Granted Patent B2

US 12,387,337 · App. 17/714,170 · Granted Aug 12, 2025

Inventors: Kris Siemionow (Chicago, IL); Marek Kraft (Poznan, PL); Dominik Pieczynski (Tulce, PL); Paul Lewicki (Tulsa, OK); Zbigniew Malota (Zabrze, PL); Wojciech Sadowski (Zabrze, PL); Jacek Kania (Rogozno, PL)

Assignee: Kardiolytics Inc.

G06T7/10G06T5/75G06T7/0012G06T7/62G06T15/08G06T2200/04G06T2207/20021G06T2207/20024G06T2207/20084

Abstract

Claims (17)

1. A computer-implemented method for autonomous segmentation of contrast-filled coronary artery vessels, the method comprising:

a) receiving a computed tomography (CT) scan volume representing a three-dimensional (3D) volume of a region of anatomy that includes a pericardium;

b) preprocessing the CT scan volume to output a preprocessed scan volume;

c) dividing the preprocessed scan volume into a first set of subvolumes, wherein the division is performed at a resolution that corresponds to the resolution of the CT scan volume, wherein the first set of subvolumes comprises a plurality of first set subvolumes of the resolution that corresponds to the resolution of the CT scan volume;

d) extracting a region of interest (ROI) corresponding to a heart region outlined by the pericardium, by autonomous segmentation of the heart region by means of a first neural network trained on 3D subvolumes to delineate the pericardium, the extracting step comprising:

feeding each first set subvolume of the first set of subvolumes to the first neural network,

receiving an individual subvolume prediction for each first set subvolume of the first set of subvolumes and

combining the individual subvolume predictions for the first set of subvolumes to generate an ROI mask that delineates the heart region based on the pericardium;

e) combining the preprocessed scan volume with the ROI mask to obtain a masked volume;

f) dividing the masked volume to a second set of subvolumes, wherein the division is performed at a resolution that corresponds to the resolution of the CT scan volume, wherein the second set of subvolumes comprises a plurality of second set subvolumes of the resolution that corresponds to the resolution of the CT scan volume; and

g) performing autonomous coronary vessel segmentation by autonomous segmentation of each second set subvolume of the second set of subvolumes individually by means of a trained segmentation convolutional neural network, trained to process a subvolume to output a mask denoting the coronary vessels, the segmentation being performed at a resolution that corresponds to the resolution of the masked volume, to output a mask denoting the coronary vessels having the resolution of the masked volume.

2. The method according to claim 1 , wherein the step of preprocessing the CT scan includes performing at least one of: windowing, filtering and normalization.

3. The method according to claim 1 , wherein the step of preprocessing the CT scan includes computing a 3D Jerman filter response.

4. The method according to claim 1 , further comprising combining the masks denoting the coronary vessel to a segmented 3D data set representing the shape, location and size of the coronary vessels.

5. A computer-implemented system, comprising:

at least one nontransitory processor-readable storage medium that stores at least one of processor-executable instructions or data; and

at least one processor communicably coupled to the at least one nontransitory processor-readable storage medium, wherein the at least one processor is configured to perform the steps of the method of claim 1 .

Assignments (1)

Continuity (3)

References Cited (14)