Neural Network Radiation Dose Prediction for Accurate, Fast Planning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing radiation treatment plans lack a balance between accuracy and speed in determining radiation doses, as current methods do not adequately discriminate between target volumes and adjacent tissues, leading to potential collateral damage.
Innovation Solution
A neural network, such as a transformer neural network, is trained using a training corpus of radiation doses and input items like patient images, fluence maps, and target dose volume information to determine radiation doses, allowing for accurate and efficient dose calculation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional optimization processes are used to calculate radiation dose deposition, then treatment plans can be generated, but the calculation speed is slow and accuracy is insufficient to meet all application needs
Solution Approach 1:
The patent creates a neural network model that copies and learns from numerous pre-calculated dose deposition examples generated through traditional optimization processes. The network stores patterns of dose deposition in a training corpus, enabling it to rapidly predict doses for new treatment plans without performing slow traditional calculations, thus achieving both accuracy and speed
Solution Approach 2:
The system performs preliminary actions by pre-calculating and storing dose deposition data for many different treatment scenarios in advance. This training corpus is built before actual treatment planning, so when a new plan is needed, the neural network can immediately retrieve and apply relevant patterns without performing time-consuming calculations from scratch
Data Source
AI summary
A neural network (such as a transformer neural network) configured to determine radiation doses can be trained using a training corpus that comprises a plurality of different resultant radiation doses (such as, but not limited to, sparsely-written resultant radiation doses) and a plurality of different input items that each correspond to a particular one of the plurality of different resultant radiation doses. Those input items can comprise at least one, two, three, or each of a patient image (such as, but not limited to, computed tomography imagery and/or Digital Imaging and Communications in Medicine-compatible imagery), a fluence map, radiation treatment platform geometry information, and/or target dose volume information (such as, but not limited to, sparsely-read target dose volume information). A radiation dose for a patient can be generated by providing patient image information as input to a trained neural network and outputting a determined radiation dose for the patient.