Systems and methods for generating biomarkers based on multivariate classification of functional imaging and associated data

Patent metadata

Field	Value
Publication number	US-2016019693-A1
Application number	US-201514800212-A
Country	US
Kind code	A1
Filing date	Jul 15, 2015
Priority date	Jul 15, 2014
Publication date	Jan 21, 2016
Grant date	—

Abstract

Systems and methods for generating biomarkers associated with neuropsychiatric disorders, neurodevelopmental disorders, neurobehavioral disorders, or other neurological disorders are described. In general, the biomarkers are generated based on correlations between functional imaging data and clinical acquired from a subject, as computed using a multivariate classifier. Functional imaging data may include functional magnetic resonance images, or activation maps generated from such images. Clinical data generally includes data associated with a clinical or behavioral characterization of the subject. The biomarkers can be used to monitor or otherwise assess a treatment response; to provide diagnostic information, such as subtyping or classifying a disorder; to provide prognostic information, such as a prediction of treatment response or outcome; or to indicate functional or anatomical targets for treatments.

First claim

1 . A computer-implemented method for generating a biomarker associated with a neuropsychiatric, neurodevelopmental, neurobehavioral, or other neurological disorder, the steps of the method comprising:
(a) providing to a computer system, functional imaging data acquired from a subject's brain; (b) providing to the computer system, clinical data associated with the subject; and (c) generating with the computer system, a biomarker associated with the neuropsychiatric, neurodevelopmental, neurobehavioral, or other neurological disorder by computing a correlation between the functional imaging data and the clinical data using a multivariate classifier.

2 . The method as recited in claim 1 , wherein step (c) includes forming a matrix having rows that correspond to regions in the subject's brain and having columns that correspond to the functional imaging data and the clinical data, and wherein computing the correlation between the functional imaging data and the clinical data includes inputting the matrix to the multivariate classifier.

3 . The method as recited in claim 2 , wherein the computer system is used to perform at least one of dimensionality reduction or feature extraction on the matrix before inputting the matrix to the multivariate classifier.

4 . The method as recited in claim 1 , wherein the biomarker generated by the computer system includes covariance patterns based on the correlation between the functional imaging data and the clinical data computed by the computer system using the multivariate classifier.

5 . The method as recited in claim 1 , wherein the biomarker generated by the computer system includes co-varying traits based on the correlation between the functional imaging data and the clinical data computed by the computer system using the multivariate classifier.

6 . The method as recited in claim 1 , wherein the biomarker generated by the computer system includes an interregional correlation matrix based on output of the multivariate classifier.

7 . The method as recited in claim 1 , wherein the biomarker generated by the computer system includes an association matrix based on output of the multivariate classifier.

8 . The method as recited in claim 1 , wherein step (c) includes using the computer system to extract at least one of a qualitative characteristic, a quantitative characteristic, a qualitative index, or a quantitative index indicative of interaction patterns in brain network organization based on the computed correlation between the functional imaging data and the clinical data.

9 . The method as recited in claim 1 , wherein the multivariate classifier is based on a principal component analysis and the biomarker includes a group level spatial component image output from the principal component analysis.

10 . The method as recited in claim 9 , wherein the biomarker further includes a loading score computed by the computer system from the group level spatial component images.

11 . The method as recited in claim 1 , wherein the multivariate classifier is based on a thresholding correlation analysis, and the biomarker includes a group level interregional correlation map output from the thresholding correlation analysis.

12 . The method as recited in claim 11 , wherein the thresholding correlation analysis includes a seed analysis.

13 . The method as recited in claim 1 , wherein the multivariate classifier is based on a hierarchical clustering analysis, and the biomarker includes a cluster map that depicts clusters or networks of brain regions with similar activation levels across subsets of valenced conditions.

14 . The method as recited in claim 13 , wherein the biomarker further includes linearly separable co-varying patterns identified with the computer system in the cluster map.

15 . The method as recited in claim 1 , wherein the multivariate classifier is based on a machine learning algorithm, and the biomarker includes a report that indicates clusters of brain regions defined by brain-wide activity and connectivity levels across valenced conditions.

16 . The method as recited in claim 15 , wherein the machine learning algorithm is trained on a database that includes at least one of functional imaging data, other imaging data, physiological data, clinical data, genetic data, and epigenetic data associated with the neuropsychiatric, neurodevelopmental, neurobehavioral, or other neurological disorder.

17 . The method as recited in claim 15 , wherein the machine learning algorithm includes at least one of a neural network or a support vector machine.

18 . The method as recited in claim 1 , wherein the multivariate classifier is based on an algorithm that estimates graph theory-based network organizational measures, and the biomarker indicates topological features in functional connectivity patterns across dimensional domains.

19 . The method as recited in claim 1 , wherein the functional imaging data includes at least one of functional magnetic resonance images acquired from the subject while the subject was performing a functional task, or functional magnetic resonance images acquired from the subject while the subject was in a resting state.

20 . The method as recited in claim 19 , wherein step (a) includes generating from the functional magnetic resonance images and with the computer system, activation maps that depict neuronal activation patterns associated with at least one of the functional task or the resting state.

21 . The method as recited in claim 20 , wherein the activation maps are generated with the computer system by using a multi-level mixed-effects statistical model.

22 . The method as recited in claim 21 , wherein the multi-level mixed-effects statistical model includes a nested random-effects structure.

23 . The method as recited in claim 22 , wherein the multi-level mixed-effects statistical model further includes an intra-subject power variance function and an autoregressive correlation structure.

24 . The method as recited in claim 1 , further comprising providing to the computer system additional data associated with system-level biological measures of the subject, and wherein step (c) includes generating the biomarker by computing a correlation between the functional imaging data, the clinical data, and the additional data using the multivariate classifier.

25 . The method as recited in claim 24 , wherein the additional data includes at least one of other imaging data, physiological data, genetic data, or epigenetic data.

26 . The method as recited in claim 1 , wherein step (c) includes generating a classifier map using the computer system to map the correlation to a multidimensional parametric space having dimensions associated with the functional imaging data and the clinical data.

27 . The method as recited in claim 26 , wherein the biomarker includes a quantitative metric computed calculated by the computer system from the classifier map, and wherein step (c) includes generating a report based on the quantitative metric.

28 . The method as recited in claim 27 , wherein the quantitative metric is a center-of-mass of correlated data associated with a particular classified group.

29 . The method as recited in claim 27 , wherein the quantitative metric is a distance of the subject's correlated data from a center-of-mass of correlated data associated with a

Assignees

• Brigham & Womens Hospital

Inventors

• Silbersweig David
• Stern Emily
• Pan Hong

Classifications

• G16H50/50 Primary

  for simulation or modelling of medical disorders · CPC title

• G06T7/0012 Primary

  Biomedical image inspection · CPC title

• G06T11/26

  Drawing of charts or graphs · CPC title

• G16H50/20

  for computer-aided diagnosis, e.g. based on medical expert systems · CPC title

• G06T2207/30016

  Brain · CPC title