Magnetic Resonance Fingerprinting Outlier Detection Without Pre-Calculated Dictionary
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Magnetic Resonance Fingerprinting (MRF) techniques face challenges in identifying abnormal tissues without a pre-calculated dictionary, leading to errors in voxel composition analysis, particularly when unknown substances are present, making it difficult to detect anomalies such as tumorous tissue.
Innovation Solution
A machine learning algorithm is trained using MRF data from subjects without known anatomical anomalies to calculate an outlier map for each voxel, assigning an outlier score based on the deviation from normal MRF vectors, enabling identification of abnormal regions without relying on a pre-existing dictionary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a dictionary of pre-calculated signals is used in MRF, then the analysis of known substances is accurate, but unknown substances cannot be detected and errors occur in voxel composition analysis
Solution Approach 1:
The patent applies preliminary action by training a machine learning algorithm in advance using MRF data from healthy subjects. This pre-trained model establishes a baseline of normal tissue characteristics before actual analysis, enabling the system to detect deviations from normality without requiring a pre-calculated dictionary of all possible substances.
Solution Approach 2:
The patent introduces an intermediary element - a machine learning algorithm that acts as a mediator between the MRF signal and the interpretation of tissue composition. Instead of directly comparing signals to a dictionary, the algorithm learns the complex patterns of normal tissue and uses this knowledge to identify anomalies, thereby detecting unknown substances indirectly through deviation from normal patterns.
2Ease of manufacture
If a pre-calculated MRF dictionary is prepared, then conventional MRF analysis can be performed, but it becomes difficult to identify abnormal tissues without dictionary coverage
Solution Approach 1:
The machine learning algorithm serves as an intermediary that enhances the reliability of abnormal tissue detection while maintaining the ease of conventional MRF analysis. It processes the MRF signals to identify patterns indicative of pathology, providing a secondary layer of analysis that works complementarily with traditional dictionary-based methods.
Solution Approach 2:
The system implements feedback by using the trained machine learning model to continuously evaluate MRF signals and provide information about abnormal tissues. This feedback mechanism allows clinicians to identify regions requiring further investigation, improving the overall reliability of the diagnostic process while maintaining workflow simplicity.
3Difficulty of detecting and measuring
If machine learning algorithm is trained on healthy subject data, then outlier detection becomes possible, but the system requires additional training data preparation
Solution Approach 1:
The patent applies preliminary action by performing the time-consuming training data preparation and model training in advance, before actual clinical use. Once trained, the machine learning algorithm can rapidly detect anomalies in patient data without requiring repeated training, thus converting upfront time investment into ongoing operational efficiency.
Solution Approach 2:
The system creates a computational model (copy) of healthy tissue patterns through training. This learned representation serves as a reference that can be rapidly applied to detect deviations in patient data, eliminating the need for repeated analysis of training data and reducing time loss during actual diagnostic operations.
Data Source
AI summary
The invention provides for a medical imaging system comprising: a memory for storing machine executable instructions; a processor for controlling the medical instrument. Execution of the machine executable instructions causes the processor to: receive MRF magnetic resonance data acquired according to an MRF magnetic resonance imaging protocol of a region of interest; reconstruct an MRF vector for each voxel of a set of voxels descriptive of the region of interest using the MRF magnetic resonance data according to the MRF magnetic resonance imaging protocol; calculate a preprocessed MRF vector (126) for each of the set of voxels by applying a predetermined preprocessing routine to the MRF vector for each voxel, wherein the predetermined preprocessing routine comprises normalizing the preprocessed MRF vector for each voxel; calculate an outlier map for the set of voxels by assigning an outlier score to the preprocessed MRF vector using a machine learning algorithm.