Wound Area Assessment using Mobile Application

Ivan Miguel Pires

1,2

and Nuno M. Garcia

1,3

Instituto de Telecomunicações, University of Beira Interior, Covilhã, Portugal

Altranportugal, Lisbon, Portugal

ECATI, Universidade Lusófona de Humanidades e Tecnologias, Lisbon, Portugal

Keywords: Wound, Detection, Contours, Area, Mobile Device, Mobile Application, Wound Area Assessment,

OpenCV, Java, Threshold, Image Processing, Color Histograms, Gauss Filter, Mobile Platforms, Android,

iOS.

Abstract: This research aims to discover methods for the detection of the area of a wound using mobile devices. These

devices have low memory and low processing capacity and they need the use of low complexity operations

to identify a wound. The calculation of the wounded area consists of three phases, there are: image

acquisition, image processing, surface reconstruction and calculations. This research is related to the use a

mobile device to identify wound contours and area in a captured image. This image can be captured with a

camera in a smartphone and the wound area is calculated based on the distance of the surface area and the

resolution of the image captured. The main study in this research is the image processing in a mobile device,

due to the limitations of these devices. However, the application developed during this research was

developed for desktop, using the OpenCV library that is compatible with the Android platform and Java

desktop technologies. During this research, the developed code written in Java will be easily adapted to the

Android platform. The desktop application developed is available in a free repository for testing.

1 INTRODUCTION

The wound area assessment is an important research

to estimate the evolution of wound healing and the

healthcare professional can change the treatment of

the patients (Ousey and Cook, 2011; Russell, 1999).

Currently this topic is under a lot of research, but a

method to estimate the wound area hasn’t been

discovered. This is a challenging task due to the

complexity of the wound, variable lighting

conditions and time constraints in clinical

laboratories (Loizou et al., 2013). Wounds have

some characteristics related to color and texture, but

these characteristics aren’t the same in all wounds

images and people in the world. Therefore, research

in this area hasn´t finished and future research can

improve the wounds characterization. The research

studies in this area have many purposes, such as

verify the existence of a chronic wound, the

existence of an infected wound, the origin of the

wound and other aspects that classify and

characterize a wound.

This analysis is related to the image capture and

processing of a wound (Lazarus et al., 1994). The

image processing and consequent wound area

estimation has several phases, such as: image

acquisition, image processing, surface reconstruction

and calculations.

The image acquisition includes the process by

witch the user makes use of a camera to capture a

wound image. This process has some problems, such

as ambient lighting, digital camera quality and

resolution, as well as other problems. To minimize

this problem, the next phase is image processing.

This phase includes, the application of filters

(blurring the images to minimize the imperfections

of the images) and the thresholding of the images

(converting the images to black and white, related to

the pixel color intensity). In the next phase, some

authors are doing the surface reconstruction, after

the minimization of possible errors in calculations in

the image-processing phase. In the calculations

phase, the contours of a wound are identified and the

wound area is the area within the contours.

This study is very important (Ousey and Cook,

2011; Russell, 1999), because it allows healthcare

professionals to control their patient’s state of

wound healing and it can improve the treatment of

271

Miguel Pires I. and M. Garcia N..

Wound Area Assessment using Mobile Application.

DOI: 10.5220/0005236502710282

In Proceedings of the International Conference on Biomedical Electronics and Devices (SmartMedDev-2015), pages 271-282

ISBN: 978-989-758-071-0

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

wounds, which this occurs with a correct method

and within a short period of time. In the 22nd annual

meeting of the Wound Healing Society (WHS) held

in 2012, the standards for wound healing procedures

and proposed recommendations for evaluating

optimal wound treatments were set (Loizou et al.,

2013).

The challenge of wound area assessment is very

complex, because the captured images contains

various sources of environmental noise, which are

difficult to control, such as lightning, contrast,

distance to the objects and quality of images. Many

people have done some research, but this research

hasn’t established a consensus in external factor

correction in images. The research studies have a

purpose to create an automatic algorithm,

programmatically implemented and mathematically

validated to calculate the wound area, but this

research depends on the approach of the researcher

to consider some variables or not. Manually, this

research is complex too, because the wound area

hasn’t a defined geometric shape. Thus, in some

cases of research without images, people preferred

to measure the wound perimeter instead of the

wound area.

Mobile technology is currently the most used and

the mobile devices in current days have a camera

with a good quality of image capture. Some devices

have a proximity sensor. The presence of the

proximity sensor allows the estimation of the wound

area with better quality. Using a mobile device

equipped with sensors the value of wound area

calculated is more approximate to the real value.

This purpose can be attained based on various

programming languages and frameworks. To

estimate the wound area there exist various methods

for image processing. Initially, in this paper, it is

analysed the wound area measurement using

MATLAB with Image Processing library. Next, in

this research is analysed the wound area

measurement with Java language and OpenCV

library compatible with the most used mobile

platforms (i.e., Android and iOS operating systems).

This paper was organized in five sections. In

section 2, other researches done by other authors are

presented. In section 3, an overview of the methods

for wound area estimation using images is described.

Section 4 relates the wound images analysis and

wound area measurement to mobile devices,

presenting new methods or sensors utilization with

these devices. Section 5 shows the conclusions of

this research. At the end, the literature references

used for this paper will be presented.

2 RELATED WORK

During the last years, various research authors

addressed the wound area measurement and they

have different perspectives about this topic. In

general, research on this topic consists in the use of

images to estimate the wound area measurement

using a device (e.g., computer, laptop, smartphone

or tablet) for image processing (Krouskop et al.,

2002). Some research focuses in the use of other

sensors, but it wasn’t very investigated and

validated. In relation to the use of images, the

approach to this topic can be divided in some parts,

there are: verification of a existence of a wound in

the image obtained for image processing,

identification of the contours of the wound and

calculate the area between the contours (this

corresponds to the wound area). This process isn’t

linear and objective and depends on the way the

research authors view the problems. Some authors

attempt to identify a wound with a relation to

standard colors of wounds, other authors attempt to

identify by textures and other authors use the join of

standard features of texture and color of wound.

Some research presents the identification of the

existence of infection in a wound region and the type

of wound.

This research area is important for the healthcare

professionals to control, with an easy method, the

wound-healing rate in a patient during the time of

healing (Ousey and Cook, 2011; Russell, 1999).

Thus, it allows a healthcare professional to verify the

wound state and change the treatment used to the

correct treatment related to the evolution of the

wound healing. This is really important in chronic

wounds (Casas et al., 2011; Papazoglou et al., 2010),

because the area of this type of wounds doesn’t

decrease constantly during the wound healing time

and it can increase and decrease inconstantly during

the time of wound healing. The chronic wounds are

more frequently in elderly people, diabetic people or

people with chronic diseases. The health defences of

these people are weakened and their wounds take

longer to heal.

Thus, some research authors have done various

studies related to this topic. The wound-healing rate

is primarily quantified by the change of wound area

and the authors attempt to define a standardized and

objective technique to assess the progress of wound

healing, by means of texture image analysis (Loizou

et al., 2013). The texture of wounds depends of the

location of wounds in people’s body. In the studies

done by the authors of (Loizou et al., 2013) some

tasks where done related to the image processing,

BIODEVICES2015-InternationalConferenceonBiomedicalElectronicsandDevices

272

these were image pre-processing, segmentation,

texture and geometrical analysis together with visual

expert’s to assess the wound healing evaluation.

These authors use a total of 77 digital images

collected in 11 different subjects with foot wounds.

These images were taken every third day, for 21

days, by an inexpensive digital camera under

different lighting conditions (Loizou et al., 2013).

The images collected were intensity normalized, and

wounds were automatically segmented using a

segmentation system based on snakes. With these

experiments, the authors of (Loizou et al., 2013), in

order to identify features that quantify the rate of

wound healing, had extracted 56 different texture

features and 4 different geometrical measures. Thus,

the authors of (Loizou et al., 2013) discovered that

the texture features indicate the progression of

wound healing and some texture features increase

(mean, contrast, roughness and radial sum), while

some other texture features decrease (sum of squares

variance, sum variance, sum average, entropy,

coarseness, EE-laws texture energy measures and

the Hurst coefficients for fractal dimension one and

two analysis) with the progression of wound healing

process. When they compare the different features at

two different time points during wound healing

process, they access the rate of wound healing, but

the comparison of all geometrical measures

extracted from wounds at two different points

doesn’t present important information about the

wound healing. So, these authors create a simple

method that uses some texture features to monitor

the wound healing process, to reduce costs, provide

standardization and improve the treatment quality

for patients and provide a valuable tool in clinical

wound evaluation (Loizou et al., 2013).

Related to the importance of wound healing

control in chronic wounds (Russell, 1999), the

authors of (Papazoglou et al., 2010) present a new

algorithm implemented in MATLAB software

validated in, approximately, 50 animal images and

100 human images. The images for the tests were

captured with a common inexpensive digital camera

and in various lighting conditions. These authors

make a comparison of results in animal images and

human images and compare the manual wound

boundary (obtained in Adobe Photoshop) and the

automatic wound boundary (obtained in MATLAB

software), obtaining very small errors in wound area

measurement. This research depends of the

resolution of the processed wound images, but the

authors of (Papazoglou et al., 2010) proposed and

evaluated a highly accurate algorithm for wound

segmentation which requires a minimal manual

input by using a combination of both red-green-blue

(RGB) and L*a*b color spaces, as well as a

combination of threshold and pixel-based color

comparing segmentation methods.

Other authors have developed systems with

automatic algorithms to measure various parameters

of wounds, such as area, perimeter, width and height

of wounds using images. A example of a system

developed by authors of (Plassmann and Jones,

1998) is the MAVIS system used to automatically

measure the dimensions of skin wounds. In this

system, the method of measurements is based on

color segmentation algorithms and this method is

able to segment images related to healthy skin,

wound tissue and epithelialisation tissue. The

method considers the RGB color planes, hue,

saturation and grey-level intensity. The RGB color

planes were only examined in isolation, showing

that straightforward thresholding of color planes

cannot produce a good segmentation, which

distinguishes between wound and skin tissues. The

wound segmentation with this method is only

partially successful if only the one-dimensional

color histograms were taken into consideration,

while using a 3-dimesional (3D) RGB histogram

space, the color volume clusters may be more widely

separated and a better segmentation result can be

achieved. Some authors, such as (Mekkes and

Westerhof, 1995; Nayak et al., 2009; Wannous et al.,

2008; Wannous et al., 2007), consider they made

some progress using 3D RGB color histogram

clustering technique to assess the wounds healing.

The research of (Mekkes and Westerhof, 1995)

shows that clusters in RGB space for a given tissue

type formed an irregularly shaped 3D cloud, and

therefore simple thresholding along the R, G and B

axis wouldn’t help to segment the image into some

tissue types. The segmentation of wounds in color

images based on the use of the black-yellow-red

classification scheme to evaluate the debridement

activity of wounds have some techniques presented

by other authors (Gammal and Popp, 1995;

Gallagher, 2012). The segmentation of wound

images consists in the image-processing phase.

The method of pre-processing images

corresponds to the first phase of wound area

measurement. The second phase consists in the

identification of contours of wounds, but some

research join the first and second phases and in the

first phase (corresponding to the segmentation,

threshold and other tasks for pre-processing images)

identifies the contours of the wound. The Support

Vector Machine classifiers (SVM) can be used to

perform region segmentation of the wound tissue

WoundAreaAssessmentusingMobileApplication

273

followed by extraction of the contours of the wound

(Kolesnik and Fexa, 2004). In the study presented in

(Kolesnik and Fexa, 2004), the authors used 50 RGB

images, which were manually delineated by experts

as training data, and then tested their method using

23 new RGB wound images. The SVM algorithm

used by (Kolesnik and Fexa, 2004) was able to

correctly classify roughly 94% of the pixels as either

wound or non-wound, compared with the expert’s

manual tracings.

Many research used SVM algorithms (Kolesnik

and Fexa, 2006) with various purposes on this topic.

Authors of (Giger et al., 2008) start using a 3D

model for wound measurements using uncalibrated

vision techniques and a color classification wound

tissues, combining shape and color analysis in a

single tool for real tissue surface measurements. A

database with images of different tissue types in

uncontrolled lighting environments was created

(Giger et al., 2008), applying a correction method to

reduce color shifts. The problem unsolved by all

authors researching in this topic is the difficulty to

control environmental conditions in the experiments

and in the use of the system by other people. Then,

color and texture tissue descriptors are extracted

from tissue regions of the images database, for the

learning stage of an SVM region classifier, and

apply unsupervised color region segmentation on

wound images and classify the tissue regions. The

SVM algorithm used by (Giger et al., 2008) obtains

an overlap score in the result of automatic

segmentation driven classification, (66 % to 88%) of

tissue regions higher than that obtained by

clinicians.

Multidimensional color histograms in SVM

classifiers for automatic extraction of wound region

from an image are used in some research (Kolesnik

and Fexa, 2005). The authors of (Kolesnik and Fexa,

2005) compare the performance of the multi-

dimensional histogram sampling with several

existing techniques for quantization of 3D color

space and this increased the performance of wound

segmentation by about 25%. Many research authors

researched about the creation of systems to measure

size and tissue type of wounds, using images taken

by a digital camera and complex systems were

created. For example, the system constructed by

(Wild et al., 2008) takes about 90 seconds per lesion

and, if the user needs a report with suggestions for

therapy, the system needs 4 minutes.

Instead of the use of SVM classifiers, some

research uses Artificial Neural Networks (ANNs)

algorithms (Acha et al., 2005; Navas et al., 2013;

Song, 2012; Song and Sacan, 2012), such as the

Multi-Layer Perceptron (MLP) and the Radial Basis

Function (RBF) with parameters determined by a

cross-validation approach. There are then applied

with supervised learning in the prediction procedure

for the wound identification, and their results are

compared. The results obtained with ANNs are

satisfactory and this reveals that this method can, in

the future, improve the techniques of wound area

measurement and identification, making it a

promising tool to assist in the field of clinical wound

evaluation.

The automatic systems for wound area

measurements are very useful for telemedicine

systems, because the patient can send a digital

photography to the system and the healthcare

professional is able to check the patient’s wound

state and change the treatments at distance

(Wannous et al., 2010). These systems need to allow

for practical image acquisition conditions, such as

digital camera type, lighting, and viewpoint of

wounds. In general, the telemedicine system consists

in a website or platform for users/patients and

professionals interaction. The systems presented in

(Wannous et al., 2010) were obtaining results of

79.3% between classified tissues and the medical

reference, which compares favorably with the

average score of 69.1% obtained by a single

clinician during the validation tests. The research

authors conclude that these systems are very

important for improving the health treatment in a

world with more people. These systems are very

useful for e-learning systems, because the use of

Web platforms allows to explaining this topic to the

students. The e-learning systems apply the

techniques present in some research and consider the

phases of image processing already present. These

systems improve the teaching-learning relations and

provide a better assessment among students than

traditional methods (Prodan et al., 2010b). All

systems implemented are Web-based (Kim et al.,

2003; Prodan et al., 2009) or software developed in

MATLAB software or in other programming

languages, such as the Java programming language

(Cuautle, 2007; Prodan et al., 2009; Prodan et al.,

2006; Prodan et al., 2010a) and others, using various

frameworks. These systems for classifying the

wounds, in general, comprise four phases, and these

are (Kumar et al., 2013): pre-processing images,

image segmentation, feature extraction and

classification. At the end of this, the system will be

able to measure the wound area and other features of

wounds. The web-based systems (Kim et al., 2003,

Prodan et al., 2009) consist in three-tier layer system

for the user to send a digital image and the image is

BIODEVICES2015-InternationalConferenceonBiomedicalElectronicsandDevices

274

processed in a server, showing the results of wounds

area during the time of wound healing.

Other authors (D, 2006) developed a desktop

application in the VB.NET language for Microsoft

systems. This algorithm consists in various steps,

these are: the digitalization of the outline (perimeter)

of an image from right to left, the digitalization of

the outline (perimeter) of an image from left to right,

the digitalization of the outline (perimeter) of an

image from top to bottom, the digitalization of the

outline (perimeter) of an image from bottom to top

and, finally, the calculation of the area enclosed by

that outline. This system has an objective to use a

database and it consists in a non-invasive, accurate,

consistent, efficient and easy wound measurement

system.

The differential evolution method for estimating

the wound healing is used in some research and

obtains best results. This method uses the K-Nearest

Neighbor (KNN) algorithm to classify the wound

healing. This method includes many phases, there

are (Aslantas and Tunckanat, 2007): read image,

detect entire wound, fill gaps, dilate the image, fill

interior gaps, remove connected objects on border

and smooth the object. This method obtains good

results during the validation tests and it obtains very

low errors.

The planimetric techniques for assessing the

wound area and perimeter with reliability and low

errors are used by authors of (Mayrovitz and

Soontupe, 2009). The automated systems are very

important for a correct wound area measurement,

because, in some cases, the manual measurement is

very difficult.

Nowadays, the use of smartphones equipped

with camera and various sensors is very common the

people and these equipments are practical for wound

measurement (Wannous et al., 2011; Cuautle, 2007;

Foltynski et al., 2013; Hettiarachchi et al., 2013;

Perera and Chakrabarti, 2013; Sikka et al., 2012).

These devices can improve the telemedicine

techniques and treatment at distance (Vivanco et al.,

2011). The mobile applications structure is the same

of desktop applications using frameworks designed

for the wound area measurement. In general, the

captured images are sent to an images database to

improve the reliability of the method and send the

processing tasks to the server that has more capacity

to do complex tasks of image processing. These

applications can apply all research studies already

presented, but the user can access to the wound area

over time in various places. These systems have the

same problem of digital images, because all

variables in the environment are difficult to control.

In fact, the use of these practical devices is low-cost

and it will be able to adapt in hospitals for correct

and practical measurements, because, in recent days,

the research in this topic has improved. Recently,

many authors developed various frameworks to

estimate the wound area using the Java

programming language (Cuautle, 2007), and XML

(eXtended Markup Language) descriptors (Prodan et

al., 2009; Prodan et al., 2006) as reference, for

Android platform or other mobile platforms. The

mobile health improves the treatments in various

areas, such as control of a healing wound and the

patient’s health state in various parameters (Friesen

et al., 2013, Perera and Chakrabarti, 2013). For

desktop applications, mobile applications or web-

based applications, this needs a study and

development of a framework for the programming

language used and an image descriptor (in general, it

is defined in XML) for classifying the images. This

work is very difficult, because images have many

features, such as granularity, texture, color and the

wounds depends of various factors.

In current days, many applications, frameworks

and methods have been developed and are in

advanced research state. The tracings of the Visitrak

method were quick, easy, and inexpensive to

perform and noninvasive for the patient (H et al.,

2009). The Visitrak method considers the foot

curvature and removed the subjectivity associated

with manual square counting. The method was both

valid and repeatable in the measurement of wounds

> 25mm

in size. The Pressure Ulcer Scale for

Healing tool was designed to track pressure ulcer

healing by monitoring wound parameters of length

times width, exudate amount and tissue type (H et

al., 2009). The PSST system was designed to

describe wound healing in pressure ulcers,

consisting of 15 scored (used to assess variables of

wound size and depth, tissue characteristics and

wound exudate, whereas the non-scored items

examined wound location and shape) and two non-

scored items (H et al., 2009). The Sessing Scale is a

seven-stage scale designed to measure progress in

wound healing over time, with each stage describing

wound tissue attributes throughout the wound

healing process (H et al., 2009). The Sussman

Wound Healing Tool is based on an acute model of

wound healing, which describes tissue status and

size throughout the wound healing process (H et al.,

2009). Other mobile application for mobile devices

is MOWA (Mobile Wound Analyzer) (Healthpath,

2011). MOWA differentiates types of tissues found

in pressure ulcer, analyzes photos taken with the on

board camera or uploaded pictures from other

WoundAreaAssessmentusingMobileApplication

275

sources, identifies three types of tissues in the bed of

the lesion (necrotic, fibrinous and granulation) and

calculates the area of the lesion and indicates the

treatment. WoundRight (Technologies, 2013) is an

other mobile wound care application that offers

advanced wound, ostomy, and continence

documentation with the ability to add individual

treatments, and perform detailed assessments.

WoundRight (Technologies, 2013) performs a

powerful accurate and consistent assessments for

wounds and generates progress, area and dimension

charts of wounds and calculates and analyzes wound

data to drive better care and better results. These

applications improve ulcer therapy, telemedicine,

assistance quality, follow up of the cure,

communication, collaboration, home care assistance

and medical/nurse training and education, reducing

use of ineffective products, care times and

hospitalization time.

The next sections present some methods of

wound area estimation using images and the

inclusion of the use of mobile equipments for

executing this analysis of wound area.

3 METHODS OF ESTIMATION

OF WOUND AREA USING

IMAGES

The estimation of wound area has already a lot of

research focus on the use of images for creating

automatic algorithms to identify the wound and

estimate the wound area. For this process, various

research identify a lot of methods to estimate the

wound area with more or less accuracy and

reliability, depending of the research authors. In this

section, a lot of methods of estimation of wound

area using images are described.

The methods of estimation of wound area are

part of software implemented in desktop and mobile

applications and exists various types of methods.

The process of the methods researched can do

automatic or manual procedures to identify the

wound area and measure the wound area.

The basic method of wound area measurement

consists in the phases focused by other authors in

other research studies presented in the section 2. So,

after segmenting images by color or texture or a

mixed segmentation by color and texture and

consequent wound contours detection, the results of

wounds area are in an enclosed contour over the area

of the plain image (Hettiarachchi et al., 2013). In

order to calculate the enclosed pixel area, a flood fill

is used to separate internal and external pixels

(Hettiarachchi et al., 2013). After this process,

histogram method, that consists in separate the

colors by intensity, is used to calculate the number

of pixels within the wound, which is then scaled to

the actual size using the initial calibration triangle

(Hettiarachchi et al., 2013).

A free hand (FH) drawing (Van Poucke et al.,

2010) method is based on the simply holding down

of the mouse button and dragging to draw the

margin of the wound bed. In fact of the difficulty to

calculate the wound area, this method was compared

with other method in (Van Poucke et al., 2010) and

the mean of two wound area values obtained by two

methods is considered the approximated value of the

wound area. The method used in the comparison is a

method based on a closed polygon (CP) (Van

Poucke et al., 2010) graph algorithm, which is a

technique where the margin of the wound bed is

drawn with multiple lines that eventually meet. For

this comparison, the authors of (Van Poucke et al.,

2010) used a set of 2285 images of wounds and any

method that is considered clinically accepted,

because the values are very different, as it is possible

see in figure 1.

Figure 1: Correlation of area measured with free hand and

closed polygon with line of equality (Van Poucke et al.,

2010).

In figure 1 is possible to view that the wound

areas identified differs in the use of free hand

drawing method and a closed polygon graph

algorithm. So, these methods are acceptable to some

healthcare professionals, but not accepted for others,

because these methods obtain large errors in wound

area measurement.

The other computer-aided method for wound

area measurement consists in the use of tools related

to the image processing, such as Adobe Photoshop,

to open a JPG image, defining manually the border

of wound in image and calculate the pixel value of

the region selected as wound and record this to a

Microsoft Excel sheet (Li et al., 2012). This process

BIODEVICES2015-InternationalConferenceonBiomedicalElectronicsandDevices

276

is repeated and some independent research authors

statistically measure the values of wound area based

on the values in the sheet. The traditional

transparency-based wound healing assessment is an

efficient method and clinically accepted by

healthcare professionals (Li et al., 2012). This

method uses a transparency film for marking out the

margins of individual wounds. The outlines regions

as well as the suitable standard area control(s) were

cut off along the margin with an electronic cutter

and weighed by an analytical balance and the

weights of transparency pieces were converted to the

areas one by one by dividing the weight of the

checked region (Y) with the weight of standard unit

(X) and then times the known area of the standard

unit (Z), applying, in the end, statistical corrections

for the obtain a better wound area measurement (Li

et al., 2012).

The chronic wounds have very irregular

dimensions and various methods created aren’t very

accurate. But, for chronic wounds, it is important to

control the wound healing. Thus, a reasonable

approach to determining wound size during a brief

patient encounter would be to document the wound’s

linear measurement – that is, perpendicular linear

dimensions (D, 2006). Normally, in the wound’s

linear measurement, the researchers measure a

wound as a shape, such as a rectangle or an ellipse.

The area of an ellipse is calculated by measuring two

perpendicular diameters, such as maximum diameter

(major diameter) and maximum diameter

perpendicular to the first diameter (minor diameter)

(D, 2006) and this area have an error between 16%

and 40% of the real area (Casas et al., 2011). This

method is simple and relatively cheap, but the

method isn’t precise, because it assumes the wound

area as a simple shape and the wound can have an

irregular form (D, 2006). By other research authors,

this method is called ruler based method (Nemeth et

al., 2010). If the area selected is a rectangle, the area

may be overestimated by 10% to 45% with less

accuracy for smaller wounds (Casas et al., 2011).

An other method consists in the placing of a

transparent film over the wound and tracing the

outline with a permanent marker (Casas et al., 2011,

Nemeth et al., 2010). After this process, the

transparent film is placed on a metric grid and the

area is calculated by counting the number of squared

millimeters contained within the outline (Casas et

al., 2011). This method has a large probability to

have a human error in tracing and the trace is

subjective, depending of the person that does the

trace (Casas et al., 2011). After this, the area can be

measured with a digital photography of the

transparent film with the trace to measure the value

of wound area (Casas et al., 2011). The wound area

can be estimated with a planimeter (Nemeth et al.,

2010).

Other vision-based techniques use either

stereophotogrammetry (SPG) or structured lighting

to obtain wound images (Nemeth et al., 2010). For

stereophotogrammetry, two or more photographs of

the same wound are taken from slightly different

angles and the photographs are used to produce a 3D

model of the wound in a computer (Nemeth et al.,

2010). Then, a computer traces the wound border

and the wound area can be calculated.

Various research authors attempt to create an

algorithm to estimate the wound area using images,

but all methods have influence of external factors,

such as lightning, shadow, and others, in the image

and these cause errors in wound area estimation.

Despite errors, some algorithms are clinical accepted

to help the measurement of wound area.

In the next section, this document will explain

how to use mobile devices, such as smartphones,

tablets and other hand-held devices, for the

estimation of the wound area and the methods,

languages and frameworks, which is possible to use.

4 WOUND AREA ESTIMATION

USING A MOBILE DEVICE

In the last decade, the use of mobile devices has

been increasing, because these equipments

experienced a reduction in price and an increase of

memory and processing capacities (Heggestuen,

2013). Now, one in every 5 people in the world own

a smartphone and one in every 17 people own a

tablet (Heggestuen, 2013). The two platforms

responsible for the largest market share are Android

operating system (owned by Google) and iOS

operating system (owned by Apple) (Bosomworth,

2013). Smartphones usually integrate various

sensors to perform tasks that are related to the use of

the phone in a telecommunications or multimedia-

browsing context, such as camera, accelerometer,

proximity sensor and others. These equipments

allow the user to do complex tasks in movement

without dependency of a desktop computer, because

these equipments can connect to the Internet to send

data for processing tasks that needs a server for

remote processing, storing into a remote database

and visualization of remote data processed.

These devices have a lot of applications available

in the online application stores, to manage and

WoundAreaAssessmentusingMobileApplication

277

processing data and do other tasks. Applications

related to the wound area measurement are low,

because it is very difficult to measure the wound

area and various research don’t have a consensus

about the wound area measurement techniques.

Generally, these equipments have a capacity to take

photographic images of a wound with the embedded

camera in the smartphone. Recently, various

research authors saw the benefits of the use of

smartphones for wound area measurement and a lot

of research was done in this topic. The two mobile

applications available in application online stores

about wound area measurement are MOWA (Mobile

Wound Analyzer) (Healthpath, 2011) and

WoundRight (Technologies, 2013).

MOWA is a non-invasive software that makes

use of a camera of a smartphone to allow the user to

take photos or upload photos of the wound to

analyze, differentiating the types of tissues (necrotic

tissue, fibrinous tissue and granulation tissue) and

calculating the wound area and indicating the

treatment. This application does some tasks

automatically and other tasks are manual (needs

human interaction). The manual tasks are taking a

photo, designing the mask, setting parameters and

sending a JPG and PDF file via e-mail. The

automatic tasks are analyzing tissues, calculating the

wound area, defining directions/suggestions in

treatment and creating an analysis report in PDF.

MOWA is registered as a medical device, is fast

(analysis process takes less than 3 minutes), is easy

to use, improves ulcer therapy, improves the quality

of assistance, helps to identify and measures the

ulcer tissues, defines the priorities in treatment,

suggests the therapeutic treatment, improves the

follow up of the cure, automates the clinical

documentation, improves the communication and

collaboration, helps medical/nurse training and

education, facilitates sector study investigations,

eases effectiveness monitoring of new products,

reduces the use of ineffective products, reduces care

times, reduces hospitalization time, supports home

care assistance and supports telemedicine. Figure 2

shows the screen of application, calculating the

wound area of a digital image uploaded to the

application and using the automatic method of

measurement. MOWA application is only for iPhone

and it is paid.

WoundRight is a software application that offers

advanced wound, ostomy, and continence

documentation with the ability to add individual

treatments, performs detailed assessments and brings

convenient and immediate wound care to the patient.

This application takes the tablet anywhere and

collects your data with or without the Internet,

performs powerful, accurate and consistent

assessments for wounds, generated progress, area

and dimension charts, shares patients data securely

with their affiliated accounts, tracks vitals, medical

conditions, and report on open or closed wounds,

calculates and analyzes wound data to drive better

care and better results, reduces redundancy, keeps

compliant, improves revenues and decrease wound

care costs by lowering re-hospitalization rates and

increasing referral rates and decreases maintenance

costs. WoundRight application is only for tablets, is

supported for Android and iOS operating systems

and it is free.

Figure 2: Screen of MOWA application calculating the

wound area (Healthpath, 2011).

Many other research studies exist about the

wound area measurement related to mobile devices.

These authors define algorithms implemented in

various programming languages and use various

frameworks. The Android development should be

done in Java programming language and in this

programming language exists a lot of studies,

frameworks and libraries available for wound image

processing to measure the wound area. The iOS

development should be done in Objective-C

programming language and in this programming

language exists a minor number of studies,

frameworks and libraries available for wound image

processing to measure the wound area.

The study presented in (Hettiarachchi et al.,

2013) consists in an application, implemented for

the Android operating system, to provide a practical

fast and non-invasive technique to monitor the

wound healing process. The process starts with the

pre-processing of an image captured with camera of

the mobile device. In pre-processing phase, the

active contour is predefined and centered on the

wound, cropping is utilized for centering the wound.

In this phase, the unnecessary artifacts such as

clothing, limb borders and backgrounds are removed

and users demarcate the crop section by marking the

diagonal points of a rectangle on the device’s touch

BIODEVICES2015-InternationalConferenceonBiomedicalElectronicsandDevices

278

screen. Next, all images are resized to 500 pixels,

but maintain the original ratio. Finally, in this phase,

the saturation plane of the HSV color model is

extracted as the base for the active contour algorithm

since it displayed the best contrast between infected

and normal skin and the image is further smoothed

using a Gaussian filter of 31x31 dimensions with

σ=5 in order to remove artifacts that may have

otherwise attracted the active contour erroneously.

The next phase corresponds to the segmentation of

the image. The segmentation is based on active

contour models, which identifies the wound border

irrespective of coloration and shape, and the user,

providing higher control and accuracy, can modify

the segmentation. Finally, the wound area

measurement is further normalized to remove effects

of camera distance and angle and the area

corresponds to the area between the contours. The

results obtained in this application have an accuracy

level of 90%.

The mobile applications for estimating the

wound area have a lot of research that explain the

benefits of these applications (Perera and

Chakrabarti, 2013; Friesen et al., 2013; Casas et al.,

2011). About the method analyzed in the section 3,

ruler measures and transparency tracings were used

by authors of (Nemeth et al., 2010) for measuring

wound size often have low accuracy and reliability

and they created a new method used in a wound

measurement device (WMD) designed by them. The

device designed in (Nemeth et al., 2010) allows the

wound area measurement with to assess for accuracy

over distance from the wound surface, as well as

camera angle skew, using the inter-rater and a intra-

rater methods. Both intra-rater and inter-rater

reliability proved to be significantly higher than

conventional methods, such as ruler measures and

transparency tracings and these methods have an

average 2.65% error rate in the accuracy

measurement based on two black and white shapes

with known areas. The two methods that used a

rectangular approximation of area had positive bias

meaning that they typically over-estimated the area,

whereas the two techniques that traced the borders

under-estimated the areas (Nemeth et al., 2010).

Usually, the process of wound area measurement

using a mobile device consists in three main phases,

these are (Andrade et al., 1999, Gonzalez and

Woods, 1992; Zaffari, 2006): image pre-processing,

image segmentation and wound area measurement.

In the image pre-processing, to obtain better results

of the contours, a filter for remove the noise is

applied, such as the Gaussian Blur. After applying

the filter, the adaptive threshold should be done to

convert the image to black and white colors. Next,

the dilate operation is applied to fill the cracks. After

these three tasks, closed contours can be achieved.

For wound area measurement exists a framework,

developed for all platforms in the market to do the

image processing with easy methods. This

framework is named OpenCV (Bradski and Kaehler,

2008; Marengoni and Stringhini, 2009) and it allows

to do the tasks referring to pre-processing image

phase. In the image segmentation phase, this

framework has methods to identify the contours of a

wound, using various algorithms, such as Canny

Edge Detection or others. And for the end phase, this

framework has a method to calculate the area

between the contours. So, the use of OpenCV

framework is easy and facilitates the tasks for

wound area measurement. This is compatible with

Android applications (developed in Java) and iOS

applications (developed in Objective-C). Other

frameworks are developed in the Java programming

language (Cuautle, 2007; Prodan et al., 2006) and

can be adapted to mobile applications for Android

operating system.

Most of all methods researched for desktop

applications can be adapted to mobile device

applications (Friesen et al., 2013) and have

possibility to estimate the wound area, processing

the image on the device, after capturing the image

with the embedded camera, or sending to a server,

via Internet, and the image processing will be

performed remotely (Foltynski et al., 2013; Sikka et

al., 2012; Vivanco et al., 2011). The user can see the

results on the device, such as AreaMe (Foltynski et

al., 2013), that the results was compared with the

results of Visitrak and SilhouetteMobile (Wannous

et al., 2011) systems.

The use of mobile technology increases the ease

and accuracy in monitoring a wound treatment. This

is especially important in chronic wounds, because

they require greater control and adaptation of

treatment. This is a topic that is in constant

investigation and evolution to improve the existent

methods.

At the end of this research, a Java Desktop

application was available at a free repository and the

source code is available for future research studies.

5 CONCLUSIONS

In conclusion, the wound area assessment is a good

topic for research, because in the last years there

have been some improvements, but the algorithms

developed in some cases have big errors. The

WoundAreaAssessmentusingMobileApplication

279

existence of a lot of research indicates that this

research in this topic isn’t ended and is a very

interesting research.

The research of the wound area assessment can

be done manually (low precision) or automatically

(in the last years has a evolution in this methods). To

manually process for wound area measure, the

healthcare processional use metrics to measure the

wound area, which commonly is measured as a

geometric form, such as a rectangle. For automatic

process, the healthcare professional uses software

for image processing that identifies a wound based

on color and/or texture of wound images. The

wound image characteristics are stored in a database

for future comparisons. This process is done in three

phases, these are: image pre-processing,

segmentation and wound area measurement.

In the image pre-processing phase is identified

the existence of a wound in the image, by color

histograms or texture comparisons and the image

needs to be applied a low-pass filter, threshold and

dilatation of image to obtain a closed contour. To

verify the existence of a wound in the image a K-NN

or a SVM algorithm is applied for the verification.

In the segmentation phase, the contours of the

wound image are identified. For identify the

contours various methods exist, for example using a

geometric shape, discarding some parts out of the

wound or identify the pixels by the color.

For the wound area measurement phase, the

result obtained is the area between the contours. This

research is very important especially for the area of

chronic wounds, because these wounds need to be

monitored during the healing time, because this area

doesn’t decrease constantly during the healing time.

The use of mobile devices allows the wound area

measurement with better precision, because is good

to identify the camera distance of the wound, so it

allows to estimate the real wound area anywhere in

movement, using a proximity sensor of the mobile

device. For the process to estimate the wound area in

a mobile device, various frameworks exist in various

programming languages for help to the development

of the applications, such as OpenCV framework and

others.

This research is very difficult, because is not

possible to control some environmental variables,

such as lightning, noise and quality of camera, but

some algorithms implemented are clinically

accepted to help the healthcare professionals in

telemedicine. The wound image processing in

mobile device can be done with processing image in

the application or send the image by Internet to a

server and receive the results data in the smartphone.

As future work, this research topic needs to be

continuing improve for these applications can be

used commonly in a hospital to improve the

treatments of the patients. The use of automatic

systems has advantages and disadvantages, but

normally the use of automatic systems is more

precise than manual measurements.

ACKNOWLEDGMENTS

This work was supported by FCT projectPEst-

OE/EEI/L A0008/2013 (Este trabalho foi suportado

pelo projecto FCT PEst-OE/EEI/LA0008/2013).

The authors would also like to acknowledge the

contribution of the COST Action IC1303 –

AAPELE – Architectures, Algorithms and Protocols

for Enhanced Living Environments.

REFERENCES

Acha, B., Serrano, C., Acha, J. I. & Roa, L. M. 2005.

Segmentation And Classification Of Burn Images By

Color And Texture Information. J Biomed Opt, 10,

034014.

Andrade, E. Processamento Digital De Imagens - Pdi.

Centro De Estudos – FundaçÃo São Lucas.

Aslantas, V. & Tunckanat, M. 2007. Differential

Evolution Algorithm For Segmentation Of Wound

Images. 1-5.

Bosomworth, D. 2013. Mobile Marketing Statistics 2013

[Online]. Available:

Http://Www.Smartinsights.Com/Mobile-

Marketing/Mobile-Marketing-Analytics/Mobile-

Marketing-Statistics/ [Accessed January 12th 2014].

Bradski, G. & Kaehler, A. 2008. Learning Opencv, United

States Of America, O'reilly.

Casas, L., Castaneda, B. & Treuillet, S. 2011. Imaging

Technologies Applied To Chronic Wounds: A Survey.

4th International Symposium On Applied Sciences In

Biomedical And Communication Technologies.

Barcelona, Spain.

Cuautle, N. F. 2007. System For Face Detection On A

Mobile Phone Using Java Technology. Master Thesis,

University Of Oslo.

D, L. 2006. Design And Implementation For Wound

Measurement Application. Primary Intention 2006,

14, 56-58, 60-63, 66.

Filho, O. M. & Neto, H. V. 1999. Processamento Digital

De Imagens, Rio De Janeiro, Brasport.

Foltynski, P., Ladyzynski, P. & Wojcicki, J. M. 2013. A

New Smartphone-Based Method For Wound Area

Measurement. Artif Organs.

Friesen, M. R., Hamel, C. & Mcleod, R. D. 2013. A

Mhealth Application For Chronic Wound Care:

Findings Of A User Trial. Int J Environ Res Public

BIODEVICES2015-InternationalConferenceonBiomedicalElectronicsandDevices

280

Health, 10, 6199-214.

Gallagher, B. A. 2012. Wound Bed Assessment Using

Calibrated Images And Representation In Openehr.

Master Of Science In Health Informatics, University

Of Dublin.

Gammal, S. & Popp, R. 1995. A Color Image Analysis

System (Cd-Cwa) To Quantify Wound Healing Of

Ulcers. Skin Resear Techn, 1.

Giger, M. L., Wannous, H., Lucas, Y., Treuillet, S. &

Karssemeijer, N. 2008. Efficient Svm Classifier Based

On Color And Texture Region Features For Wound

Tissue Images. 6915, 69152t-69152t-10.

Gonzalez, R. C. & Woods, R. E. 1992. Processamento De

Imagens Digitais, Brasil, Editora Edgar Bucher Ldta.

H, P., M, M., J, T., P, P., S, S. & Ji, S. 2009. Assessment

Of Wound Healing: Validity, Reliability And

Sensitivity Of Available Instruments. Wound Practice

And Research, 17, 208-217.

Healthpath. 2011. Mowa - Mobile Wound Analyzer For

Iphone And Ipad - Healthpath [Online]. Available:

Http://Www.Healthpath.It/Imowa.Html [Accessed

January 12th 2014].

Heggestuen, J. 2013. One In Every 5 People In The World

Own A Smartphone, One In Every 17 Own A Tablet

[Chart] [Online]. Available:

Http://Www.Businessinsider.Com/Smartphone-And-

Tablet-Penetration-2013-10 [Accessed January 12th

2014].

Hettiarachchi, N. D. J., Mahindaratne, R. B. H., Mendis,

G. D. C., Nanayakkara, H. T. & Nanayakkara, N. D.

2013. Mobile Based Wound Measurement. 2013 Ieee

Point-Of-Care Healthcare Technologies (Pht), 298-

301.

Kim, H. M., Lowery, J. C., Hamill, J. B. & Wilkins, E. G.

2003. Accuracy Of A Web-Based System For

Monitoring Chronic Wounds. Telemed J E Health, 9,

129-40.

Kolesnik, M. & Fexa, A. 2004. Segmentation Of Wounds

In The Combined Color-Texture Feature Space.

Medical Imaging 2004: Image Processing, Pts 1-3,

5370, 549-556.

Kolesnik, M. & Fexa, A. 2005. Multi-Dimensional Color

Histograms For Segmentation Of Wounds In Images.

Image Analysis And Recognition, 3656, 1014-1022.

Kolesnik, M. & Fexa, A. How Robust Is The Svm Wound

Segmentation? In: Ieee, Ed. Signal Processing

Symposium, 2006 Rejkjavik. Ieee, 50-53.

Krouskop, T. A., Baker, R. & Wilson, M. S. 2002. A

Noncontact Wound Measurement System. J Rehabil

Res Dev, 39, 337-45.

Kumar, K. S., Asokan, L., M.P, P. & Reddy, B. E. 2013.

Assessment Of The Wound-Healing Process By

Accurate Single View Issue Classification And Depth

Estimation For Telemedicine. International Journal

Of Engineering And Advanced Technology (Ijeat), 2,

467-475.

Lazarus, G. S., Cooper, D. M., Knighton, D. R., Margolis,

D. J., Pecoraro, R. E., Rodeheaver, G. & Robson, M.

C. 1994. Definitions And Guidelines For Assessment

Of Wounds And Evaluation Of Healing. Archives Of

Dermatology, 130, 489-493.

Li, P. N., Li, H., Wu, M. L., Wang, S. Y., Kong, Q. Y.,

Zhang, Z., Sun, Y., Liu, J. & Lv, D. C. 2012. A Cost-

Effective Transparency-Based Digital Imaging For

Efficient And Accurate Wound Area Measurement.

Plos One, 7, E38069.

Loizou, C. P., Kasparis, T. & Polyviou, M. 2013.

Evaluation Of Wound Healing Process Based On

Texture Image Analysis. Journal Of Biomedical

Graphics And Computing, 3.

Marengoni, M. C. & Stringhini, D. 2009. Tutorial:

IntroduçÃo À Visão Computacional Usando Opencv.

Rita, Xvi.

Mayrovitz, H. N. & Soontupe, L. B. 2009. Wound Areas

By Computerized Planimetry Of Digital Images:

Accuracy And Reliability. Adv Skin Wound Care, 22,

222-9.

Mekkes, J. R. & Westerhof, W. 1995. Image Processing In

The Study Of Wound Healing. Clinics In

Dermatology, 13, 401-407.

Navas, M., Luque-Baena, R. M., Morente, L., Coronado,

D., Rodriguez, R. & Veredas, F. J. 2013. Computer-

Aided Diagnosis In Wound Images With Neural

Networks. Advances In Computational Intelligence, Pt

Ii, 7903, 439-448.

Nayak, R., Kumar, P. & Galigekere, R. R. 2009. Towards

A Comprehensive Assessment Of Wound-

Composition Using Color-Image Processing. 2009

16th Ieee International Conference On Image

Processing, Vols 1-6, 4133-4136.

Nemeth, M., Sprigle, S. & Gajjala, A. 2010. Clinical

Usability Of A Wound Measurement Device. 36th

Annual American Spinal Injury Conference. Nashville.

Ousey, K. & Cook, L. 2011. Understanding The

Importance Of Holistic Wound Assessment. Practice

Nursing, 22, 308-314.

Papazoglou, E. S., Zubkov, L., Mao, X., Neidrauer, M.,

Rannou, N. & Weingarten, M. S. 2010. Image

Analysis Of Chronic Wounds For Determining The

Surface Area. Wound Repair Regen,

18, 349-58.

Perera, C. & Chakrabarti, R. 2013. The Utility Of Mhealth

In Medical Imaging. Journal Of Mobile Technology In

Medicine, 2, 4-6.

Plassmann, P. & Jones, T. D. 1998. Mavis: A Non-

Invasive Instrument To Measure Area And Volume Of

Wounds. Measurement Of Area And Volume

Instrument System. Med Eng Phys, 20, 332-8.

Prodan, A., Campean, R., Revnic, C. & Prodan, R. 2009.

Strategies For Wound Image Understanding. 2009

Ieee/Acs International Conference On Computer

Systems And Applications, Vols 1 And 2, 1018-1024.

Prodan, A., Campean, R., Rusu, M., Revnic, C., Mitrea, P.

& Prodan, R. 2010a. Toward A Model For Wound

Healing Simulation. 115-120.

Prodan, A., Rusu, M., Campean, R. & Prodan, R. 2006. A

Java Framework For Analysing And Processing

Wound Images For Medical Education. 20th European

Conference On Modelling And Simulation Ecms 2006,

421-426.

Prodan, A., Rusu, M., Revnic, C., Campean, R., Mitrea, P.

WoundAreaAssessmentusingMobileApplication

281

& Prodan, R. 2010b. Intelligent E-Tools For Wound

Image Understanding And Evaluation. 8-13.

Russell, L. 1999. The Importance Of Wound

Documentation And Classification. Br J Nurs, 8,

1342-3, 1346, 1348 Passim.

Sikka, N., Carlin, K. N., Pines, J., Pirri, M., Strauss, R. &

Rahimi, F. 2012. The Use Of Mobile Phones For

Acute Wound Care: Attitudes And Opinions Of

Emergency Department Patients. J Health Commun,

17 Suppl 1, 37-42; Quiz 42-3.

Song, B. 2012. An Automated Wound Identification

System Based On Image Segmentation And Artificial

Neural Networks. Master Of Science In Biomedical

Engineering, Faculty Of Drexel University.

Song, B. & Sacan, A. 2012. Automated Wound

Identification System Based On Image Segmentation

And Artificial Neural Networks. 1-4.

Technologies, W. 2013. Woundright - Mobile Wound Care

Software [Online]. Woundright Technologies.

Available:

Http://Www.Woundrightapp.Com/Woundright.Html

[Accessed January 12th 2014].

Van Poucke, S., Nelissen, R., Jorens, P. & Vander

Haeghen, Y. 2010. Comparative Analysis Of Two

Methods For Wound Bed Area Measurement. Int

Wound J, 7, 366-77.

Vivanco, J., Neighbour, R., Hamel, C., Mukhi, S., Friesen,

M. R. & Mcleod, R. D. 2011. A Smartphone

Application For Remote Wound Treatment And

Documentation Compliance. In: Ieee (Ed.) Resna.

Toronto, Canada: Ieee.

Wannous, H., Lucas, Y. & Treuillet, S. 2011. Enhanced

Assessment Of The Wound-Healing Process By

Accurate Multiview Tissue Classification. Ieee Trans

Med Imaging, 30, 315-26.

Wannous, H., Lucas, Y., Treuillet, S. & Albouy, B. 2008.

Mapping Classification Results On 3d Model: A

Solution For Measuring The Real Areas Covered By

Skin Wound Tissues. 2008 3rd International

Conference On Information And Communication

Technologies: From Theory To Applications, Vols 1-5,

930-935.

Wannous, H., Treuillet, S. & Lucas, Y. 2007. Supervised

Tissue Classification From Color Images For A

Complete Wound Assessment Tool. Conf Proc Ieee

Eng Med Biol Soc, 2007, 6032-5.

Wannous, H., Treuillet, S. & Lucas, Y. 2010. Robust

Tissue Classification For Reproducible Wound

Assessment In Telemedicine Environments. Journal

Of Electronic Imaging, 19.

Wild, T., Prinz, M., Fortner, N., Krois, W., Sahora, K.,

Stremitzer, S. & Hoelzenbein, T. 2008. Digital

Measurement And Analysis Of Wounds Based On

Colour Segmentation. European Surgery, 40, 5-10.

Zaffari, C. A. 2006. Visualização E Processamento

Digital De Imagens Médicas. Master Of Science,

Universidade Católica Do Rio Grande Do Sul.

BIODEVICES2015-InternationalConferenceonBiomedicalElectronicsandDevices

282