Computerized segmentation and measurement of chronic wound images

Highlights

• We proposed a method to segment chronic wounds based on 4D probability map.

• The algorithm characterizes wounds into granulation, slough and eschar tissues.

• Experiment on 80 wound images gives an average accuracy of 75.1%.

• We also analyze the agreement between experts on wound and tissue boundaries.

• For wound boundary, mean agreement between experts is between 67.4% and 84.3%.

Abstract

An estimated 6.5 million patients in the United States are affected by chronic wounds, with more than US$25 billion and countless hours spent annually for all aspects of chronic wound care. There is a need for an intelligent software tool to analyze wound images, characterize wound tissue composition, measure wound size, and monitor changes in wound in between visits. Performed manually, this process is very time-consuming and subject to intra- and inter-reader variability. In this work, our objective is to develop methods to segment, measure and characterize clinically presented chronic wounds from photographic images. The first step of our method is to generate a Red-Yellow-Black-White (RYKW) probability map, which then guides the segmentation process using either optimal thresholding or region growing. The red, yellow and black probability maps are designed to handle the granulation, slough and eschar tissues, respectively; while the white probability map is to detect the white label card for measurement calibration purposes. The innovative aspects of this work include defining a four-dimensional probability map specific to wound characteristics, a computationally efficient method to segment wound images utilizing the probability map, and auto-calibration of wound measurements using the content of the image. These methods were applied to 80 wound images, captured in a clinical setting at the Ohio State University Comprehensive Wound Center, with the ground truth independently generated by the consensus of at least two clinicians. While the mean inter-reader agreement between the readers varied between 67.4% and 84.3%, the computer achieved an average accuracy of 75.1%.

Introduction

Computer-aided measurement of the size and characteristics of chronic wounds is a novel approach to standardizing the accuracy of chronic wound assessment. A chronic wound, as defined by Centers for Medicare and Medicaid Services, is a wound that has not healed in 30 days. An estimated 6.5 million patients in the United States are affected by chronic wounds, and it is claimed that an excess of US$25 billion is spent annually on treatment of chronic wounds. The burden is growing rapidly due to increasing health care costs, an aging population and a sharp rise in the incidence of diabetes and obesity worldwide [1]. As such, there is a need for a timely and accurate method to document the size and evolving nature of chronic wounds in both the inpatient and outpatient settings. Such an application can potentially reduce clinicians׳ workload considerably; make the treatment and care more consistent and accurate; increase the quality of documentation in the medical record and enable clinicians to achieve quality benchmarks for wound care as determined by the Center for Medicare Services.

The current state of the art approach in measuring wound size using digital images, known as digital planimetry, requires the clinician to identify wound borders and wound tissue type within the image. This is a time-intensive process and is a barrier to achieving clinical quality benchmarks. Our group is developing image analysis tools that will enable the computer to perform this analysis rather than requiring user input. Developing an accurate method of measuring wound size and tissue characteristics serially over time will yield clinically meaningful information in relation to the progression or improvement of the wound. The focus of the work reported in this paper is the segmentation of the wounds.

A wound exhibits a complex structure and may contain many types of tissue such as granulation, slough, eschar, epithelialization, bone, tendon and blood vessels, each with different color and texture characteristics. In this paper, we proposed a novel probability map that measures the likelihood of wound pixels belonging to granulation, slough or eschar (see Fig. 1), which can then be segmented using any standard segmentation techniques. In this work, we focus on the granulation, slough and eschar tissues as these are the three most commonly seen tissues in wounds. A preliminary version of this work has been reported in [2]. This paper extended the previous work significantly with an extensive literature review, more elaborate explanation of the proposed method, employing two segmentation techniques to show that the probability map is adaptable to many different techniques, comparison with other existing method, comprehensive analyses on inter-reader variability between clinicians, a much bigger dataset used for performance evaluation (which was divided into three sets for analysis purpose), as well as more elaborate discussions on the results.

The paper is organized as follows: Section 2 presents the review of the literature on wound image analysis. In Section 3, we present our proposed probability map approach to wound segmentation and integrate it with two different segmentation techniques. 4 Experimental setup, 5 Experimental results and discussion discusses the experimental setup, results and discussion. Finally, Section 6 concludes the paper and describes future work.