Estimating Lung Perfusion from Non-Contrast CT Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Thoracic CT scans do not directly capture functional information about pulmonary perfusion, making it difficult to estimate lung perfusion and detect perfusion defects without additional expensive and elaborate imaging procedures.
Innovation Solution
A computer-implemented method that uses a trained algorithm to estimate lung perfusion from non-contrast CT images by learning from reference perfusion information from modalities like SPECT or dual-energy CT, allowing the algorithm to detect structural changes indicative of perfusion issues without requiring direct perfusion imaging during analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional CT imaging is used to assess lung structure, then anatomical detail is captured, but functional perfusion information is not obtained
Solution Approach 1:
The patent uses a trained machine learning algorithm as an intermediary to translate structural CT image data into functional perfusion information. The algorithm learns the relationship between anatomical structures and perfusion patterns from training data, enabling indirect inference of functional information from structural images without requiring direct functional imaging
Solution Approach 2:
The method creates a virtual copy of perfusion information by training the algorithm to reproduce perfusion patterns observed in reference imaging studies. Once trained, the algorithm generates synthetic perfusion maps that replicate the functional information originally obtained from expensive perfusion imaging modalities
2Loss of information
If expensive perfusion imaging modalities are used to obtain functional information, then direct perfusion data is available, but cost and complexity increase
Solution Approach 1:
The patent makes the conventional CT scanner multi-functional by enabling it to provide both anatomical and functional perfusion information through the integrated algorithm. The same hardware platform performs both structural imaging and functional assessment, eliminating the need for specialized perfusion imaging equipment
Solution Approach 2:
The algorithm creates virtual perfusion images that replicate the information obtained from expensive modalities like SPECT or contrast-enhanced perfusion CT. This copying approach allows functional assessment using only the conventional CT scanner, avoiding the need for additional specialized equipment
3Loss of information
If multiple imaging modalities are used to obtain both structural and functional information, then comprehensive data is available, but workflow complexity and time increase
Solution Approach 1:
The patent merges structural and functional information acquisition into a single imaging workflow. The algorithm processes standard CT images to simultaneously provide both anatomical visualization and functional perfusion assessment, combining what previously required separate imaging sessions into one unified process
Solution Approach 2:
The CT imaging system is enhanced to perform multiple functions - anatomical imaging, perfusion assessment, and defect detection - all within the same imaging session and processing pipeline, eliminating the need for separate functional imaging studies
Data Source
AI summary
The present invention relates to a computer implemented method for estimating lung perfusion from CT images, comprising the steps of: providing a CT image of at least a part of the lung, in particular a CT scan taken at inspiration, and more in particular a non-contrast CT scan taken at inspiration; providing the CT image to a trained computer implemented algorithm to estimate lung perfusion based on the CT image, wherein the trained computer implemented algorithm is trained by providing a set of CT images from at least a part of the lung, in particular a CT scan taken at inspiration, and more in particular a non-contrast CT scan taken at inspiration; providing perfusion information corresponding to the CT image; and training the computer implemented algorithm to learn to estimate perfusion in a CT image based on the reference perfusion information provided during training.