FFR Computation via ML-Guided Anatomical Model Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for diagnosing coronary artery disease, such as Quantitative Coronary Angiography, lack functional assessment of blood flow and are invasive, while computational fluid dynamics require long computation times for non-invasive hemodynamic index calculations like fractional flow reserve.
Innovation Solution
A method and system for fast non-invasive computation of hemodynamic indices using patient-specific anatomical models generated from medical image data, with trained machine learning models to predict regions requiring user feedback for accurate computation, reducing user interaction and computation time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If computational fluid dynamics algorithms are used for non-invasive hemodynamic index computation, then functional assessment accuracy is improved, but computation time increases significantly
Solution Approach 1:
The system performs preliminary automated generation of patient-specific anatomical models from medical images before hemodynamic computation, and uses machine learning to pre-identify regions requiring user feedback, thereby preparing the computational framework in advance to reduce overall computation time while maintaining accuracy
Solution Approach 2:
The system segments the arterial model into different regions based on machine learning predictions, identifying specific areas where user feedback is required versus areas that can be processed automatically, allowing parallel processing and reducing overall computation time while maintaining functional assessment accuracy in critical regions
2Measurement precision
If pressure wire intervention is performed for FFR measurement, then functional assessment accuracy is improved, but patient risk and procedure complexity increase
Solution Approach 1:
The system creates a digital copy of the patient's arterial anatomy from non-invasive medical images and performs virtual hemodynamic measurements on this digital model, eliminating the need for physical pressure wire insertion while maintaining FFR measurement accuracy through computational simulation
Solution Approach 2:
The system replaces the mechanical pressure wire intervention with a computational fluid dynamics-based virtual measurement system that processes medical images and anatomical models to determine FFR values, thereby eliminating procedural risks while preserving measurement accuracy
3Productivity
If automated anatomical model generation is performed, then user interaction time is reduced, but model accuracy for complex cases may deteriorate
Solution Approach 1:
The system implements an interactive feedback mechanism where machine learning models predict regions requiring user feedback, and clinicians can provide corrective input for those specific areas while leaving automatically processed regions unchanged, allowing the system to learn from and improve upon automated generation while maintaining high processing speed
Solution Approach 2:
The system applies different processing qualities to different regions of the anatomical model based on machine learning predictions, providing high-level automated processing for simple regions while enabling targeted user refinement for complex or critical areas, thereby optimizing both processing speed and model accuracy locally
Data Source
AI summary
A method and system for fast non-invasive computer-based computation of a hemodynamic index, such as fractional flow reserve (FFR) from medical image data of a patient is disclosed. A patient-specific anatomical model of one or more arteries of a patient is automatically generated based on medical image data of the patient. Regions in the automatically generated patient-specific anatomical model for which user feedback is required for accurate computation of a hemodynamic index are predicted using one or more trained machine learning models.