Method and system of analyzing medical images

Patent metadata

Field	Value
Publication number	US-11080852-B2
Application number	US-201916544479-A
Country	US
Kind code	B2
Filing date	Aug 19, 2019
Priority date	Aug 19, 2018
Publication date	Aug 3, 2021
Grant date	Aug 3, 2021

Abstract

The present invention seeks to provide a method of analyzing medical image, the method comprises receiving a medical image; applying a model stored in a memory; analyzing the medical image based on the model; determining the medical image including a presence of fracture; and, transmitting an indication indicative of the determination.

First claim

The invention claimed is:

1. A method of analyzing a medical image, wherein the medical image is a whole scale radiograph, the method comprising:
receiving the whole scale radiograph;
applying a model stored in a memory;
analyzing the whole scale radiograph based on the model without identifying the femoral neck in the whole scale radiograph;
determining the whole scale radiograph including a presence of fracture; and,
transmitting an indication indicative of the determination;
wherein applying the model comprising:
receiving training data from a dataset, the training data including a plurality of training images, each training image includes diagnosis data;
developing the model using the training data; and
storing the model in the memory, and
wherein analyzing the whole scale radiograph based on the model comprising augmenting the whole scale radiograph, said augmenting is at least one of:
zooming of the whole scale radiograph;
flipping the whole scale radiograph horizontally;
flipping the whole scale radiograph vertically; and
rotating the whole scale radiograph.

2. The method of claim 1 , wherein applying the model further comprising:
identifying a portion of each training image, wherein the portion includes the diagnosis data; and
developing the model using the training data and the portion identified.

3. The method of claim 1 , wherein developing the model comprising using machine learning technique or deep neural network learning technique.

4. The method of claim 1 , further comprising identifying a lesion site.

5. The method of claim 4 , further comprising generating a heatmap to identify the lesion site.

6. The method of claim 1 , wherein the presence of fracture comprises fracture in a hip or pelvic region.

7. The method of claim 1 , wherein the whole scale radiograph is a frontal pelvic radiograph and 500×500 pixels to 3000×3000 pixels.

8. A system of analyzing a medical image, wherein the medical image is a whole scale radiograph, the system comprising:
a scanner for receiving the whole scale radiograph;
a processor being configured to:
apply a model;
analyze the whole scale radiograph based on the model without identifying the femoral neck in the whole scale radiograph; and
determine the whole scale radiograph comprising a presence of fracture; and
a display for displaying an indication indicative of the determination;
wherein the processor is configured to apply the model comprising
receiving training data from a dataset, the training data including a plurality of training images, each training image includes diagnosis data;
developing the model using the training data; and
storing the model in the memory, and
wherein the processor is configured to analyze the whole scale radiograph based on the model comprising augmenting the whole scale radiograph, said augmenting is at least one of:
zooming of the whole scale radiograph;
flipping the whole scale radiograph horizontally;
flipping the whole scale radiograph vertically; and,
rotating the whole scale radiograph.

9. The system of claim 8 , wherein the processor is configured to retrieve the model further comprising:
identifying a portion of each training image; and
developing the model using the training data and the portion identified.

10. The system of claim 8 , wherein developing the model comprising using machine learning technique or deep neural network learning technique.

11. The system of claim 8 , wherein the processor is further configured to identify a lesion site.

12. The system of claim 11 , wherein the processor is further configured to generate a heatmap to identify the lesion site.

13. The system of claim 8 , wherein the presence of fracture comprise fracture in a hip or pelvic region.

14. The system of claim 13 , wherein the whole scale radiograph is a frontal pelvic radiograph and 500×500 pixels to 3000×3000 pixels.

Assignees

• Chang Gung Memorial Hospital Linkou

Inventors

• Liao Chien-Hung
• Cheng Chi-Tung
• Ho Tsung-Ying
• Lee Tao-Yi
• Chou Ching-Cheng

Classifications

• G06T2207/10116

  X-ray image · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T7/0012 Primary

  Biomedical image inspection · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2207/30008

  Bone · CPC title