Single-Energy CT Image Transformation Using Anatomical Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Single-energy CT imaging systems lack the increased contrast visualization and are prone to noise and image artifacts, while dual-energy systems are not universally available and limited in scan types, necessitating a method to transform images from a single energy level to multiple energy levels effectively.
Innovation Solution
A deep learning-based approach using a contrast phase classifier and energy transformation models to identify and transform images from a single energy level to a desired energy level, incorporating regression models for tissue-specific mappings and blending techniques to enhance contrast visibility and reduce artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single-energy CT imaging system is used, then the system is widely available and cost-effective, but the image quality lacks contrast visualization and suffers from noise and artifacts
Solution Approach 1:
The patent uses deep learning models to generate synthetic dual-energy images by copying the characteristics of actual dual-energy images from training data. The energy transformation model takes single-energy images and synthesizes corresponding dual-energy images, effectively copying the contrast and quality features without requiring physical dual-energy hardware.
Solution Approach 2:
The patent transforms images by changing the energy level parameter from a single energy level to multiple energy levels. Through the deep learning-based energy transformation model, the system modifies the energy parameter of the images to simulate dual-energy imaging conditions, thereby improving contrast visualization and reducing artifacts.
2Reliability
If a dual-energy CT imaging system is used, then contrast visualization and image quality are improved, but the system is not universally available and limited in scan types
Solution Approach 1:
The patent makes the single-energy CT system multi-functional by enabling it to perform dual-energy imaging tasks through software-based deep learning transformation. The energy transformation model allows the same hardware to generate images at different energy levels, providing versatility without requiring multiple physical systems.
Solution Approach 2:
The system copies the imaging capabilities of dual-energy systems by using trained deep learning models that replicate the contrast and quality characteristics. The energy transformation model generates synthetic dual-energy images from single-energy inputs, effectively copying the functional capabilities without the hardware complexity.
3Reliability
If deep learning models are applied for image transformation, then image quality and contrast are enhanced, but the processing time and computational complexity increase
Solution Approach 1:
The patent performs preliminary action by pre-training deep learning models offline using extensive training data and computational resources. Once trained, the models can be deployed for inference, which significantly reduces the processing time required during actual image transformation compared to training time.
Solution Approach 2:
The patent replaces complex mechanical dual-energy imaging systems with a software-based deep learning approach. This substitution trades computational complexity during training for faster inference time, effectively replacing hardware complexity with optimized software processing.
Data Source
AI summary
Methods and systems are provided for transforming images from one energy level to another. In an example, a method includes obtaining an image at a first energy level, identifying a contrast phase of the image, entering the image as input to a segmentation model trained to output an anatomy mask that identifies each tissue type in the image, generating a guide image from the image and the anatomy mask using a regression model, entering the image and the guide image as input into an energy transformation model trained to output a transformed image at a different, second energy level, the energy transformation model selected from among a plurality of energy transformation models based on the contrast phase, and displaying a final transformed image and/or saving the final transformed image in memory, wherein the final transformed image is the transformed image or is generated based on the transformed image.