, limiting their applicability across different examination protocols and organs.
Innovation Solution
A medical image processing device and method that estimates the elapsed period from contrast agent injection using image analysis, allowing for contrast state determination even without accessible imaging start time information, employing trained regression models and integrating multiple image inputs for robustness and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If imaging time point information is obtained from DICOM tags, then the process is simple and fast, but the information may be missing or incorrect reducing reliability
Solution Approach 1:
The patent introduces an intermediary machine learning model that processes image data to estimate temporal information when DICOM tag information is unavailable or incorrect. The model acts as a mediator between the image data and the required temporal information, using learned patterns from training data to provide accurate estimates without relying solely on potentially flawed metadata.
Solution Approach 2:
The system performs self-service by using the image data itself to generate temporal information estimates when external metadata is unreliable. The machine learning model learns to extract temporal characteristics directly from the image content, allowing the system to determine contrast phases autonomously without depending on external DICOM tag information.
2Adaptability or versatility
If manual data preparation is used for learning-based algorithms, then the algorithm can be trained, but the process is time-consuming and limits applicability across different examination protocols and organs
Solution Approach 1:
The patent implements a universal machine learning model trained on diverse training data encompassing multiple examination protocols and organ types. This single model can estimate temporal information across different CT examinations (e.g., chest, abdomen, pelvis) and various contrast protocols without requiring protocol-specific customization or manual data preparation for each case.
Solution Approach 2:
The system performs preliminary action by pre-training the machine learning model on comprehensive training data that includes various examination protocols and organs before actual use. This advance preparation allows the model to be immediately applicable to new examinations without requiring additional manual data preparation or training for each specific protocol or organ type.
3Reliability
If contrast state determination depends on accurate imaging time point information, then the determination is straightforward, but the method fails when metadata is missing or incorrect
Solution Approach 1:
The patent applies partial action by using only the necessary subset of information for contrast state determination. When DICOM tag information is unreliable, the system relies solely on image data analysis through the machine learning model, using exactly the amount of information needed without requiring complete or perfect metadata. This approach maintains robustness by not depending on excessive or complete metadata availability.
Data Source
AI summary
Provided are a medical image processing device, a medical image processing method, and a program that can estimate temporal information in an image to be processed even in a case where it is difficult to use information attached to the image to be processed. A medical image processing device includes one or more processors and one or more memories that store a program to be executed by the one or more processors. The one or more processors execute commands of the program to receive an input of an image generated by performing contrast imaging (1002) and to estimate an elapsed period from start of injection of a contrast agent in the image on the basis of image analysis of the image (1004).