Granted Patent B2

US 12,548,658 · App. 18/045,772 · Granted Feb 10, 2026

Inventors: Manish Saggar (Stanford, CA); Samir Chowdhury (Stanford, CA); Caleb Geniesse (Stanford, CA)

Assignee: The Board of Trustees of the Leland Stanford Junior University

G16H30/20A61B5/165A61B5/4064G06F18/2323G06T7/0012G16H50/20G06T2207/30016

Abstract

Claims (78)

1 . A method to identify mental state in an individual, comprising:

obtaining neuroimaging data from an individual;

constructing a shape graph of the neuroimaging data by:

computing pairwise distances of each data point in a data matrix, D, of the neuroimaging data;

constructing a reciprocal k-nearest neighbors (kNN) graph based on the pairwise distances;

determining geodesic distances based on the kNN graph;

landmarking the geodesic distances using farthest point sampling to generate landmarks on the neuroimaging data;

binning data within the neuroimaging data into bins based on the geodesic distances between data points in the neuroimaging data by partitioning the data into r overlapping bins, where partitioning the data comprises: for each landmark, defining

B

i

:=

{

x

:

D

′

(

x

i

,

x

)

≤

4

∈

·

g

1

⁢

0

⁢

0

}

,

 where B i is a bin for a first landmark x i , x is a second landmark, D′ is a transformed matrix produced from data matrix D, where D′ approximates a geometry of temporal trajectories through brain activity space in the neuroimaging data, and g is a gain parameter that controls a level of overlap between each bin; and

clustering the bins; and

identifying a mental state of the individual based on the shape graph.

2 . The method of claim 1 , wherein binning data comprises intrinsic binning or extrinsic binning, wherein intrinsic binning uses landmarks and intrinsic metrics of the data points, and extrinsic binning uses a d-dimensional grid with overlapping cells that fully covers a d-dimensional projection.

3 . The method of claim 1 , wherein the neuroimaging data is obtained from a CT scan, an MRI scan, an fMRI scan, or combinations thereof.

4 . A system for identifying mental state in an individual, comprising:

a processor and a memory, wherein the memory contains instructions that when executed by the processor instructs the processor to:

obtain neuroimaging data from an individual;

construct a shape graph of the neuroimaging data, where to construct the shape graph, the processor is further instructed to:

compute pairwise distances of each data point in a data matrix, D, of the neuroimaging data;

construct a reciprocal k-nearest neighbors (kNN) graph based on the pairwise distances;

determine geodesic distances based on the kNN graph;

landmark the geodesic distances using farthest point sampling to generate landmarks on the neuroimaging data;

bin data within the neuroimaging data based on the geodesic distances between data points in the neuroimaging data by partitioning the data within the neuroimaging data into r overlapping bins, where partitioning the data comprises: for each landmark, defining

B

i

:=

{

x

:

D

′

(

x

i

,

x

)

≤

4

∈

·

g

1

⁢

0

⁢

0

}

,

 where B i is a bin for a first landmark x i , x is a second landmark, D′ is a transformed matrix produced from data matrix D, where D′ approximates a geometry of temporal trajectories through brain activity space in the neuroimaging data, and g is a gain parameter that controls a level of overlap between each bin; and

cluster the bins; and

identify a mental state of the individual based on the shape graph.

5 . The system of claim 4 , wherein binning data comprises intrinsic binning or extrinsic binning, wherein intrinsic binning uses landmarks and intrinsic metrics of the data points, and extrinsic binning uses a d-dimensional grid with overlapping cells that fully covers a d-dimensional projection.

Assignments (2)

Continuity (2)

References Cited (126)