Systems and methods for topographical characterization of medical image data
AssigneeThe Regents of the University of Michigan
InventorsBenjamin A. Hoff, Craig Galban, Brian D. Ross
Life of this patent
Estimating the term…
Abstract
Computer-implemented methods are used to analyze a medical image to assess the state of the sample region. In at least one embodiment, the method comprises receiving at least one medical image collected previously from an image source, the at least one medical image comprising a plurality of voxels, each characterized by a signal value; classifying the signal value of each voxel as representing one of healthy tissue or diseased tissue based on a threshold value; and analyzing at least one topographical feature of the at least one medical image.
Claims20
A computer-implemented method of analyzing a medical image to assess the state of the sample region, the method comprising:
- receiving a medical image collected previously from an image source, the medical image comprising a plurality of voxels, each characterized by a signal value;
- classifying the signal value of each voxel as representing one of a first disease state or a second disease state by comparing the signal value to a threshold value;
- and
- performing a topographical feature analysis on at least one of a first set of voxels representing the first disease state and a second set of voxels representing the second disease state.
A computer-implemented method of analyzing a sample region of lung tissue to determine the condition of the sample region, the method comprising:
- receiving, using a medical imaging device, a first image data set of the sample region during inspiration, the first image data set comprising a first plurality of voxels each characterized by a signal value;
- receiving, using the medical imaging device, a second image data set of the sample region during expiration, the second image data set comprising a second plurality of voxels each characterized by a signal value;
- deformably registering the first image data set and the second image data set to produce a co-registered image data set that comprises a plurality of co-registered voxels, wherein each of the co-registered voxels includes the signal value of the voxel associated with the first image data set, and the signal value of the voxel associated with the second image data set;
- performing a first threshold analysis on the co-registered voxels to identify a first set of co-registered voxels indicating a first disease state;
- performing a second threshold analysis on the co-registered voxels to identify a second set of co-registered voxels indicating a second disease state;
- and
- performing a topographical feature analysis on at least one of the first set of co-registered voxels and the second set of co-registered voxels.
A computer-based method for analyzing a parametric response map comprising:
- receiving a first set of parametric measurement data for a tissue region, the first set of parametric measurement data comprising a plurality of voxels;
- receiving a subsequent set of parametric measurement data for the tissue region, the subsequent set of parametric measurement data comprising a plurality of voxels;
- deformably registering the subsequent set of parametric measurement data with the first set of parametric response data;
- identifying the voxels within the tissue region, on a voxel-by-voxel basis, by at least one classification based on a change in parametric measurement data, using a defined threshold for each classification, wherein each of the changes is determined by comparing voxels of the subsequent set of parametric measurement data to the voxels of the first set of parametric measurement data;
- and
- performing at least one topographical feature analysis of the voxels for at least one of the classifications.
Dependencies are inferred from the claim text.