Neural Network Irradiation Map Prediction for Interventional Radiology
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting radiation exposure during interventional radiology are not real-time and lack sufficient accuracy, failing to account for patient-specific anatomy and device position, leading to inadequate dose control.
Innovation Solution
A system utilizing a multilayer neural network that learns from associations between radiology images and acquisition parameters to predict irradiation maps in real-time, combining Monte Carlo simulations and U-Net architecture for fast and accurate dose prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Monte Carlo simulation methods are used to predict irradiation maps, then measurement precision is improved, but computing time increases making it unsuitable for real-time prediction
Solution Approach 1:
The patent pre-calculates and stores irradiation maps for various device positions and acquisition parameters before the actual interventional procedure. These pre-computed maps are stored in a database and can be quickly retrieved and interpolated during the procedure, eliminating the need for real-time Monte Carlo simulations while maintaining high accuracy
Solution Approach 2:
The patent creates a simplified 3D model copy of the patient's anatomy from pre-acquired imaging data (CT, MRI, or ultrasound). This digital twin is then used for rapid prediction of irradiation maps without requiring complex real-time simulations, as the anatomical structure is already captured and stored
2Productivity
If conventional dose tracking systems are used, then computing time is reduced, but measurement precision deteriorates by not accounting for patient anatomy and device position
Solution Approach 1:
The patent introduces a dedicated computing processor that acts as an intermediary between the simple dose tracking system and the complex Monte Carlo simulation. This processor pre-computes and stores lookup tables of irradiation maps for various conditions, enabling fast retrieval and interpolation that maintains accuracy without the computational burden of full simulations
Solution Approach 2:
The patent changes the approach from dynamic real-time simulation to static pre-computation with parameter-based retrieval. By storing irradiation maps for various device positions and acquisition parameters in advance, the system can quickly retrieve and interpolate results based on current parameters without performing complex calculations during the procedure
Data Source
AI summary
A method for obtaining an irradiation map of a patient during interventional radiology, comprising:a learning phase consisting in submitting to a neural network a learning set comprising associations between a first input tensor comprising data of a first radiology image of a patient's intervention area, and first acquisition parameters of an interventional radiology device, and labels corresponding to an irradiation map obtained by simulation from the said first radiology image and the said first acquisition parameters,a prediction phase on a given patient, comprising the acquisition of a stream of second acquisition parameters of said interventional radiology device, the preparation of a second input tensor comprising data of a second radiology image of said given patient and of said second acquisition parameters, the submission of said second input vector to said neural network and the retrieval of an irradiation map prediction.