Back to results
US 7,123,760 B2Granted
Method and apparatus for removing obstructing structures in CT imaging
AssigneeGeneral Electric Company
InventorsRakesh Mullick, Vidya Pundalik Kamath, Yogisha Mallya
Life of this patent
Estimating the term…
Abstract
A technique is provided for automatically identifying regions of bone or other structural regions within a reconstructed CT volume data set. The technique identifies and labels regions within the data set and computes various statistics for the regions. A rule-based classifier processes the statistics to classify each region. Incidental connections between disparate regions are eliminated. A structure mask, such as a bone mask, is then constructed after exclusion of regions of interest, such as a vascular map. The structure mask may then be used to construct a volume rendering free of the structure, such as bone-free.
Claims76
A method for automatically identifying one or more structural regions in a volume imaging slice, comprising:
- labeling two or more contiguous regions within a volume imaging slice to form a labeled slice, wherein at least one of the regions generally corresponds to a non-vascular region;
- flood-filling the interior of each region of the labeled slice to form a flood-filled slice;
- computing a distance map of each region of the flood-filled slice;
- computing one or more statistics for each region using at least one of the volume imaging slice, the distance map, or the flood-filled slice;
- and
- applying a rule-based classifier to the one or more statistics for each region to classify each region as one of bone, vessel, or indeterminate.
3 more elements
A method for automatically classifying a region, comprising:
- applying a rule-based classifier to a set of statistics derived for a region of an image slice, wherein the rule-based classifier classifies the region as one of bone, vessel, or indeterminate and wherein the set of statistics comprises at least one of an area, a mean intensity, a standard deviation of intensity, a growth factor, a mean intensity associated with the periphery of the region, a mean intensity associated with the interior of the region, a standard deviation of intensity associated with the periphery of the region, a standard deviation of intensity associated with the interior of the region, an intensity covariance, a maximum intensity, a percentage of voxels exceeding a threshold intensity, a mean radius of the region or a standard deviation of the radius of the region.
A method for constructing a three-dimensional bone map, comprising:
- identifying one or more bone-labeled regions in a plurality of image slices;
- performing a connectivity analysis for the one or more bone-labeled regions in each image slice to determine if each bone-labeled region in the image slice is connected to a respective region in one or more proximate slices;
- automatically re-classifying bone-labeled regions which fail the connectivity analysis as non-bone regions;
- and
- constructing a three-dimensional bone map comprising the remaining bone-labeled regions.
2 more elements
A CT imaging system, comprising:
- an X-ray source configured to emit a stream of radiation;
- a detector array configured to detect the stream of radiation;
- a system controller comprising an X-ray controller, a motor controller, and a data acquisition system, wherein the system controller controls the X-ray source and receives signal data from the detector array;
- and
- a computer operably connected to the system controller and to memory circuitry;
- wherein at least one of the system controller or the computer is configured to label two or more contiguous regions within a volume imaging slice, wherein at least one of the regions generally corresponds to a non-vascular region, to flood-fill the interior of each labeled region, to compute a distance map of each flood-filled region, to compute one or more statistics for each region using at least one of the volume imaging slice, the distance map, or the flood-filled region, and to apply a rule-based classifier to the one or more statistics for each region to classify each region as one of bone, vessel, or indeterminate.
3 more elements
A CT imaging system, comprising:
- an X-ray source configured to emit a stream of radiation;
- a detector array configured to detect the stream of radiation;
- a system controller comprising an X-ray controller, a motor controller, and a data acquisition system, wherein the system controller controls the X-ray source and receives signal data from the detector array;
- and
- a computer operably connected to the system controller and to memory circuitry;
- wherein at least one of the system controller or the computer is configured to apply a rule-based classifier to a set of statistics derived for a region of an image slice, wherein the rule-based classifier classifies the region as one of bone, vessel, or indeterminate.
3 more elements
A CT imaging system, comprising:
- an X-ray source configured to emit a stream of radiation;
- a detector array configured to detect the stream of radiation;
- a system controller comprising an X-ray controller, a motor controller, and a data acquisition system, wherein the system controller controls the X-ray source and receives signal data from the detector array;
- and
- a computer operably connected to the system controller and to memory circuitry;
- wherein at least one of the system controller or the computer is configured to identify one or more bone-labeled regions in a plurality of image slices, to perform a connectivity analysis for the one or more bone-labeled regions in each image slice to determine if each bone-labeled region in the image slice is connected to a respective region in one or more proximate slices, to automatically re-classify bone-labeled regions which fail the connectivity analysis as non-bone regions, and to construct a three-dimensional bone map comprising the remaining bone-labeled regions.
3 more elements
A CT imaging system, comprising:
- means for labeling two or more contiguous regions within a volume imaging slice to form a labeled slice, wherein at least one of the regions generally corresponds to a non-vascular region;
- means for flood-filling the interior of each region of the labeled slice to form a flood-filled slice;
- means for computing a distance map of each region of the flood-filled slice;
- means for computing one or more statistics for each region using at least one of the volume imaging slice, the distance map, or the flood-filled slice;
- and
- means for applying a rule-based classifier to the one or more statistics for each region to classify each region as one of bone, vessel, or indeterminate.
3 more elements
A CT imaging system, comprising:
- an X-ray source configured to emit a stream of radiation;
- a detector array configured to detect the stream of radiation;
- a system controller comprising an X-ray controller, a motor controller, and a data acquisition system, wherein the system controller controls the X-ray source and receives signal data from the detector array;
- a computer operably connected to the system controller and to memory circuitry;
- and
- means for applying rule-based classifier to a set of statistics derived for a region of an image slice, wherein the rule-based classifier the region as one of bone, vessel, or inderterminate.
3 more elements
A CT imaging system, comprising:
- means for identifying one or more bone-labeled regions in a plurality of image slices;
- means for performing a connectivity analysis for the one or more bone-labeled regions in each image slice to determine if each bone-labeled region in the image slice is connected to a respective region in one or more proximate slices;
- means for automatically re-classifying bone-labeled regions which fail the connectivity analysis as non-bone regions;
- and
- means for constructing a three-dimensional bone map comprising the remaining bone-labeled regions.
2 more elements
One or more computer readable media comprising routines for automatically identifying one or more structural regions in a volume imaging slice, comprising:
- a routine for labeling two or more contiguous regions within a volume imaging slice to form a labeled slice, wherein at least one of the regions generally corresponds to a non-vascular region;
- a routine for flood-filling the interior of each region of the labeled slice to form a flood-filled slice;
- a routine for computing a distance map of each region of the flood-filled slice;
- a routine for computing one or more statistics for each region using at least one of the volume imaging slice, the distance map, or the flood-filled slice;
- and
- a routine for applying a rule-based classifier to the one or more statistics for each region to classify each region as one of bone, vessel, or indeterminate.
3 more elements
One or more computer readable media comprising routines for automatically classifying a region comprising:
- A routine for applying a rule-based classifier to a set of statistics derived for a region of an image slice, wherein the rule-based classifier classifies the region as one of bone, vessel, or indeterminate and wherein the set of statistics for the region comprises at least one of an area, a mean intensity, a standard deviation of intensity, a growth factor, a mean intensity associated with the periphery of the region, a mean intensity associated with the interior of the region, a standard deviation of intensity associated with the periphery of the region, a standard deviation of intensity associated with the interior of the region, an intensity covariance, a maximum intensity, a percentage of voxels exceeding a threshold intensity, a mean radius of the region or a standard deviation of the radius of the region.
One or more computer readable media comprising routines for constructing a three-dimensional bone map, comprising:
- a routine for identifying one or more bone-labeled regions in a plurality of image slices;
- a routine for performing a connectivity analysis for the one or more bone-labeled regions in each image slice to determine if each bone-labeled region in the image slice is connected to a respective region in one or more proximate slices;
- a routine for automatically re-classifying bone-labeled regions which fail the connectivity analysis as non-bone regions;
- and
- a routine for constructing a three-dimensional bone map comprising the remaining bone-labeled regions.
2 more elements
Dependencies are inferred from the claim text.