Generating functional brain mappings and accompanying reference information
A method for mapping functionally related brain regions of a subject. The method includes receiving a structural brain image and a dataset of resting-state functional MRI (rs-fMRI) three-dimensional (3D) image frames of a brain of the subject includes 3D image frames of the subject's brain over time. A functional connectivity map identifying groupings of functionally connected voxels is overlaid on the structural brain image. Spontaneous brain activations associated with each voxel in a grouping of functionally connected voxels are time-correlated with spontaneous brain activations in one or more other voxels in the grouping of functionally connected voxels. A reference location map for a pre-defined resting state network is generated, including reference information output. The reference location map includes brain reference locations that are informative of the one or more resting-state networks to assist a user in identifying the resting state network of the functional connectivity map.
1 . A computer implemented method for generating outputs and accompanying reference information in a software-based system, comprising:
generating a primary system output using a machine-learning based algorithm created through an algorithm training process that uses training data, wherein the primary system output comprises a mapping of defined functional brain networks within a brain of an individual subject based upon image data pertaining to the brain of the individual subject; and
separately generating an accompanying reference output of reference information informative of the training data used to create the machine-learning based algorithm, wherein the reference output comprises population-averaged location information for the defined functional brain networks.
2 . The computer implemented method of claim 1 , wherein the primary system output comprises the mapping of defined functional brain networks overlaid upon a structural image including the brain of the individual subject.
3 . The computer implemented method of claim 1 , wherein the population-averaged location information for the defined functional brain networks comprises, for each defined network, a set of population-averaged reference locations falling within bounds of that defined network, wherein the reference output comprises the sets of population-averaged reference locations overlaid upon a structural image of a brain, and wherein the structural image of the brain upon which the sets of population-averaged reference locations are overlaid is a structural image of the brain of the individual subject.
4 . The computer implemented method of claim 1 , wherein the image data are functional magnetic resonance imaging (fMRI) data, wherein the fMRI data are resting-state fMRI (rs-fMRI) data acquired during a time the individual subject is in a state of rest, wherein the functional brain networks are a defined set of one or more resting state networks (RSNs) comprising a topology of RSNs, and wherein the RSNs included in the topology are macro-scale RSNs present within a broad range of healthy subjects.
5 . The computer implemented method of claim 4 , wherein a number of macro-scale RSNs included in the topology of RSNs equals six to eight RSNs.
6 . The computer implemented method of claim 4 , wherein the macro-scale RSNs include brain networks making up an eloquent cortex of the brain of the individual subject, wherein the brain networks making up the eloquent cortex include at least a sensorimotor network (SMN) and a language network (LAN), wherein the brain networks making up the eloquent cortex further include a vision network (VIS).
7 . The computer implemented method of claim 4 , wherein the macro-scale RSNs comprise one or more of: a ventral attention network (VAN), a dorsal attention network (DAN), a frontoparietal control network (FPC), and a default mode network (DMN).
8 . The computer implemented method of claim 4 , wherein generating the primary system output comprises generating a set of functional connectivity maps, and wherein each functional connectivity map identifies functionally connected voxels of the brain of the individual subject included in one RSN of the topology of RSNs.
9 . The computer implemented method of claim 8 , wherein generating the primary system output comprises generating, based upon the rs-fMRI data for the individual subject, a voxel-wise correlation map of the brain of the individual subject comprising measures of correlation between MR signals at different voxel pairs throughout the brain, wherein generating the voxel-wise correlation map comprises calculating, for each voxel pair, a Pearson product-moment correlation coefficient for an MR signal at one voxel compared to an MR signal at a second voxel, wherein the calculation of the correlation coefficient yields a single scalar value representing a measure of strength in linear association between the two MR signals of the voxel pair.
10 . The computer implemented method of claim 8 , wherein generating the primary system output further comprises generating the set of functional connectivity maps by assigning voxels of the brain of the individual subject to the defined set of one or more RSNs of the topology of RSNs, wherein the assignment of voxels to the defined set of one or more RSNs involves use of the machine-learning based algorithm, wherein the machine-learning based algorithm is trained, using a supervised learning process, to assign voxels using pattern matching that applies weight in determining the assignment to the following factors: within-network MR signal patterns being correlated in time, and global patterns in MR signals throughout the brain and between different networks.
11 . The computer implemented method of claim 10 , wherein the supervised learning process uses, for each RSN of the topology of RSNs, a set of reference locations in the brain representing population-based locations that fall within the RSN, and wherein the supervised learning process uses rs-fMRI datasets acquired from a number of individual subjects.
12 . The computer implemented method of claim 11 , wherein the supervised learning process includes generating, based upon the rs-fMRI data of one rs-fMRI dataset, a voxel-wise correlation map of the brain of the individual subject comprising measures of correlation between MR signals at different voxel pairs throughout the brain, and wherein the supervised learning process further includes generating training data for each rs-fMRI dataset that includes supervised training data comprising: the voxel-wise correlation map for voxel locations corresponding to the reference locations, and for supervision, network assignment information comprising a RSN identity to which the voxel location corresponding to the reference location is a member.
13 . The computer implemented method of claim 11 , wherein the reference output comprises a set of RSN reference maps, one for each RSN of the topology of RSNs, wherein each RSN reference map includes the corresponding set of reference locations in the brain representing population-based locations that fall within the RSN.
14 . The computer implemented method of claim 13 , wherein each RSN reference map comprises the reference locations for that RSN overlaid upon an anatomical brain image, and wherein the anatomical brain image upon which the reference locations for that RSN are overlaid is an anatomical brain image of the same individual subject for which output RSN mappings are generated.
15 . A computer system comprising:
a processor; and
a non-transitory computer readable medium having stored thereon program code executable by the processor, the program code causing the processor to:
generate a primary system output using a machine-learning based algorithm created through an algorithm training process that uses training data, wherein the primary system output comprises a mapping of defined functional brain networks within a brain of an individual subject based upon image data pertaining to the brain of the individual subject; and
separately generate an accompanying reference output of reference information informative of the training data used to create the machine-learning based algorithm, wherein the reference output comprises population-averaged location information for the defined functional brain networks.
16 . The computer system of claim 15 , wherein the primary system output comprises a functional connectivity map of resting state networks derived from functional magnetic resonance imaging (fMRI) data, wherein the primary system output is overlaid on a structural magnetic resonance imaging (MRI) image of the brain of the individual subject.
17 . The computer system of claim 15 , wherein the reference output comprises statistical population-averaged location information for the defined functional brain networks.
18 . A non-transitory computer storage medium having stored thereon program code executable by a processor, the program code embodying a method comprising:
generating a primary system output using a machine-learning based algorithm created through an algorithm training process that uses training data, wherein the primary system output comprises a mapping of defined functional brain networks within a brain of an individual subject based upon image data pertaining to the brain of the individual subject; and
separately generating an accompanying reference output of reference information informative of the training data used to create the machine-learning based algorithm, wherein the reference output comprises population-averaged location information for the defined functional brain networks.
19 . The computer storage medium of claim 18 , wherein the primary system output comprises a functional connectivity map of resting state networks derived from functional magnetic resonance imaging (fMRI) data, wherein the primary system output is overlaid on a structural magnetic resonance imaging (MRI) image of the brain of the individual subject.
20 . The computer storage medium of claim 18 , wherein the reference output comprises baseline information for the defined functional brain networks associated with a healthy subject.