Medical Image Processing Apparatus for Low-Dose CT Artifact Reduction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional X-ray CT image processing methods face challenges with low-dose image capturing, particularly at sites with high X-ray absorption, leading to reduced data counts, dark band artifacts, streak artifacts, and decreased accuracy of CT values, especially when noise is present.

Innovation Solution

A medical image processing apparatus utilizing a trained machine learning model applies artifact reduction to X-ray CT scan data, specifically trained using projection data with artificially generated low count artifacts and noise, to produce image data with reduced artifacts and noise.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of energy

If low-dose X-ray CT image capturing is performed, then radiation exposure is reduced, but data count is reduced leading to artifacts and decreased accuracy

Engineering Contradiction:

Improveradiation exposureVSAvoidaccuracy of CT value

Core Design Contradiction:

Loss of energyVSMeasurement precision

Solution Approach 1:

A machine learning model is introduced as an intermediary between the low-count projection data and the final reconstructed image. The model learns the mapping relationship from low-count data to high-quality images through training with simulated data, enabling artifact reduction without requiring high-dose acquisition

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

Training data with low count artifacts is generated in advance through simulation before actual low-dose imaging. The machine learning model is pre-trained with this simulated data to learn how to correct artifacts, enabling effective artifact reduction when processing actual low-dose images

Inventive Principle:

Principle #10Preliminary action

2Productivity

If conventional artifact reduction methods are used with low count data, then processing speed is maintained, but artifact reduction effectiveness is insufficient

Engineering Contradiction:

Improveprocessing speedVSAvoidartifact reduction effectiveness

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

Conventional iterative reconstruction methods are replaced with a machine learning-based approach. The trained model directly processes low-count data to generate artifact-reduced images, achieving both speed and effectiveness by leveraging learned patterns rather than iterative computation

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS20240169531A1Medical image processing apparatus, medical image processing method, and model generation method

Publication Date: 2024.05.23 CANON KK

AI summary

A medical image processing apparatus according to one embodiment includes processing circuitry. The processing circuitry acquires second image data in which a low count artifact is reduced, by applying a trained machine learning model to first image data that is obtained by X-ray CT scan. The processing circuitry outputs image data based on the second image data. The machine learning model is trained by using training data that includes third image data and fourth image data, where the third image data is reconstructed based on projection data that is obtained by X-ray CT scan and the fourth image data is based on the projection data and includes a generated low count artifact.