System and method for cardiovascular risk prediction and computer readable medium thereof

Patent metadata

Field	Value
Publication number	US-12198340-B2
Application number	US-202217804839-A
Country	US
Kind code	B2
Filing date	May 31, 2022
Priority date	May 31, 2022
Publication date	Jan 14, 2025
Grant date	Jan 14, 2025

Abstract

Provided are a system and a method for cardiovascular risk prediction, where artificial intelligence is utilized to perform segmentation on non-contrast or contrast medical images to identify precise regions of the heart, pericardium, and aorta of a subject, such that the adipose tissue volume and calcium score can be derived from the medical images to assist in cardiovascular risk prediction. Also provided is a computer readable medium for storing a computer executable code to implement the method.

First claim

What is claimed is:

1. A system for cardiovascular risk prediction, comprising:
a segmentation module configured to segment a region from a medical image, wherein the segmentation module is implemented with a machine learning model to segment the region from the medical image, and wherein the machine learning model has a network architecture comprising an encoder part, a decoder part, an attention mechanism, and a variational auto-encoder decoder branch; and
an extraction module configured to extract an analysis result from the region of the medical image, wherein the attention mechanism is configured to highlight salient features passed through skip connections between the encoder part and the decoder part, and wherein the variational auto-encoder decoder branch is configured to reconstruct the medical image based on features from endpoint of the encoder part during training of the machine learning model.

2. The system of claim 1 , wherein the medical image is a non-contrast or contrast computed tomography image.

3. The system of claim 1 , further comprising a model training module configured to provide training to the machine learning model through steps of:
pre-processing a training data into a predetermined consistency;
augmenting the training data by performing random cropping, random spatial flipping and/or random scaling or shifting of intensity on the training data;
training the machine learning model using the training data; and
validating a training result of the machine learning model using a loss function.

4. The system of claim 3 , wherein the training data is generated via labeling the medical image manually and/or with assistance of an auxiliary annotation model.

5. The system of claim 1 , wherein the analysis result comprises an adipose tissue volume of the region, and wherein the extraction module comprises a fat extraction unit configured to quantify the adipose tissue volume within the pericardium in the region through steps of:
calculating a Hounsfield unit value for the pericardium based on an attenuation coefficient under computed tomography;
defining a range of positive and negative standard deviation for the Hounsfield unit value on a basis of noise tolerance; and
determining the adipose tissue volume within the pericardium based on the range.

6. The system of claim 1 , wherein the analysis result comprises a calcium score of the region, and wherein the extraction module comprises a calcium extraction unit configured to quantify the calcium score of a heart or an aorta from the region through steps of:
identifying a calcium region from the region based on a cut point defined by an Agatston score;
capturing the calcium regions as a 3D image;
analyzing the 3D image by a classifier to determine a classification of the calcium region;
assigning a calcium score for the calcium region; and
generating a heatmap to illustrate the calcium region and the calcium score.

7. The system of claim 1 , further comprising a pre-processing module configured to pre-process the medical image into a predetermined consistency through steps of:
resampling a 3D volume of the medical image into a spacing of 2×2×2 mm or any predetermined size; and
normalizing an intensity of the 3D volume into unit standard deviation with zero mean.

8. The system of claim 1 , further comprising an output module configured to present a cardiovascular risk prediction score based on the analysis result.

9. A method for cardiovascular risk prediction, comprising:
segmenting a region from a medical image by a segmentation module, wherein the segmentation module is implemented with a machine learning model to segment the region from the medical image, and wherein the machine learning model has a network architecture comprising an encoder part, a decoder part, an attention mechanism, and a variational auto-encoder decoder branch; and
extracting analysis result from the region of the medical image by an extraction module, wherein the attention mechanism is configured to highlight salient features passed through skip connections between the encoder part and the decoder part, and wherein the variational auto-encoder decoder branch is configured to reconstruct the medical image based on features from endpoint of the encoder part during training of the machine learning model.

10. The method of claim 9 , wherein the medical image is a computed tomography image.

11. The method of claim 9 , further comprising a model training module configured to provide training to the machine learning model through steps of:
pre-processing a training data into a predetermined consistency, wherein the training data is generated via labeling the medical image manually and/or with assistance of an auxiliary annotation model;
augmenting the training data by performing random cropping, random spatial flipping and/or random scaling or shifting of intensity on the training data;
training the machine learning model using the training data; and
validating a training result of the machine learning model using a loss function.

12. The method of claim 9 , wherein the analysis result comprises an adipose tissue volume of the region, and wherein the extraction module comprises a fat extraction unit configured to quantify the adipose tissue volume within the pericardium in the region through steps of:
calculating a Hounsfield unit value for the pericardium based on an attenuation coefficient under computed tomography;
defining a range of positive and negative standard deviation for the Hounsfield unit value on a basis of noise tolerance; and
determining the adipose tissue volume within the pericardium based on the range.

13. The method of claim 9 , wherein the analysis result comprises a calcium score of the region, and wherein the extraction module comprises a calcium extraction unit configured to quantify the calcium score of a heart or an aorta from the region through steps of:
identifying a calcium region from the region based on a cut point defined by an Agatston score;
capturing the calcium regions as a 3D image;
analyzing the 3D image by a classifier to determine a classification of the calcium region;
assigning a calcium score for the calcium region; and
generating a heatmap to illustrate the calcium region and the calcium score.

14. The method of claim 9 , further comprising a pre-processing module configured to pre-process the medical image into a predetermined consistency through steps of:
resampling a 3D volume of the medical image into a spacing of 2×2×2 mm or any predetermined size; and
normalizing an intensity of the 3D volume into unit standard deviation with zero mean.

15. The method of claim 9 , further comprising an output module configured to present a cardiovascular risk prediction score based on the analysis result.

16. A non-transitory computer readable medium, which stores a computer executable code, the computer executable code implementing the method according to claim 1 after being executed.

Assignees

• Univ Nat Taiwan

Inventors

• Wang Tzung-Dau
• Lee Wen-Jeng
• Huang Yu-Cheng
• Tseng Chiu-Wang
• Lee Cheng-Kuang
• Wang Wei-Chung
• Chou Cheng-Ying

Classifications

• G06T7/90

  Determination of colour characteristics · CPC title

• G06T2207/30048

  Heart; Cardiac · CPC title

• G06T2207/10072

  Tomographic images · CPC title

• G16H15/00

  ICT specially adapted for medical reports, e.g. generation or transmission thereof · CPC title

• G16H50/20

  for computer-aided diagnosis, e.g. based on medical expert systems · CPC title