Perfusion MRI AIF Estimation Using Deep Learning Post-Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating arterial input function (AIF) in perfusion MRI are sensitive to acquisition effects, require detailed knowledge of perfusion sequences, or compromise image resolution, and involve iterative processes that are computationally costly and time-consuming.
Innovation Solution
A post-processing method using deep learning to generate a viable arterial input function (AIF) by learning the disruptive effects of perfusion sequence acquisition and tissue signal redundancy, allowing for a non-iterative, unconstrained estimation of AIF, reducing the number of required tissue signals, and enhancing execution speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing methods are used to estimate AIF in perfusion MRI, then the estimation can be obtained, but the estimation is sensitive to acquisition effects and requires detailed knowledge of perfusion sequences
Solution Approach 1:
The patent introduces an intermediary deep learning model that acts as a mediator between the raw perfusion MRI data and the AIF estimation. This model learns the complex relationship between tissue signals and AIF during training, eliminating the need for detailed knowledge of perfusion sequences during deployment while maintaining estimation accuracy.
Solution Approach 2:
The patent uses a training phase where the model learns from synthetic or labeled data to create an internal representation of the AIF-tissue signal relationship. This learned model then copies this knowledge to process new perfusion data without requiring explicit knowledge of the acquisition parameters.
2Measurement precision
If iterative processes are used to estimate AIF, then the estimation can be obtained, but the process is computationally costly and time-consuming
Solution Approach 1:
The patent performs preliminary action by training the deep learning model offline before actual AIF estimation is needed. During the training phase, the model learns optimal parameter mappings and relationships. When deployed, the pre-trained model can rapidly estimate AIF from new perfusion data without requiring iterative optimization, thus achieving both accuracy and speed.
3Reliability
If conventional perfusion sequences are used, then the acquisition can be performed, but the AIF estimation requires many tissue signals and high computational resources
Solution Approach 1:
The patent changes the parameter representation by transforming the problem from estimating AIF directly from raw perfusion curves to estimating AIF parameters from tissue signal characteristics. The deep learning model learns to map tissue signal parameters to AIF parameters, reducing the complexity of data processing while maintaining reliability.
Data Source
AI summary
A method for post-processing a sampled time-dependent experimental perfusion signal to generate a pharmacokinetic parameter is implemented by a processing unit of a medical-imaging analysis system, said unit having been trained beforehand in a process allowing the disrupting effect of acquisition of a perfusion sequence on arterial signals and redundancy of information related to an arterial input function shared by a set of at least two tissual signals to be learnt. Such an arterial input function is produced directly in a step by said processing unit thus trained from a first arterial input function and from tissual signals selected beforehand.