Granted Patent B2

US 12,579,644 · App. 18/446,925 · Granted Mar 17, 2026

Inventors: Sourav Halder (Evanston, IL); Ethan M. Johnson (Evanston, IL); Jun Yamasaki (Evanston, IL); Peter J. Kahrilas (Evanston, IL); Michael Markl (Evanston, IL); John Erik Pandolfino (Evanston, IL); Neelesh A. Patankar (Evanston, IL)

Assignee: Northwestern University

G06T7/0012G06T7/11G06T2207/20081G06T2207/30004

Abstract

Claims (20)

1 . A method for analyzing flow through a tubular organ of a subject, the method comprising:

(a) accessing, with a computer system, medical imaging data acquired from the subject, the medical image data depicting transport of a bolus through a tubular organ of the subject;

(b) generating segmented medical imaging data with the computer system by segmenting the medical imaging data in order to segment the bolus as it is transported through the tubular organ of the subject;

(c) computing, with the computer system from the segmented medical imaging data, spatiotemporal parameters associated with the bolus as it is transported through the tubular organ of the subject;

(d) accessing with the computer system a machine learning model that has been trained to estimate quantitative flow analysis parameter data from spatiotemporal parameters associated with a bolus transport;

(e) generating quantitative flow analysis parameter data by inputting the spatiotemporal parameter data to the machine learning model using the computer system, generating output data as the quantitative flow analysis data; and

(f) presenting the quantitative flow analysis parameter data to a user.

2 . The method of claim 1 , wherein the quantitative flow analysis parameter data comprise at least one of cross-section areas of the tubular organ, fluid velocity of the bolus, or fluid pressure of the bolus.

3 . The method of claim 2 , wherein the quantitative flow analysis parameter data comprise cross-section areas of the tubular organ, fluid velocity of the bolus, and fluid pressure of the bolus.

4 . The method of claim 1 , wherein the spatiotemporal parameter data comprise spatial position along the bolus and time.

5 . The method of claim 1 , wherein generating the segmented medical imaging data comprises inputting the medical imaging data to a trained machine learning algorithm, generating output data as the segmented medical image data, wherein the trained machine learning algorithm has been trained on training data to identify and segment a bolus from background image data.

6 . The method of claim 5 , wherein the trained machine learning algorithm implements an artificial neural network.

7 . The method of claim 6 , wherein the artificial neural network is a convolutional neural network having a 3D U-Net architecture.

8 . The method of claim 1 , wherein the machine learning model comprises a physics-informed neural network (PINN).

9 . The method of claim 8 , wherein the PINN is based on a one-dimensional fluid mechanics-based model.

10 . The method of claim 1 , wherein the tubular organ is an esophagus.

11 . The method of claim 1 , wherein the medical imaging data comprise magnetic resonance images.

12 . The method of claim 1 , wherein the medical imaging data comprise computed tomography (CT) images.

13 . The method of claim 1 , wherein the tubular organ is an esophagus and further comprising computing at least one of an esophageal wall stiffness or an active relaxation from the quantitative flow analysis parameter data.

14 . The method of claim 13 , further comprising generating a report that quantifies esophageal function based on the quantitative flow analysis parameter data and at least one of esophageal wall stiffness or active relaxation.

Assignments (2)

Continuity (2)

References Cited (48)