X-Ray AI Training from Multimodal Imaging Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing x-ray imaging lacks sufficient training data for AI solutions due to limited availability of x-ray data sets, especially for less prevalent diseases like idiopathic pulmonary fibrosis, leading to delayed diagnosis and inadequate patient management.
Innovation Solution
Adapt large collections of data from other imaging modalities like CT, MRI, and ultrasound to generate synthetic x-ray projections for training AI networks, enhancing x-ray imaging capabilities in patient management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If projection x-ray imaging is used as the primary modality, then ease of operation and accessibility are improved, but diagnostic accuracy and detection capability for certain diseases deteriorate
Solution Approach 1:
The patent introduces AI solutions as an intermediary between projection x-ray imaging and CT imaging. The AI system processes projection x-ray images to detect disease conditions that would otherwise require CT imaging, thereby maintaining the accessibility and ease of operation of x-ray while improving diagnostic accuracy through intelligent analysis algorithms
2Quantity of substance
If AI solutions are trained using only x-ray data, then data availability for training is improved, but training data quantity and diversity deteriorate
Solution Approach 1:
The patent merges multiple data sources including projection x-ray images, CT images, and electronic health record data into a unified training dataset. This combination approach increases both the quantity and diversity of training data available for AI model development, allowing the system to learn from multiple imaging modalities and clinical contexts simultaneously
Solution Approach 2:
The patent transitions from two-dimensional projection x-ray images to three-dimensional CT volumetric data as part of the training dataset. This dimensional expansion provides the AI system with deeper anatomical information and spatial context, enriching the training data beyond what is available from flat x-ray projections alone
3Measurement precision
If CT imaging is used for differential diagnosis, then diagnostic accuracy is improved, but healthcare costs and patient accessibility deteriorate
Solution Approach 1:
The patent creates a virtual copy of CT-level diagnostic capability through AI algorithms that analyze projection x-ray images. Instead of requiring actual CT imaging for every diagnostic scenario, the AI system generates CT-equivalent diagnostic information from more accessible x-ray images, thereby maintaining diagnostic accuracy while reducing the need for expensive and less accessible CT scanners
Data Source
AI summary
A method to enhance the clinical value and role of x-ray imaging as part of the diagnostic and patient management chain using artificial intelligence (AI). The present invention adapts large collections of data acquired with modalities other than projection x-ray to enhance AI training that will enable x-ray imaging to more effectively be used in the full span of patient management from detection and diagnosis to management and therapy.