Pediatric X-Ray Bone Suppression Using Synthetic CT Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
X-ray tomography images, particularly in chest X-rays, suffer from varying image quality due to factors like X-ray intensity, capturing conditions, patient posture, and anatomical complexity, making it difficult to obtain clear and useful information for diagnosis.
Innovation Solution
A method involving a bone suppression model trained using CT image pairs to suppress bone structures in X-ray images, utilizing a style conversion model to enhance image quality and a supervised learning approach to improve visibility of internal organs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If dual energy X-ray capture is used to suppress bone structures, then visibility of internal organs is improved, but device complexity and cost increase
Solution Approach 1:
The patent creates synthetic dual-energy X-ray images by projecting 3D CT data (which contains bone suppression information) onto 2D planes to generate synthetic views that mimic dual-energy X-ray images. This copying approach allows the system to achieve bone suppression effects without requiring actual dual-energy X-ray hardware, thus resolving the contradiction between improved organ visibility and device complexity
Solution Approach 2:
The patent introduces CT data as an intermediary medium. Instead of directly using complex dual-energy X-ray capture, the system uses CT scans (a different imaging modality) as an intermediate step to extract bone suppression information, which is then applied to enhance the visibility of internal organs in X-ray images
2Measurement precision
If dual energy X-ray capture is used to suppress bone structures, then image processing capability is improved, but ease of manufacture deteriorates
Solution Approach 1:
The patent synthesizes dual-energy X-ray image pairs by projecting 3D CT data onto 2D planes. This copying method allows the system to generate training data and achieve bone suppression effects without manufacturing complex dual-energy X-ray capture systems, thereby improving image processing capability while maintaining ease of manufacture
Solution Approach 2:
The patent changes the approach from hardware-based dual-energy capture to software-based synthesis by altering the data representation parameters. By working with projected 2D images derived from 3D CT data instead of actual dual-energy X-ray pairs, the system achieves advanced image processing capability through parameter transformation rather than complex hardware manufacturing
3Measurement precision
If bone suppression is applied to X-ray images, then visibility of internal organs is improved, but image quality varies due to capturing conditions
Solution Approach 1:
The patent generates synthetic dual-energy X-ray images by projecting 3D CT data, which provides consistent and reliable source material. This copying approach ensures that the bone suppression effect is derived from stable 3D anatomical data rather than variable 2D X-ray capture conditions, thereby improving both organ visibility and image quality consistency
Solution Approach 2:
The patent performs bone suppression processing in advance by working with 3D CT data before generating the final 2D projected images. By pre-processing the 3D volumetric data to separate bone structures, the system ensures consistent suppression effects that are not affected by variations in 2D X-ray capturing conditions such as patient posture or breathing
Data Source
AI summary
A medical image processing method for processing a pediatric simple X-ray image using a machine learning model and a medical image processing apparatus therefor are provided. The apparatus applies a style conversion model to a CT image pair including a basic CT image and a suppression CT image where at least a portion of a bone of the basic CT image is suppressed to convert the CT image pair into a conversion image pair and trains a bone suppression model for a simple X-ray image based on training data including the conversion image pair.