Hindawi
Computational and Mathematical Methods in Medicine
Volume 2021, Article ID 4208254, 9 pages
https://doi.org/10.1155/2021/4208254
Research Article
Image Processing for mHealth-Based Approach to Detect the
Local Tissue Inflammation in Cutaneous Leishmaniasis: A
Proof of Concept Study
Hermali Silva ,1 Kalaivani Chellappan ,2 and Nadira Karunaweera 1
1
Department of Parasitology, Faculty of Medicine, University of Colombo, Colombo, Sri Lanka
2
Department of Electrical, Electronic & Systems Engineering, Faculty of Engineering & Built Environment, National University
of Malaysia, Bangi, Selangor, Malaysia
Correspondence should be addressed to Nadira Karunaweera; nadira@parasit.cmb.ac.lk
Received 28 September 2021; Accepted 9 November 2021; Published 27 November 2021
Academic Editor: Luminita Moraru
Copyright © 2021 Hermali Silva et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Skin lesions are a feature of many diseases including cutaneous leishmaniasis (CL). Ulcerative lesions are a common manifestation
of CL. Response to treatment in such lesions is judged through the assessment of the healing process by regular clinical
observations, which remains a challenge for the clinician, health system, and the patient in leishmaniasis endemic countries. In
this study, image processing was initially done using 40 CL lesion color images that were captured using a mobile phone
camera, to establish a technique to extract features from the image which could be related to the clinical status of the lesion.
The identified techniques were further developed, and ten ulcer images were analyzed to detect the extent of inflammatory
response and/or signs of healing using pattern recognition of inflammatory tissue captured in the image. The images were
preprocessed at the outset, and the quality was improved using the CIE L ∗ a ∗ b color space technique. Furthermore, features
were extracted using the principal component analysis and profiled using the signal spectrogram technique. This study has
established an adaptive thresholding technique ranging between 35 and 200 to profile the skin lesion images using signal
spectrogram plotted using Signal Analyzer in MATLAB. The outcome indicates its potential utility in visualizing and assessing
inflammatory tissue response in a CL ulcer. This approach is expected to be developed further to a mHealth-based prediction
algorithm to enable remote monitoring of treatment response of cutaneous leishmaniasis.
1. Introduction Leishmaniasis is a neglected tropical disease affecting
about 88 countries mainly in the tropics and subtropics
Interdisciplinary approaches are becoming increasingly pop- (http://www.cdc.gov). A bite of an infected sand fly can cause
ular in the health sector, specially to improve the diagnosis the disease in humans, and the type of disease manifestations
and management of various diseases [1–4]. Variety of may vary from localized lesions or ‘wounds’ on skin to visceral
methods of image processing and analyses are available for form which affects the internal organs of the patient. The most
extracting valuable information from raw images taken from common disease form in the world is cutaneous leishmaniasis
cameras [5]. With the advent of mobile phones, the concept (CL) which affects the skin [8]. According to their morpholog-
of mobile health or mHealth emerged and it could be ical appearance, CL lesions are commonly categorized as pap-
broadly described as a medical and public health practice ules, nodules, plaques, and ulcers. Of these, the most common
supported by mobile devices such as mobile phones [6]. is ulcers [8]. A CL ulcer has a ‘volcanic’ appearance with a cen-
With the increased usage of smart phones and tablets among tral crater and a raised border [9].
people, their contribution to image analysis applications and Visual and clinical assessment of the lesions is generally
mHealth has also been noteworthy [7]. used in CL for monitoring the treatment response. This
, 2 Computational and Mathematical Methods in Medicine
requires a skilled, trained medical personnel. However, some
of the regions burdened with leishmaniasis are remote where Transform to Extract
Crop RGB
RGB image CIE 1976 lab luminosity
mobility and access to specialized dermatology treatment image
color space channel
centres and specialized medical personnel are limited. Even
in areas with access to the above, some patients may opt to
seek follow-up through mHealth due to the convenience,
economic reasons, or when movements are restricted such
as during the current COVID-19 pandemic. Furthermore,
accessibility to mobile devices even in tropical countries, Pattern Grayscale
including Sri Lanka has exponentially increased during Thresholding Histogram
recognition PCA
the past decade. A series of surveys on the access and
use of information and communications technology, con-
ducted from 2017 to 2019, found that most of the Asian Figure 1: System block diagram of the agile integration technique
for image processing and analysis.
countries surveyed had about two-thirds of their 15–65-
year-old population owning a mobile phone, with 78% to
be exact in Sri Lanka [10]. This survey furthermore iden- and analysis techniques to process mobile captured JPEG
tified that Sri Lankan rural dwellers had a similar likeli- images of cutaneous leishmaniasis lesions to detect the
hood of owning a mobile phone as the urban dwellers. inflamed tissue and relating it to the treatment response
The need to seek digitalized solutions for health-related of the lesions.
issues that may be accessed remotely is felt stronger now
than ever due to the ‘new normal’ adaptations required
to survive the pandemic situation. 2. Materials and Methods
mHealth approaches using mobile phone applications 2.1. Patients and Image Acquisition. Fifty patients who
for presumptive diagnosis of leishmaniasis and using image attended the District General Hospital, Hambantota, Sri
processing for assessment of skin lesions have been reported Lanka, for treatment of cutaneous leishmaniasis were
with emphases on the need and space for further improve- selected for this study. Parasitological confirmation of leish-
ment [11–14]. Various image processing techniques are maniasis was made by microscopy and/or culture. Color
employed at different stages of conversion of a raw image images of the lesions were taken from a mobile phone,
(viz image acquisition, image preprocessing, clustering, and before starting the treatment with weekly injections of intra-
classification), into making it more useful for extracting infor- lesional sodium stibogluconate (IL-SSG), which is an anti-
mation [15]. Techniques such as color space, principal compo- mony containing drug used as the standard treatment for
nent analysis (PCA), and pattern recognition have been used CL in Sri Lanka [21]. A preliminary image processing was
for image processing in various medical fields [16–18]. done for 40 images which included all phenotypes (papules
Studies have shown the presence of a local inflammatory n = 10, nodules n = 10, plaques n = 10, and ulcers n =10) as
reaction at the lesion site in CL [19, 20]. This inflammatory given below under section (A), to establish an image pro-
response includes increased vascular permeability, dilatation cessing technique to extract features from the images which
of blood vessels in the dermis, and infiltration of the lesion could be related to the inflammatory response and/or treat-
site with immune cells, which would contribute to the swell- ment response of all types of CL lesions. Since it was noted
ing and the morphology of the lesion [13]. Even though the through the results of the preliminary analysis that the areas
inflammatory changes are not pathognomonic to CL, the with the inflamed tissue in ulcers could be visualized differ-
reduction in the inflammatory response is an indication of ently and clearly from the healthy tissue, the image process-
healing or a good response to treatment in CL. While utilizing ing methodology was further developed into an agile
microscopy and other parasitological laboratory methods to integration methodology of eight blocks to detect the extent
confirm the diagnosis of leishmaniasis, a technique to visualize of inflammatory response and/or signs of healing in ulcers,
the inflamed tissue in a CL lesion will be valuable in treatment using 10 images of ulcers (Section (B)). Ethical approval
monitoring and follow-up of patients. Appreciating the for this study was obtained from the Ethics Review Commit-
mHealth concept, in the simplest scenario, the patients or tee (EC/16/080), Faculty of Medicine, University of
the primary health care workers can capture the lesion image Colombo (http://www.med.cmb.ac.lk), Sri Lanka.
using a mobile phone and upload to a cloud based diagnostic
centre for an analysis report to be forwarded to a consultant (A) Preliminary image processing (n = 40, phenotypes:
medical specialists for treatment monitoring and evaluation papules, nodules, plaques, ulcers)
purposes as part of clinical management.
Based on the currently available information in digital An image processing technique using MATLAB (The
and health fields, the authors hypothesized that image pro- MathWorks, Inc., Natick, Massachusetts, United States)
cessing could be used to visualize the inflammatory tissue was adapted to process 40 mobile captured JPEG images.
in the CL lesions and could be utilized as a noncontact The images were resized to 256 × 256 pixel resolution, and
assessment or a self-assessment method to detect the grayscale conversion was done to standardize the variation
inflammatory response in CL. Thus, the aim of this study of images. The standardized images were enhanced by using
was to investigate the concept of using image processing a contrast stretching algorithm to identify the boundary of
