Vessel lumen sub-resolution segmentation
An imaging system ( 100 ) includes a sub-resolution luminal narrowing detector ( 112 ) which detects sub-resolution narrowing of a vessel lumen in an image volume by a centerline profile analysis and computes a sub-resolution determined diameter by modifying an approximated visible lumen diameter with the detected sub-resolution narrowing.
1. An imaging system for detecting narrowing of a vessel lumen and segmenting the vessel lumen, comprising:
one or more processors configured to:
receive image data;
generate a centerline through the vessel lumen from the image data;
estimate a visible lumen diameter by approximating the image data and the centerline;
detect, at a greater precision than a visible resolution of the imaging system, the narrowing of the vessel lumen in the image data using a centerline profile analysis, wherein the centerline profile analysis detects the narrowing of the vessel lumen based on intensity profiles that are lower than a predetermined minimum threshold or greater than a predetermined maximum threshold, the intensity profiles being obtained by sampling voxels in cross-sections along the centerline; and
calculate sub-resolution diameters in the cross-sections along the centerline by modifying the estimated visible lumen diameter with the detected narrowing of the vessel lumen.
2. The imaging system according to claim 1 , wherein the one or more processors are further configured to segment the vessel lumen of the image data into a first portion that represents the vessel lumen and a second portion that represents background, wherein each voxel in the first portion is assigned first and second values, wherein the first value indicates how likely for a voxel to be included in the first portion based on an intensity of a closest centerline voxel used by a trained model, and a second value indicates how likely for the voxel to be included in the first portion based on a spatial distance of the voxel from a closest centerline point and a sub-resolution diameter.
3. The imaging system according to claim 1 , wherein the sub-resolution diameters are calculated based on diameters calculated using a full-width at half maximum intensity of the intensity profiles for each cross-section along the centerline of the vessel lumen.
4. The imaging system according to claim 2 , wherein the trained model models characteristics of intensities of the voxels in the cross-sections along the centerline.
5. The imaging system according to claim 4 , wherein the trained model includes a linear function fitted to a minimum threshold of intensities of the voxels in the cross sections along the centerline.
6. The imaging system according to claim 4 , wherein the trained model includes a linear function fitted to a maximum threshold of intensities of the voxels in the cross sections along the centerline.
7. A method of detecting narrowing of a vessel lumen and segmenting the vessel lumen by an imaging system, comprising:
receiving image data;
generating a centerline through the vessel lumen from the image data;
estimating a visible lumen diameter by approximating the image data and the centerline;
detecting, at a greater precision than a visible resolution of the imaging system, the narrowing of the vessel lumen in the image data using a centerline profile analysis, wherein the centerline profile analysis detects the narrowing of the vessel lumen based on intensity profiles that are lower than a predetermined minimum threshold or greater than a predetermined maximum threshold, the intensity profiles being obtained by sampling voxels in cross-sections along the centerline; and
calculating sub-resolution diameters in the cross-sections along the centerline by modifying the estimated visible lumen diameter with the detected narrowing of the vessel lumen.
8. The method according to claim 7 , further comprising segmenting the vessel lumen of the image data into a first portion that represents the vessel lumen and a second portion that represents background, wherein each voxel in the first portion is assigned first and second values, wherein the first value indicates how likely for a voxel to be included in the first portion based on an intensity of a closest centerline voxel used by a trained model, and a second value indicates how likely for the voxel to be included in the first portion based on a spatial distance of the voxel from a closest centerline point and the sub-resolution diameter.
9. The method according to claim 7 , wherein the sub-resolution diameters are calculated based on diameters calculated using a full-width at half maximum intensity of the intensity profiles for each cross-section along the centerline of the vessel lumen.
10. The method according to claim 7 , wherein the trained model models characteristics of intensities of the voxels in the cross-sections along the centerline.
11. The method according to claim 10 , wherein the trained model includes a linear function fitted to a minimum threshold of intensities of the voxels in the cross sections along the centerline.
12. The method according to claim 10 , wherein the trained model includes a linear function fitted to a maximum threshold of intensities of the voxels in the cross sections along the centerline.