Systems, devices, and methods for harmonization of imaging datasets including biomarkers
Inventors
Clark, Samuel • WENGLER, Kenneth • HORGA HERNANDEZ, Guillermo
Assignees
Columbia University in the City of New York • Research Foundation for Mental Hygiene Inc • Terran Biosciences Inc
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12079960-B2
Publication Date
2024-09-03
Expiration Date
Abstract
An exemplary system, method and computer-accessible medium for harmonizing neuromelanin (NM) data using combat directly on a NM database or using combat generated coefficients to harmonize future data can include, for example, receiving imaging information of a brain of the patient(s), from one MRI scanner, receiving imaging information of a brain of the patient(s), from a second MRI scanner and using combat to harmonize the data between scanners against a reference dataset. The Neuromelanin (NM) concentration of the patient(s) can then be determined based on the harmonized data. The NM concentration can be determined using a voxel-wise analysis procedure. The voxel-wise analysis procedure can be used to determine a topographical pattern(s) within a substantia nigra (SN) of the brain of the patient(s).
Core Innovation
The invention provides systems and methods for cross-scanner harmonization of neuromelanin-sensitive MRI (NM-MRI) biomarker data. A neuromelanin dataset is harmonized to reduce scanner-induced nonbiological variability while preserving biological variability. The described approach uses an empirical Bayesian ComBat algorithm to generate coefficients for harmonization across MRI machines and vendors.
In one aspect, neuromelanin associated with a first dataset obtained using a set of MRI machines manufactured by a first vendor is measured, and neuromelanin associated with a reference dataset obtained using a reference MRI machine manufactured by a second vendor different from the first vendor is measured. A vendor specific ComBat coefficient is generated based on at least one of the measured levels, and the vendor specific coefficient is applied directly to a third dataset obtained from an MRI machine manufactured by the first vendor to obtain an adjusted neuromelanin level for comparison to the reference levels.
In another aspect, an apparatus and a non-transitory processor-readable medium compute a first set of measures based on plurality of images and imaging source information, provide or search for coefficients associated with the imaging source, apply the coefficients to transform the first set of measures into a harmonized second set having reduced variability associated with identified sources of variability, and link and store the harmonized measures with the coefficients and imaging-source information. The disclosed workflow supports voxel map based masking and biomarker-presence measure generation, and enables downstream neuromelanin analyses such as voxel-wise analysis and machine learning outputs for scanner classification and age prediction.
Claims Coverage
The partial content includes four independent claims (clm-00001, clm-00002, clm-00007, clm-00012), with inventive features covering vendor specific ComBat coefficient harmonization for neuromelanin datasets, remote-server NM-MRI concentration assessment with harmonization and reporting, an apparatus workflow for masked voxel-map measure computation followed by coefficient-based harmonization, and a non-transitory processor-readable medium implementing a database-coefficient harmonization workflow with coefficient lookup and storage. Across these claims, the central inventive theme is harmonizing neuromelanin measures by transforming computed measures using imaging-source-associated coefficients that produce reduced variability.
Vendor specific ComBat coefficient harmonization using measured neuromelanin levels
Measuring a first level of neuromelanin associated with a first dataset obtained using a set of MRI machines manufactured by a first vendor; measuring a second level of neuromelanin associated with second dataset obtained using a reference MRI machine manufactured by a second vendor different from the first vendor; implementing a machine learning (ComBat) algorithm to generate a vendor specific coefficient associated with the set of MRI machines, the vendor specific coefficient being based on at least one of (1) the first level of neuromelanin associated with a set of MRI machines or (2) the second level of neuromelanin associated with the reference MRI machine; obtaining a third dataset associated with a first MRI machine manufactured by the first vendor; and applying the vendor specific coefficient directly on the third dataset from the first MRI machine manufactured by the first vendor to measure an adjusted third level of neuromelanin associated with first MRI machine compared to at least one of (1) the second level of neuromelanin associated with the reference MRI machine or (2) the second dataset.
Remote NM-MRI neuromelanin dataset harmonization with coefficient-based analysis and report generation
Performing a Neuromelanin-Magnetic Resonance Imaging (NM-MRI) scan on a subject; acquiring a neuromelanin dataset from the NM-MRI scan; optionally encrypting the neuromelanin dataset; uploading the neuromelanin dataset to a remote server; optionally decrypting the dataset; harmonizing the dataset by (i) running a combat algorithm on the entire dataset to generate and apply a coefficient, or (ii) applying a scanner specific coefficient to the dataset, or (iii) applying a vendor specific coefficient to the dataset; performing an analysis of the neuromelanin dataset, wherein the analysis comprises one or more of (i) comparing the neuromelanin dataset with one or more previously acquired neuromelanin datasets from the said subject, (ii) comparing the neuromelanin dataset with a control dataset, (iii) comparing the neuromelanin dataset with one or more previously acquired neuromelanin datasets from different subjects; generating a report comprising the neuromelanin analysis; and uploading the report to remote server.
Mask-based voxel map measures with coefficient-based harmonized storage tied to imaging source
Receiving information associated with an imaging source; receiving a first dataset including a first plurality of images associated with the imaging source, each image in the plurality of images includes a set of voxels; applying a mask on the first plurality of images to generate a voxel map, the voxel map being based in an average of values associated with a subset of voxels from the set of voxels; calculating a first set of measures based on the voxel map, the first set of measures being based on indications of a presence of a biomarker in the plurality of images based on which the voxel map was generated; providing the first set of measures to a data harmonization algorithm, the data harmonization algorithm configured to calculate a set of coefficients, the set of coefficients when applied to the first set of measures configured to transform the first set of measures into a harmonized second set of measures, the second set of measures having a reduced variability associated with one or more identified sources of variability, compared to the first set of measures; receiving the second set of measures and the set of coefficients associated with the transformation of the first set of measures into the second set of measures; and linking and store the second set of measures and the set of coefficients with the information associated with the imaging source.
Coefficient lookup and application for harmonized reduced-variability measures stored with imaging-source association
Receive a dataset including a plurality of images; receive information associated with an imaging source associated with the dataset; compute a first set of measures based on the plurality of images; search a database to determine a set of coefficients associated with the imaging source, the set of coefficients configured to transform the first set of measures into a harmonized second set of measures, the second set of measures having a reduced variability associated with one or more identified sources of variability, compared to the first set of measures; obtain the set of coefficients associated with the imaging source; apply the set of coefficients to the first set of measures to generate the second set of measures having the reduced variability associated with the one or more identified sources of variability; and store the second set of measures in association with the set of coefficients and the information associated with an imaging source.
Across clm-00001, clm-00002, clm-00007, and clm-00012, the claims focus on computing neuromelanin-related measures from NM-MRI images and transforming those measures into harmonized measures by applying ComBat-based coefficients that are associated with an imaging source (scanner/vendor). The claims further specify remote upload/report workflows (clm-00002), mask/voxel-map based measure computation (clm-00007), and coefficient retrieval and storage linked to imaging-source information (clm-00012).
Stated Advantages
Reduced variability associated with one or more identified sources of variability compared to the first set of measures.
Improved reproducibility.
Increased statistical power for age-related patterns.
Preserving biological variability while reducing scanner-induced nonbiological variability.
Documented Applications
Cross-scanner harmonization of neuromelanin-sensitive MRI (NM-MRI) biomarker data from MRI machines manufactured by different vendors (e.g., GE/Siemens/Philips).
Downstream analyses including univariate voxelwise analysis and MVPA/SVM scanner classification outputs.
Downstream analysis using SVR for age prediction.
Clinical/hospital setting workflow supporting onboarding/registering new scanners and automation using scanner-specific coefficients and vendor specific coefficients.
Remote portal/workflow with uploading neuromelanin datasets to a remote server and uploading a generated report to the remote server.
Interested in licensing this patent?