AI-based label generating system and methods for use therewith

Patent metadata

Field	Value
Publication number	US-11669965-B2
Application number	US-202117446863-A
Country	US
Kind code	B2
Filing date	Sep 3, 2021
Priority date	Nov 21, 2018
Publication date	Jun 6, 2023
Grant date	Jun 6, 2023

Abstract

A label generating system operates to generate an artificial intelligence model by: training on a training data set that includes the plurality of medical scans with the corresponding global labels; generating testing global probability data by performing an inference function that utilizes the artificial intelligence model on the plurality of medical scans with the corresponding global labels, wherein the testing global probability data indicates a testing set of global probability values corresponding to the set of abnormality classes, and wherein each of the testing set of global probability values indicates a probability that a corresponding one of the set of abnormality classes is present in each of the plurality of medical scans with the corresponding global labels; comparing the testing set of global probability values to a corresponding confidence threshold for each of the plurality of medical scans selected based on the corresponding one of the global labels; generating an updated training data set by correcting ones of the plurality of medical scans having a corresponding one of the testing set of global probability values that compares unfavorably to the corresponding confidence threshold; and retraining the artificial intelligence model based on the updated training set.

First claim

What is claimed is:

1. A label generating system, comprising:
at least one processor; and
a memory that stores operational instructions that, when executed by the at least one processor, cause the label generating system to:
store a plurality of medical scans with corresponding global labels, wherein the global labels each correspond to one of a set of abnormality classes;
generate an artificial intelligence model by:
training on a training data set that includes the plurality of medical scans with the corresponding global labels;
generating testing global probability data by performing an inference function that utilizes the artificial intelligence model on the plurality of medical scans with the corresponding global labels, wherein the testing global probability data indicates a testing set of global probability values corresponding to the set of abnormality classes, and wherein each of the testing set of global probability values indicates a probability that a corresponding one of the set of abnormality classes is present in each of the plurality of medical scans with the corresponding global labels;
comparing the testing set of global probability values to a corresponding confidence threshold for each of the plurality of medical scans selected based on the corresponding one of the global labels;
generating an updated training data set by correcting ones of the plurality of medical scans having a corresponding one of the testing set of global probability values that compares unfavorably to the corresponding confidence threshold; and
retraining the artificial intelligence model based on the updated training set;

receive, via the receiver, a new medical scan;
generate global probability data based on the artificial intelligence model, wherein the global probability data indicates a set of global probability values corresponding to the set of abnormality classes, and wherein each of the set of global probability values indicates a probability that a corresponding one of the set of abnormality classes is present in the new medical scan; and
transmit, via a transmitter, the global probability data to a client device for display via a display device.

2. The label generating system of claim 1 , wherein the global probability data is generated based on probability matrix data generated by performing an inference function that utilizes the artificial intelligence model on the new medical scan, wherein the probability matrix data includes, for each of a set of image patches of the new medical scan, a set of patch probability values corresponding to the set of abnormality classes, and wherein each of the set of patch probability values indicates a probability that a corresponding one of the set of abnormality classes is present in the each of the set of image patches.

3. The label generating system of claim 2 , wherein the operational instructions, when executed by the at least one processor, further cause the label generating system to:
determine a subset of the set of abnormality classes are present in the new medical scan in response to a corresponding subset of the set of global probability values comparing favorably to a corresponding set of probability thresholds.

4. The label generating system of claim 3 , wherein the operational instructions, when executed by the at least one processor, further cause the label generating system to:
transmit, via the transmitter, abnormality data that indicates the subset of the set of abnormality classes to the client device for display via the display device.

5. The label generating system of claim 3 , wherein a size of the subset of the set of abnormality classes determined to be present in the new medical scan is greater than one.

6. The label generating system of claim 3 , wherein at least two of the set of probability thresholds are different values.

7. The label generating system of claim 3 , wherein abnormality correlation data corresponding to the set of abnormality classes is generated in conjunction with training the artificial intelligence model.

8. The label generating system of claim 7 , wherein generating the global probability data utilizes the abnormality correlation data, and wherein an additional abnormality class is added to the subset of the set of abnormality classes in response to the abnormality correlation data indicating that a correlation between the additional abnormality class and at least one of the subset of the set of abnormality classes comparing favorably to a correlation threshold.

9. The label generating system of claim 3 , wherein the operational instructions, when executed by the at least one processor, further cause the label generating system to:
determine a region of interest corresponding to each of the subset of the set of abnormality classes based on the probability matrix data, wherein the abnormality data further includes the region of interest.

10. The label generating system of claim 9 , wherein an interface is displayed by the display device, wherein the interface displays the new medical scan, and wherein the interface further indicates the region of interest in response to receiving the abnormality data.

11. A method, comprising:
storing a plurality of medical scans with corresponding global labels, wherein the global labels each correspond to one of a set of abnormality classes;
generating an artificial intelligence model by:
training on a training data set that includes the plurality of medical scans with the corresponding global labels;
generating testing global probability data by performing an inference function that utilizes the artificial intelligence model on the plurality of medical scans with the corresponding global labels, wherein the testing global probability data indicates a testing set of global probability values corresponding to the set of abnormality classes, and wherein each of the testing set of global probability values indicates a probability that a corresponding one of the set of abnormality classes is present in each of the plurality of medical scans with the corresponding global labels;
comparing the testing set of global probability values to a corresponding confidence threshold for each of the plurality of medical scans selected based on the corresponding one of the global labels;
generating an updated training data set by correcting ones of the plurality of medical scans having a corresponding one of the testing set of global probability values that compares unfavorably to the corresponding confidence threshold; and
retraining the artificial intelligence model based on the updated training set;

receiving, via the receiver, a new medical scan;
generating global probability data based on the artificial intelligence model, wherein the global probability data indicates a set of global probability values corresponding to the set of abnormality classes, and wherein each of the set of global probability values indicates a probability that a corresponding one of the set of abnormality classes is present in the new medical scan; and
transmitting, via a transmitter, the global probability data to a client device for display via a display device.

12. The method of claim 11 , wherein the global probability data is generated based on probability matrix data generated by performing an inference function that utilizes the artificial intelligence model on the new medical scan, wherein the probability matrix data includes, for each of a set of image patches of the new medical scan, a set of patch probability values corresponding to the set of abnormality classes, and wherein each of the set of patch probability values indicates a probability that a corresp

Assignees

• Enlitic Inc

Inventors

• Yao Li
• Lyman Kevin
• Jadhav Ashwin
• Poblenz Eric C
• Upton Anthony

Classifications

• G06T2207/10104

  Positron emission tomography [PET] · CPC title

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06T7/187

  involving region growing; involving region merging; involving connected component labelling · CPC title

• G06V10/25

  Determination of region of interest [ROI] or a volume of interest [VOI] · CPC title

• G06N3/088

  Non-supervised learning, e.g. competitive learning · CPC title