Virtual Monochromatic CT Imaging From Single-Energy Scans
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Dual energy CT devices are expensive, and many medical institutions have not introduced CT systems compatible with dual energy CT, necessitating a method to generate virtual monochromatic X-ray images with different energy levels using single energy CT systems.
Innovation Solution
A neural network is trained with dual energy CT data to infer virtual monochromatic X-ray images, allowing single energy CT systems to generate images with varying energy levels by inputting a single energy CT image and outputting multiple virtual monochromatic X-ray images with different energy levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If dual energy CT systems are used to generate virtual monochromatic X-ray images with different energy levels, then lesion detection accuracy is improved, but device cost and complexity increase
Solution Approach 1:
The patent creates virtual copies of dual-energy CT imaging capabilities by training a neural network to synthesize virtual monochromatic images at different energy levels from single-energy CT data. This allows institutions with only single-energy CT systems to access the diagnostic benefits of multi-energy imaging without purchasing expensive dual-energy equipment
Solution Approach 2:
The patent changes the energy level parameter of virtual monochromatic images through neural network processing. By inputting a single-energy CT image and having the trained neural network generate multiple output images at different virtual energy levels (e.g., 40 keV, 70 keV, 100 keV), the system effectively varies the energy parameter without physical hardware changes
2Adaptability or versatility
If dual energy CT systems are introduced to produce multiple energy level images, then material differentiation capability is improved, but device cost increases
Solution Approach 1:
The patent makes single-energy CT systems universally capable of producing multi-energy virtual monochromatic images through neural network processing. The same single-energy CT hardware can generate images at multiple virtual energy levels, enabling material differentiation capabilities previously exclusive to dual-energy systems
Solution Approach 2:
The neural network creates virtual copies of dual-energy imaging functionality, allowing single-energy CT systems to simulate the material differentiation capabilities of dual-energy systems by generating synthetic multi-energy images from single-energy input data
3Device complexity
If single energy CT is used to image subject body, then device cost is reduced, but CT value differentiation between different materials deteriorates
Solution Approach 1:
The patent compensates for the limited CT value differentiation of single-energy CT by using neural network processing to generate virtual images at different energy levels. This allows post-processing enhancement of material differentiation without requiring expensive dual-energy hardware
Data Source
AI summary
Systems and methods are described, which generate a plurality of virtual monochromatic X-ray images having different energy levels even with a CT system with single energy CT. An example CT system includes an X-ray tube in which a prescribed tube voltage (120 (kVp)) is applied and one or more processors. The one or more processors perform an operation including inputting a CT image generated based on the single energy CT data collected from a subject body to a trained neural network (94), and causing the trained neural network to infer 40 (keV), 50 (keV), 60 (keV), 80 (keV), 90 (keV), and 100 (keV) virtual monochromatic X-ray images based on the CT image.