AI Analysis of Photon-Counting CT Using Virtual Spectral Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional AI and machine learning systems are not suited for processing medical imaging data acquired via photon counting in CT imaging, which lacks effective methods for analyzing spectral imaging data to assess coronary microvascular dysfunction.
Innovation Solution
A system utilizing AI/ML-based networks to perform medical imaging analysis tasks on photon counting CT (PCCT) data, generating multiple virtual images and combining their results through statistical, learned, or context-based weightings to achieve comprehensive medical imaging analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional AI and machine learning systems are used for processing medical imaging data, then general medical image analysis can be performed, but they are not suited for processing spectral imaging data acquired via photon counting
Solution Approach 1:
The patent transforms spectral imaging data by converting photon counting data into virtual monoenergetic images at different energy levels (e.g., 40-140 keV). This parameter transformation enables conventional AI systems to process the data effectively while preserving spectral information, thus resolving the incompatibility between conventional AI systems and photon counting data
2Measurement precision
If multiple virtual images are generated and processed through multiple machine learning networks, then comprehensive medical imaging analysis is achieved, but the system complexity increases
Solution Approach 1:
The patent divides the complex analysis task into multiple sub-tasks, each handled by a specialized machine learning network. Different virtual images (virtual non-contrast, virtual iodine, virtual pure lumen images) are processed by separate networks, and the results are combined through weighted fusion. This segmentation allows each network to focus on specific features, improving diagnostic accuracy while managing system complexity through modular architecture
Solution Approach 2:
The patent combines results from multiple machine learning networks through weighted fusion, where each network's output is assigned a weight based on its reliability or performance metrics. This merging strategy integrates complementary information from different virtual images and networks, achieving comprehensive diagnostic accuracy while systematically managing the complexity through structured result integration
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Systems and methods for performing a medical imaging analysis task from photon counting computed tomography, PCCT, imaging data are provided. PCCT imaging data acquired from a PCCT imaging device is received. A plurality of PCCT virtual images is generated from the PCCT imaging data. A plurality of medical imaging analysis sub-tasks is performed based on the plurality of PCCT virtual images using a plurality of machine learning based networks. Results of the medical imaging analysis sub-tasks are combined to generate results of a medical imaging analysis task. The results of the medical imaging analysis task are output.