X-ray Image Generation Device Using DRR Trained Machine Learning for Bone Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing X-ray imaging techniques face challenges with misalignment artifacts due to subject motion, particularly in bone suppression and angiographic imaging, and struggle to clearly display bioabsorbable stents, which are not recognizable in X-ray fluorography.
Innovation Solution
An image generating device using machine learning with DRR images to simulate geometric fluoroscopy conditions, generating trained models for recognizing specific regions such as bone, blood vessels, or stents, and converting X-ray images to isolate or enhance these regions, thereby reducing artifacts and exposure dose.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If dual energy subtraction imaging is performed to remove bone portions, then bone suppression is achieved, but misalignment artifacts occur due to subject motion between two imaging times
Solution Approach 1:
The system performs preliminary segmentation of bone portions from a single X-ray image using machine learning models trained on DRR images, eliminating the need for sequential imaging and thus preventing motion-induced misalignment artifacts while achieving bone suppression
2Manufacturing precision
If digital subtraction angiography is performed to visualize blood vessels, then vessel visibility is improved, but misalignment artifacts occur due to subject motion between mask and live image acquisition
Solution Approach 1:
The system performs preliminary segmentation of blood vessels from a single X-ray image using machine learning models trained on DRR images, eliminating the need for sequential mask and live imaging and thus preventing motion-induced misalignment artifacts while achieving vessel visualization
3Manufacturing precision
If multiple X-ray images are accumulated to display stent, then stent visibility is improved, but exposure dose increases
Solution Approach 1:
The system extracts stent information from a single X-ray image using machine learning models trained on DRR images, eliminating the need for multiple image accumulations and thus reducing exposure dose while maintaining stent visibility
4Object-affected harmful factors
If bioabsorbable stent is used, then patient safety is improved, but stent visibility in X-ray fluorography is lost
Solution Approach 1:
The system uses machine learning models trained on DRR images as an intermediary to infer and visualize stent locations from standard X-ray fluorography images, enabling visualization of bioabsorbable stents that are inherently invisible in conventional X-ray imaging while maintaining patient safety
Data Source
AI summary
An image generating device for generating an image which is an X-ray image of an area including a bone portion of a subject with the bone portion removed has a control unit 70 including: a DRR imager 81 configured to generate a first DRR image of an area including a bone portion and a second DRR image showing the bone portion, by performing, for a set of CT image data of an area including the bone portion of a subject, a virtual fluoroscopic projection simulating a geometric fluoroscopy condition of an X-ray irradiator and an X-ray detector for the subject; a training section 82 configured to generate a machine learning model for recognizing the bone portion, by performing machine learning using the first DRR image and the second DRR image serving as a training image; an image converter 83 configured to perform conversion of the X-ray image of the area including the bone portion of the subject, using the machine learning model trained in the training section 82, to generate an image showing the bone portion; and a bone portion subtractor 84 configured to subtract the image showing the bone portion from the X-ray image of the area including the bone portion of the subject.