System for machine learning-based model training and prediction for evaluation of pain
AssigneeAclarion, Inc.
InventorsJames Charles Abel, James Clayton Peacock, III, Ricardo Dario Pradenas
Life of this patent
Estimating the term…
Abstract
Systems and methods are provided for using machine learning (ML) models to evaluate data generated during MRI procedures, such as magnetic resonance spectroscopy (MRS) data, to evaluate patient disc pain. An ML-based disc assessment system can evaluate MRS spectrum data ML models trained to classify the MRS spectrum based on diagnostic and/or outcome-based ground truth data.
Claims22
A computer-implemented method for providing diagnostic information for a medical condition associated with a region of interest (ROI) in a patient, the computer-implemented method comprising:
- obtaining, by a diagnostic system comprising computer-readable memory and one or more computer processors, processed MRS spectrum data representing output of a magnetic resonance spectroscopy (MRS) system, wherein the output of the MRS system comprises MRS acquisition data for a voxel located within the ROI;
- extracting, by the diagnostic system, a subset of the processed MRS spectrum data corresponding to a chemical shift range of interest ranging from a lower chemical shift range boundary to an upper chemical shift range boundary, wherein the subset of the processed MRS spectrum data comprises a plurality of data points, and wherein a data point of the plurality of data points represents an amplitude value associated with a chemical shift value within the chemical shift range of interest;
- generating, by the diagnostic system, a feature vector comprising at least eight (8) dimensions, wherein each dimension comprises a representative value for a plurality of amplitude values in a different one of a plurality of subranges of the chemical shift range of interest;
- processing, by the diagnostic system, the feature vector using an ML model trained to determine, from the representative values for each of the plurality of subranges, a classification regarding the medical condition associated with the ROI;
- and
- generating, by the diagnostic system, the diagnostic information for the medical condition using the classification.
A system for providing diagnostic information for a medical condition associated with a region of interest (ROI) in a patient, the system comprising:
- computer-readable memory storing executable instructions;
- and
- one or more computer processors in communication with the computer-readable memory and programmed by the executable instructions to:
- obtain processed MRS spectrum data representing output of a magnetic resonance spectroscopy (MRS) system, wherein the output of the MRS system comprises MRS acquisition data for a voxel located within the ROI;
- extract a subset of the processed MRS spectrum data corresponding to a chemical shift range of interest ranging from a lower chemical shift range boundary to an upper chemical shift range boundary, wherein the subset of the processed MRS spectrum data comprises a plurality of data points, and wherein a data point of the plurality of data points represents an amplitude value associated with a chemical shift value within the chemical shift range of interest;
- generate a feature vector comprising at least eight (8) dimensions, wherein each dimension comprises a representative value for a plurality of amplitude values in a different one of a plurality of subranges of the chemical shift range of interest;
- process the feature vector using an ML model trained to determine, from the representative values for each of the plurality of subranges, a classification regarding the medical condition associated with the ROI;
- and
- generate the diagnostic information for the medical condition using the classification.
Dependencies are inferred from the claim text.