Vascular network organization via hough transform (VaNgOGH): a radiomic biomarker for diagnosis and treatment response

A non-transitory computer-readable storage device storing computer-executable instructions that, in response to execution, cause a processor to perform operations, the operations including:

• accessing a radiological image of a region of tissue demonstrating cancerous pathology, where the radiological image includes a plurality of voxels, a voxel having an intensity;
• defining an image scene in a three dimensional (3D) space based on the radiological image, where the image scene includes a plurality of voxels, a voxel having an intensity, where the image 3D space includes an intra-tumoral volume and a set of peri-tumoral sub-volumes that is disjoint from the intra-tumoral volume;
• segmenting and skeletonizing a tumor represented in the intra-tumoral volume;
• segmenting and skeletonizing a vasculature represented in the set of peri-tumoral sub-volumes;
• obtaining a first two-dimensional (2D) vessel network representation in the a Cartesian domain from the segmented and skeletonized tumor and vasculature;
• obtaining a second two-dimensional (2D) vessel network representation in the a spherical domain from the segmented and skeletonized tumor and vasculature;
• computing and aggregating a first set of localized Hough transforms in the a Cartesian space based on the first 2D vessel network representation;
• computing and aggregating a second set of localized Hough transforms on spherical projections within each member of the disjoint set of peri-tumoral sub-volumes, respectively, based on the second 2D vessel network representation;
• computing a set of first order statistics derived from the first set of localized Hough transforms and the second set of localized Hough transforms;
• computing a Vascular Network Organization via Hough transform (VaNgOGH) descriptor based on the set of first order statistics;
• providing the VaNgOGH descriptor to a machine learning classifier configured to distinguish a positive class from a negative class based on the VaNgOGH descriptor;
• and
• receiving a probability that the region of tissue belongs to the positive class or the negative class;
• generating a classification of the region of tissue based on the probability;
• and
• displaying the classification.

An apparatus comprising:

• a processor;
• a memory configured to store a digitized radiological image of a region of interest (ROI) demonstrating cancerous pathology, the digitized radiological image having a plurality of voxels, a voxel having an intensity;
• an input/output (I/O) interface;
• a set of circuits;
• and
• an interface that connects the processor, the memory, the I/O interface, and the set of circuits, the set of circuits comprising:
• an image acquisition circuit configured to:
• access a the digitized radiological image of an the ROI, where the ROI includes a tumor region and a peritumoral region, the digitized radiological image including a a plurality of voxels, a voxel having an intensity, where the digitized radiological image has a plane of acquisition z;
• a segmentation and skeletonization circuit configured to:
• define a tumoral volume by segmenting the tumor region represented in the digitized radiological image, the tumoral volume having a boundary and a centroid;
• define a peritumoral volume based on the boundary of the tumoral volume;
• and
• define a vasculature associated with the tumoral volume, where the vasculature is located within the peritumoral volume;
• a Cartesian feature circuit configured to:
• generate a Cartesian two-dimensional (2D) vessel network representation in the a Cartesian domain based on the vasculature;
• compute a first set of localized Hough transforms based on the Cartesian 2D vessel network representation;
• and
• generate a first aggregated set of peak orientations based on the first set of localized Hough transforms;
• a spherical domain feature circuit configured to:
• divide the peritumoral volume into a set of annular sub-volumes;
• generate, for each annular sub-volume, respectively, a spherical 2D vessel network representation in the a spherical domain based on the vasculature;
• compute a second, different set of localized Hough transforms based on the spherical 2D vessel network representation of each annular sub-volume, respectively;
• and
• generate a second, different aggregated set of peak orientations based on the second set of localized Hough transforms;
• a vascular network organization descriptor circuit configured to:
• generate a vascular network organization descriptor based on the first aggregated set of peak orientations and the second aggregated set of peak orientations;
• a classification circuit configured to:
• compute a probability that the ROI is a member of a positive class or negative class, where the classification circuit is configured to compute the probability based, at least in part, on the vascular network organization descriptor, using a linear discriminant analysis (LDA) machine learning approach;
• and
• generate a classification of the ROI as a member of the positive class or the negative class based on the probability;
• and
• a display circuit configured to display the classification.