System and method for segmentation and automated measurement of chronic wound images

Patent metadata

Field	Value
Publication number	US-2016284084-A1
Application number	US-201615078313-A
Country	US
Kind code	A1
Filing date	Mar 23, 2016
Priority date	Mar 23, 2015
Publication date	Sep 29, 2016
Grant date	—

Abstract

Disclosed are systems and methods for automated monitoring of the size, area or boundary of chronic wound images. The disclosure includes use of a probability map that measures the likelihood of wound pixels belonging to granulation, slough or eschar, which can then be segmented using any standard segmentation techniques. Measurement of the wound size, area or boundary occurs automatically and without user input related to outlining, filling in, or making measurement lines over the image on a display.

First claim

What is claimed is:

1 . A method for segmentation and automated measurement of chronic wound images comprising:
obtaining a digital image, wherein at least a portion of the image comprises a wound; automatically determining at least one of a boundary, size, or an area of the wound; and providing the determined at least one of the boundary, the size or the area of the wound, wherein the provided determined at least one of the boundary, the size or the area of the wound is used to monitor wound healing.

2 . The method of claim 1 , wherein the determining at least one of the boundary, the size, or the area of the wound comprises classifying one or more pixels of the digital image as belonging to the wound or not being associated with the wound.

3 . The method of claim 2 , wherein the determining at least one of the boundary, the size, or the area of the wound comprises determining a pixel size for at least the one or more pixels of the digital image that are classified as belonging to the wound.

4 . The method of claim 3 , wherein the pixel size is determined by placing an object of a known size in the digital image proximate to the wound.

5 . The method of claim 4 , wherein the object of known size comprises a label of known size.

6 . The method of claim 5 , wherein determining the pixel size comprises:
detecting the label in the digital image using image analysis algorithms; measuring a number of pixels that span a given detected edge of known size of the label; and determining the pixel size by dividing the known size of the given detected edge by the number of pixels that span the given detected edge of the label.

7 . The method of claim 5 , wherein the label further comprises thereon descriptive information of a patient associated with the wound.

8 . The method of claim 3 , wherein classifying the one or more pixels of the digital image belonging to the wound or not being associated with the wound comprises segmenting the one or more pixels of the digital image as belonging to the wound based on a color of the at least one or more pixels of the digital image.

9 . The method of claim 8 , wherein each of the one or more pixels segmented as belonging to the wound are further identified as granulation, slough or eschar tissue based on a probability of the pixel belonging to a color of a four-dimensional color map.

10 . The method of claim 9 , wherein the probability of the pixel belonging to the color of the four-dimensional color map is computed based on the distance of the image pixels to red, yellow, black and white colors in a modified Hue-Saturation-Value (HSV) color space.

11 . The method of claim 10 , wherein pixels having the highest probability of being red are associated with granulation tissue, pixels having the highest probability of being yellow are associated with slough tissue, and pixels having the highest probability of being black are associated with eschar tissue.

12 . The method of claim 11 , wherein pixels having the highest probability of being white are associated with epibole tissue, skin or an object of a known size in the digital image used to determine pixel size.

13 . The method of claim 9 , wherein the one or more pixels segmented as belonging to the wound are further identified as granulation, slough or eschar tissue based on the probability of the pixel belonging to the color of a four-dimensional color map as determined by a region-growing algorithm or an optimal thresholding algorithm.

14 . The method of claim 2 , further comprising determining the boundary of the wound and reporting the wound boundary as an ordered vector of pixel coordinates.

15 . The method of claim 1 , further comprising displaying the boundary of the wound on the digital image.

16 . The method of claim 15 , wherein the boundary of the wound is displayed as an overlay on the digital image with a color that's distinguishable from both the wound and surrounding tissue.

17 . The method of claim 14 , wherein determining the size and area of a wound comprises
determining a maximum distance between two boundary pixel values; reporting the maximum distance as a length of the wound; determining a perpendicular maximum distance between two boundary pixels, wherein the perpendicular maximum distance is a maximum distance between two boundary pixels such that a straight line drawn between the two boundary pixels that form the perpendicular maximum distance would be perpendicular to a straight line drawn between the two boundary pixels that form the length of the wound; reporting the perpendicular maximum distance as a width of the wound; calculating pixel values of pixels that belong to the wound that are within the boundary of the wound; and reporting the total pixel values within the wound boundary as the area of the wound

18 . The method of claim 17 , wherein the width is reported in centimeters, the width is reported in centimeters, and the area is reported in square centimeters.

19 . The method of claim 1 , wherein at least one expert reviews the determined at least one of the size, boundary or the area of the wound and provides feedback and said feedback is used for updating segmentation and classification parameters used for determining at least one of the size, boundary or the area of the wound.

20 . A method for segmentation and measurement of temporal changes in chronic wound images comprising:
obtaining a first digital image, wherein at least a portion of the first digital image comprises a wound; determining at least one of a first boundary, a first size, or a first area of the wound at a first time by classifying one or more pixels of the first digital image as belonging to the wound or not being associated with the wound; obtaining a second digital image of the at least the portion of the wound, wherein the second digital image is captured at a second time that is after the first digital image was captured; determining at least one of a second boundary, a second size, or a second area of the wound at the second time by classifying one or more pixels of the second digital image as belonging to the wound or not being associated with the wound; comparing the determined at least one of the first boundary, first size or first area of the wound to the determined at least one of the second boundary, second size or the second area of the wound to determine if the at least one of the boundary, size or the area of the wound is changing by getting smaller or getting larger, or if it is staying the same; and providing the results of the comparison, wherein the provided comparison is used to monitor wound healing.

21 . The method of claim 20 , wherein the wound is medically treated in accordance with the determination that the at least one of the size or the area of the wound is getting smaller, getting larger, or staying the same.

22 . The method of claim 21 , further comprising predicting at least one timeline for healing of the wound.

23 . The method of claim 22 , wherein predicting the at least one timeline for healing of the wound considers demographic and medical characteristics of a patient associated with the wound.

24 . The method of claim 23 , further comprising displaying the at least one predicted timeline for healing of the wound.

25 . The method of claim 20 , wherein providing the comparison, wherein the provided comparison is used to monitor wound

Assignees

• Ohio State Innovation Foundation

Inventors

• Gurcan Metin N
• Sen Chandan K
• Gordillo Gayle

Classifications

• G16H30/40

  for processing medical images, e.g. editing · CPC title

• G16H30/20

  for handling medical images, e.g. DICOM, HL7 or PACS · CPC title

• G06F18/2433

  Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection · CPC title

• G06T2207/20076

  Probabilistic image processing · CPC title

• G06T7/62

  of area, perimeter, diameter or volume · CPC title