Original Article

Automated hand thermal image

segmentation and feature extraction in
the evaluation of rheumatoid arthritis

Proc IMechE Part H:

J Engineering in Medicine
2015, Vol. 229(4) 319331
IMechE 2015
Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0954411915580809
pih.sagepub.com

U Snekhalatha1, M Anburajan1, V Sowmiya2, B Venkatraman3 and

M Menaka3

Abstract
The aim of the study was (1) to perform an automated segmentation of hot spot regions of the hand from thermograph
using the k-means algorithm and (2) to test the potential of features extracted from the hand thermograph and its measured skin temperature indices in the evaluation of rheumatoid arthritis. Thermal image analysis based on skin temperature measurement, heat distribution index and thermographic index was analyzed in rheumatoid arthritis patients and
controls. The k-means algorithm was used for image segmentation, and features were extracted from the segmented
output image using the gray-level co-occurrence matrix method. In metacarpo-phalangeal, proximal inter-phalangeal and
distal inter-phalangeal regions, the calculated percentage difference in the mean values of skin temperatures was found
to be higher in rheumatoid arthritis patients (5.3%, 4.9% and 4.8% in MCP3, PIP3 and DIP3 joints, respectively) as compared to the normal group. k-Means algorithm applied in the thermal imaging provided better segmentation results in
evaluating the disease. In the total population studied, the measured mean average skin temperature of the MCP3 joint
was highly correlated with most of the extracted features of the hand. In the total population studied, the statistical feature extracted parameters correlated significantly with skin surface temperature measurements and measured temperature indices. Hence, the developed computer-aided diagnostic tool using MATLAB could be used as a reliable method in
diagnosing and analyzing the arthritis in hand thermal images.

Keywords
Heat distribution index, thermographic index, rheumatoid arthritis, k-means algorithm

Date received: 1 September 2014; accepted: 13 March 2015

Introduction
Rheumatoid arthritis (RA) is a chronic autoimmune
disease that causes inflammation of blood vessels,
development of bumps called rheumatoid nodules,
weakening of bones and deformity of the joints leading
to long-term disability. It causes premature mortality,
disability and compromised quality of life in the industrialized and developing world.1 The prevalence of RA
ranges from 0.5% to 1% worldwide in the general population.1 The prevalence rate in India is 0.9%, almost
equal to the world prevalence rate.2 RA is associated
with severe disability and substantial morbidity.3,4
Symmons and Gabriel5 reported that mortality is
greater in patients with established RA in comparison
with the general population. RA directly affects physical function and mobility and results in substantial
short-term and long-term morbidity.
Thermal imaging is a non-contact, noninvasive,
diagnostic imaging procedure based on skin

temperature measurement. It is a functional imaging

method for analyzing physiological function related to
body temperature. Also, it has been used as a complimentary diagnostic tool in various clinical applications
of rheumatology such as diagnosis of patellafemoral
arthralgia6 and monitoring of skin temperature elevations in scleroderma,7 RA,8 juvenile RA9 and hand and
knee osteoarthritis.10,11
Jiang et al.12 hypothesized that thermography may
provide a sensitive, noninvasive method to find disease

Department of Biomedical Engineering, SRM University, Chennai, India

Department of Biomedical Engineering, Jerusalem College of
Engineering, Anna University, Chennai, India
3
Indra Gandhi Center for Atomic Research, Kalpakkam, India
2

Corresponding author:
M Anburajan, Department of Biomedical Engineering, SRM University,
Kattankulathur 603203, Chennai, Tamil Nadu, India.
Email: hod.biomedi@ktr.srmuniv.ac.in

Downloaded from pih.sagepub.com by guest on May 1, 2015

320

Proc IMechE Part H: J Engineering in Medicine 229(4)

severity both in animal models and in human studies.

Glehr et al.13 used computer-assisted infrared thermography to diagnose anterior knee pain following
implantation of artificial knee joints. Trafarski et al.14
demonstrated that thermography is useful in analyzing
the thermal reaction of the organism connected with
local cryostimulation and monitoring and evaluating
the progress of treatment in RA patients.
Thermography has been used as an effective tool in
quantification of inflammation in hand joints of RA. It
has been evident that thermographic changes in skin
temperature over the areas of chronic inflammation
accurately reproduce the changes in other chemical and
cellular process in RA.15 Various studies have been performed in last and recent decade for the assessment of
RA using thermography in hand regions.1618
Thermographic index (TI) and heat distribution index
(HDI) have been used as quantitative measurement of
inflammation in rheumatoid joint.19,20
Several boundary edge detection techniques such as
probability-based edge detection techniques and canny
edge detection techniques have been applied in thermal
images for the extraction of region of interest (ROI).21
The edge detection approach is not suitable for segmenting the thermal images because additional effort is
required to connect the incomplete edges into a distinct
and meaningful object boundary. Hence, it requires a
need of clustering-based segmentation techniques for
extracting the ROI. Thermograms, either grayscale or
pseudo color, are processed for detection of abnormal
regions and quantification. However, temperature variations are not visible to naked eye. Hence, it is necessary to develop and analyze feature extraction
algorithms for abnormality detection and classification
of disease state and normal.22
The aim of our study was (1) to perform an automated segmentation of hot spot region of the hand
from thermographs using the k-means algorithm and
(2) to test the potential of features extracted from the
hand thermograph and its measured skin temperature
indices in the evaluation of RA.

Patients and methods

Study design and population
This study was approved by our institutional ethical
committee (approval no. 35/iec/2010), and the informed
consent statement was signed by each participant. A
free medical screening camp to identify RA in the suburban South Indian population was conducted at SRM
Hospital and Research Centre, Kattankulathur,
Chennai, Tamil Nadu, India, on 1519 February 2011.
Patients with major systemic disease were included in
the study. However, patients who had undergone recent
physiotherapy, thermotherapy, phototherapy and fever
were excluded. According to the Indian Rheumatology

Association (IRA) consensus report 2008,23 individuals

with persistent inflammatory arthritis with more than
four joints, high Erythrocyte Sedimentation Rate
(ESR) (RA . 20 mm/h)/high C-reactive protein (CRP)
(RA . 80 mg/dL), positive immunoglobulin M (Igm)
rheumatoid factor (RA . 20 U/mL), radiographic
changes in juxta-articular osteopenia, erosions and joint
space narrowing were classified as RA. In all the
patients, both the hands were symmetrically affected
by RA.
A total number of 50 subjects, age ranging from 30
to 75 years, were registered in the camp. According to
the diagnostic criteria of the study, 30 subjects (male/
female (M/F) ratio = 1:3) were found to have known
RA, and their mean age was found to be 45.3 6 11.4
years. They had disease duration of 4.3 6 2.5 years, 15
age- and sex-matched normal subjects (M/F ratio =
1:3), their mean age was found to be 45.5 6 11.8 years,
were included in the study. The remaining five subjects
were excluded according to the exclusion criteria of the
study.

Thermal image acquisition procedure and analysis

The acquisition of thermal image was performed
according to the guidelines recommended by the
International Academy of Clinical Thermology.24 All
the subjects were asked to remove all the ornaments
and were seated with their hands exposed for 15 min in
a temperature-controlled room at 20 C with humidity
of 45%50%. The subjects were then asked to stand in
the image capturing room with their hands kept in both
dorsal view and ventral view. The thermal camera was
positioned at a distance of 1.0 m parallel to the hand,
and a thermal image of both dorsal and ventral views
of both right and left hand was taken in 5 s. All the
thermal images were acquired in the same room for a
period of 1 week in the forenoon session only, to avoid
the effect of cyclic variations in the atmospheric temperature in the study.
A hand-held thermal camera (ThermaCAM-T400;
FLIR, Wilsonville, OR, USA) was used to image the
hand region of both the groups of RA and normal.
The thermal camera T400 utilizes the 320 3 420 thermal element focal plane array (FPA) uncooled microbolometer detector system with minimum focusable
distance of 0.4 m to infinity. The camera could measure
the temperature range of 220 C to 1200 C to an
accuracy of 2% with thermal sensitivity of 0.05 C. The
images were stored and then analyzed using the software FLIR Quick report, version-1.2 and further processed with MATLAB version7.1 (MathWorks Inc.,
Natick, MA, USA).
From the thermal image of the hand, the average
skin temperature (C) of a total of 28 index joints,
which includes 10 distal inter-phalangeal (DIP) joints,
8 proximal inter-phalangeal (PIP) joints and 10

Downloaded from pih.sagepub.com by guest on May 1, 2015

Snekhalatha et al.

321

metacarpo-phalangeal (MCP) joints of both the hands

taken in dorsal view, was evaluated using FLIR software. It was measured by positioning an available
square area tool of ROI of size 8 3 8 mm in each
respective joint in dorsal view. Furthermore, the average skin surface temperature was measured in the ventral side (palm region).

5.

6.

Temperature indices

7.

The HDI was defined as twice the standard deviation

(SD) of all the temperatures within the ROI of the
MCP joint. It represents the pattern and spread of temperature across joints.25 TI describes the peak temperature of the joint and the area of each temperature band.
It was used as a quantitative evaluation of skin thermal
changes in the ROI. The TI was calculated using the
formula
P
( DT3a)
1
TI =
A
P
where
represents the summation, DT represents the
difference between the mean skin temperature over the
monitored area and the temperature of the isotherm
and a indicates the area (cm2) occupied by isotherm
and A represents the total area (cm2) of thermogram.
The TI values were used to diagnose the individual, participated in the study as follows: (1) 42: normal, (2) 3
4: osteoarthritis and (3) . 4: RA.26

Thermal image segmentation algorithm

Automated thermal image segmentation was performed
by applying k-means algorithm to hand thermal images
of both dorsal and ventral views. k-Means was an
unsupervised clustering algorithm which classifies the
data based on k number of groups based on image features. The grouping was done by minimizing the sum
of squares of distance between data points and the corresponding cluster centroids.
A summary of the k-means algorithm27 is given here
as follows:
1.
2.

3.

4.

Input image was the hand thermal image taken in

dorsal/ventral view.
The RGB image was converted to HSV color
space.
The reasons for conversion of RGB to HSV were
given as follows: (1) RGB defines color in terms of
a combination of primary colors, whereas HSV
describes it using more familiar comparisons such
as color, vibrancy and brightness; (2) furthermore,
HSV describes color similarly to how the human
eye tends to perceive it.
Three classes (k = 3) were chosen and assigned as
initial centroidsClass 1: cluster 1; Class 2: cluster
2; Class 3: cluster 3.
The distance between the centroid and each pixel
of the input was computed.

According to the minimum distance criterion, the

data were clustered as follows:
(a) Cluster 1 to separate hot spot region;
(b) Cluster 2 to separate background region;
(c) Cluster 3 to separate other than hot spot
regions as well as background.
The three centroids using the average of all HSV
points in each cluster were updated.
Steps 16 were repeated until no points switch to a
new cluster, or until we hit a maximum number of
25 iterations.

Statistical feature extraction

The feature extraction technique is implemented over
the segmented output image to extract the intensity features and statistical texture features. The features such
as mean, SD, smoothness, entropy, kurtosis, skewness,
variance, contrast, correlation and energy are extracted
from the segmented output image using gray-level cooccurrence matrix (GLCM) method28 and explained as
follows:
1.

Mean
The mean is the average intensity value of the hot
spot regions and was given as
X X
Mean =
i
jiP(i, j)
2

2.

SD
The SD is a measure of dispersion or variation
from the mean value. A low SD represents the
data points to be very close to the mean value
and a high SD represents the data points are
spread out over the large range of values
s=

N1 N
1
X
X

!1=2
2

(i  m) p(i, j)

i=0 j=0

3.

Entropy
Entropy is the measure of randomness that can
be used to characterize the texture of the image.
It is a quantitative measure of image information.
If all the image pixels have the same gray level,
minimum entropy is achieved; if the pixels have a
uniform distribution of gray levels or the image is
histogram equalized, maximum entropy is
achieved
X X
Entropy =
i
jP(i, j) logP(i, j)
4

4.

Skewness
Skewness is the measure of the asymmetry of the
pixel distribution around its mean. A symmetric
distribution has zero skewness and has equal values for mean, median and mode. If the skewness
is positive, then the data are positive skewed or
skewed right, meaning that the right tail of the

Downloaded from pih.sagepub.com by guest on May 1, 2015

322

Proc IMechE Part H: J Engineering in Medicine 229(4)

distribution is longer than the left, the mean is
greater than the median and the mode is less than
the median. If the skewness is negative, the data
are negatively skewed, meaning that the left tail
of the distribution is longer than the right, the
mean is less than the median and the mode is
greater than the median
hP
i
P(i, j)  m2 =N
Skewness =
5
s3

5.

Kurtosis
Kurtosis refers to the measure of the peak of the
array of intensity distribution. Similar to skewness, kurtosis is a descriptor of the shape of the
probability distribution. There are three different
interpretations of kurtosis: mesokurtic distribution, leptokurtic distribution and platykurtic distribution. A zero excess kurtosis indicates the
mesokurtic distribution which has a normal
peak around the mean. A positive excess kurtosis
represents the leptokurtic distribution which has a
more acute peak around the mean. A negative
excess kurtosis shows the platykurtic distribution
which has a lower, wider peak around the mean29
i
i
Xhh
Kurtosis =
P(i, j)  m4 =N =s4  3
6

6.

7.

Variance
The variance is the square of the SD. A small
variance indicates that the data points are close
to the mean, whereas the large variance
means that the data points are spread out from
the mean
X X
Var =
i
j(i  m)2 P(i, j)
7
Contrast
The contrast is the measure of the intensity contrast between the pixel and its neighbors
XX
Contrast =
i, j2 Pi, j
8
i=0 i=0

8.

Energy
Energy is given by the sum of the squared elements in the GLCM. Energy measures the textural uniformity that means pixel pairs
repetitions. It also measures the smoothness of
the image. For more uniformly distributed pixels
in image regions, the smoothness level and energy
were low. But in case of the nonuniform distributed region, the smoothness and energy are high.
Hence, in case of RA patients, due to uneven distribution of temperature in hand regions, the
energy obtained was higher compared to the
normal30

Energy =

(Pi, j )2

i, j = 0

9.

Correlation
Correlation is a measure of linear dependencies of
gray level on those of neighboring pixels
Correlation =

X X (i  m)(j  m)P(i, j)
si sj

i=0 j=0

10.

10

Homogeneity
Homogeneity is used to determine whether frequency counts are distributed identically across
different populations
Homogenity =

N1
X

pij

i, j = 0 1 + i

 j2

11

Statistical analysis
Data were analyzed using SPSS software package version 19.0 (SPSS Inc., Chicago, IL, USA). The measured
mean surface temperature, HDI, TI and feature extraction parameters were compared between the RA group
and normal group using a Students t-test. The
KolmogorovSmirnov and ShapiroWilk tests were
performed to test the normality of above-mentioned
variables which gave a significance value (p \ 0.05).
The bi-variate (Pearson correlation) analyses were used
to obtain the correlations between HDI, TI, skin temperature measurements and feature extraction parameters in the total population studied.

Results
HDI and TI analysis
The measured temperature indices correlated significantly with the feature extracted parameters in the total
population studied (Table 1). The mean values of measured HDI and TI in RA patients were highly significant (p \ 0.01) when compared to the normal group.
The mean (6SD) HDI values of both RA and normal
groups were obtained as 0.79 (60.2) and 0.38 (60.1),
respectively, and found to be statistically significant.
Also, the measured mean (6SD) values of TI obtained
for the RA group and normal groups were 4.18 (60.5)
and 2.69 (60.5), respectively (Table 2). The percentage
of patients whose diagnostic values of TI categorized
into normal, osteoarthritis and RA was 24.4%, 8.8%
and 66.6%, respectively.

Thermal image analysis and feature extraction

In the total population (n = 45), the temperature
indices such as HDI and TI correlated significantly (p
\ 0.01) with the feature extracted parameters
(Table 1). The mean average skin temperature of
MCP1, MCP3, MCP4 and MCP5 correlated

Downloaded from pih.sagepub.com by guest on May 1, 2015

Snekhalatha et al.

323

Table 1. Correlation matrix between temperature parameter and statistical feature extracted parameter measured from hand
thermal image in total population studied.
Temperature
parameters

Feature extraction parameters

Mean

Standard
deviation

Skewness

Entropy

Thermal indices
HDI
0.4**
0.42**
0.39*
0.45**
TI
0.46** 0.51**
0.44**
0.47**
Skin temperature at MCP joints, PIP joints and DIP joints
0.32*
0.31*
MCP1
0.45** 0.24#
0.34*
0.21#
MCP2
0.46** 0.25#
MCP3
0.48** 0.51**
0.52**
0.51**
MCP4
0.46** 0.48**
0.48**
0.49**
MCP5
0.41** 0.26
0.42**
0.39*
PIP2
0.3*
0.29*
0.37*
0.27#
0.38*
0.41**
0.26#
PIP3
0.27#
0.26#
0.38*
0.26#
PIP4
0.27#
PIP5
0.3*
0.3*
0.37*
0.31*
0.09#
0.11#
0.2#
DIP1
0.27#
DIP2
0.24#
0.2#
0.24#
0.25\#
#
#
#
DIP3
0.23
0.23
0.27
0.26#
DIP4
0.23#
0.16#
0.31*
0.25#
#
#
#
DIP5
0.22
0.15
0.26
0.26#

Kurtosis

Variance

Contrast

Correlation

Energy

Homogeneity

0.39*
0.44**

0.38*
0.33*

0.56**
0.61**

0.44**
0.53**

0.44**
0.54**

0.45**
0.54**

0.25#
0.20#
0.43**
0.42**
0.26#
0.48**
0.46**
0.48**
0.47**
0.05#
0.22#
0.21#
0.17#
0.14#

0.42**
0.40**
0.31*
0.26
0.39*
0.24#
0.32*
0.24#
0.29*
0.019#
0.05#
0.06#
0.0002#
0.22#

0.25#
0.27#
0.55**
0.53**
0.50**
0.29*
0.38*
0.26
0.25#
0.005#
0.16#
0.15#
0.07#
0.03#

0.30*
0.26#
0.56**
0.30*
0.33*
0.52**
0.56**
0.51**
0.45**
0.03#
0.21#
0.23#
0.24#
0.22#

0.33*
0.28*
0.56**
0.30*
0.28*
0.3*
0.32*
0.29*
0.28*
0.016#
0.12#
0.14#
0.16#
0.12#

0.39*
0.39*
0.56**
0.41**
0.46**
0.32*
0.34*
0.31*
0.29*
0.05#
0.16#
0.18#
0.19#
0.16#

HDI: heat distribution index; TI: thermographic index; MCP: metacarpo-phalangeal; PIP: proximal inter-phalangeal; DIP: distal inter-phalangeal; NS:
not significant.
*p \ 0.05; **p \ 0.01; #NS.

Table 2. Measured mean average skin surface temperature of RA and normal in required region of interest.
Temperature
parameters

Region of
interest

RA (N = 30),
mean 6 SD

Normal (N = 15),
mean 6 SD

% Difference

Statistical
significance (p)

0.38 6 0.1
2.69 6 0.5

51.89
35.64

0.001
0.001

33.93 6 0.3
33.86 6 0.2
33.64 6 0.2
33.76 6 0.2
33.86 6 0.2
33.80 6 0.2
33.71 6 0.2
33.72 6 0.2
33.67 6 0.1
33.87 6 0.4
33.86 6 0.2
33.55 6 0.4
33.57 6 0.3
33.75 6 0.1
34.43 6 0.2

3.84
4.35
5.29
4.44
4.35
4.79
4.88
4.71
4.60
3.80
4.35
4.79
4.68
3.98
2.71

0.002
0.003
0.001
0.003
0.004
0.002
0.001
0.004
0.001
0.005
0.001
0.001
0.004
0.005
0.004

2.89E 2 06 6 1.08E 2 06
2.08E 2 05 6 9.58E 2 06
0.001 6 0.0008
0.0001 6 5.61E 2 05
0.008 6 0.007
4.06E 2 13 6 8.47E 2 13
3.94E 2 08 6 1.37E 2 08
6.94E 2 06 6 2.18E 2 06
0.0004 6 0.0003
0.00044 6 0.0003
0.00046 6 0.0003

94.25
74.59
8.33
95
93.75
50.25
50.27
89.39
75
71.42
76.47

0.0003
0.0002
0.0009
0.0002
0.0004
0.06
0.03
3.66E 2 06
0.0002
0.0004
0.0002

Thermal indices
HDI
Whole hand
0.79 6 0.2
TI
Whole hand
4.18 6 0.5
Skin temperature measurements at joints
MCP
MCP1
35.39 6 0.6
MCP2
35.40 6 0.6
MCP3
35.52 6 0.7
MCP4
35.33 6 0.6
MCP5
35.40 6 0.6
PIP
PIP2
35.42 6 0.6
PIP3
35.44 6 0.4
PIP4
35.39 6 0.4
PIP5
35.22 6 0.7
DIP
DIP1
35.21 6 0.5
DIP2
35.40 6 0.6
DIP3
35.24 6 0.6
DIP4
35.22 6 0.5
DIP5
35.15 6 0.5
Ventral side
palm
35.39 6 0.7
Measured statistical features
Mean
Whole hand
5.03E 2 05 6 5.868E 2 05
Std dev
Whole hand
8.54E 2 05 6 7.63E 2 05
Skewness
Whole hand
0.006 6 0.006
Entropy
Whole hand
0.002 6 0.002
Kurtosis
Whole hand
0.128 6 0.15
Smoothness
Whole hand
5.1941E 2 11 6 1.368E 2 10
Variance
Whole hand
7.924E 2 08 6 9.08E 2 08
Contrast
Whole hand
6.5461E 2 05 6 5.054E 2 05
Correlation
Whole hand
0.0016 6 0.0014
Energy
Whole hand
0.0014 6 0.001
Homogeneity
Whole hand
0.0017 6 0.001

RA: rheumatoid arthritis; SD: standard deviation; HDI: heat distribution index; TI: thermographic index; MCP: metacarpo-phalangeal; PIP: proximal
inter-phalangeal; DIP: distal inter-phalangeal.
Downloaded from pih.sagepub.com by guest on May 1, 2015

324

Proc IMechE Part H: J Engineering in Medicine 229(4)

significantly (p \ 0.05) with features such as mean,

variance, skewness, energy, correlation, entropy and
homogeneity. As an overall comparison of all the MCP
joints, it was found that in MCP3 joint, the measured
mean average skin temperature was found to be highly
correlated with the feature extracted parameters in the
total population studied. Also, the mean skin surface
temperature of PIP2, PIP3 and PIP5 correlated with
most of feature extracted parameters in the total population studied. But DIP joints had not shown any significant correlation between the mean average skin
surface temperature and the feature extracted
parameters.
Table 2 shows the measured mean average skin temperature values at regions of DIP, PIP and MCP from
the thermograph both in RA and normal groups. It
was observed that the mean values of skin temperature,
measured at all joints of the hand, were higher in the
RA group, compared to its normal counterparts.

In MCP, PIP and DIP regions, the calculated percentage difference in the mean values of skin temperatures
was found to be higher in MCP3, PIP3 and DIP3
regions of interests in the RA group, as compared to
the normal group. These values were 5.3%
[((35.52 2 33.64)/35.52) 3 100], 4.9% [((35.44 2 33.71)/
35.44) 3 100] and 4.8% [((35.24 2 33.55)/35.24) 3 100],
respectively, and these values were statistically significant at p \ 0.001. Thus, in the RA group, the MCP3
joint region showed highest percentage difference in the
mean values of skin temperature, as compared to normal group. Figure 1(a) and (b) represents the dorsal
view of the hand thermal image of normal subject and
the RA patient, respectively. In total, 14 ROIs (square
boxes) are shown with the minimum and maximum
skin surface temperatures. These ROIs correspond to
MCP (15), PIP (25) and DIP (15) joints of each
hand of a subject in the RA and normal groups.
Similarly, Figure 1(c) and (d) represents the ventral

Figure 1. (a) Normal subjectdorsal view of the hand thermal image. The square ROIs indicate the boxes depicting the minimum
temperature ranges (30.6 C32.5 C) and maximum skin surface temperature ranges (31.5 C32.5 C) at the MCP, PIP and DIP
joints. (b) RA patientsdorsal view of the hand thermal image. The square ROIs indicate the boxes depicting the minimum
temperature ranges (31.9 C33.6 C) and maximum skin surface temperature ranges (33 C34.8 C) at the MCP, PIP and DIP
joints. (c) Normal subjectventral view of the measured thermal image. The square ROI indicates a box depicting the minimum
temperature (34.5 C) and maximum skin surface temperature (35.9 C) at palm region. (d) RA patientventral view of the hand
thermal image. The square ROI indicates a box depicting the minimum temperature (35.5 C) and maximum skin surface
temperature (36.5 C) at palm region.

Downloaded from pih.sagepub.com by guest on May 1, 2015

Snekhalatha et al.

325

Figure 2. RA patientdorsal view of various stages of the segmented output hand thermal image: (a) input hand thermal image, (b)
cluster 1 image depicts presence of hot spot region, (c) cluster 2 image depicts the background, (d) cluster 3 image depicts the other
regions, (e) segmented output image depicts presence of hot spot region and (f) segmented gray output image with hot spot region.

view of the hand thermal image of RA and normal

groups, respectively. The ROI indicated by the square
box depicts the minimum and maximum skin surface
temperature in the palm region (ventral side).
The thermal images of both hands in dorsal view
were segmented for RA and normal subjects using the
k-means algorithm (Figures 2 and 3). Figure 2 shows
the dorsal view of RA patient thermal image with the
various stages of the segmented output image; specifically Figure 2(a) shows the input image; Figure 2(b)

represents the cluster 1 image displaying the hot spot

regions (absent for normal group); Figure 2(c) illustrates the cluster 2 image depicting the background;
Figure 2(d) displays the cluster 3 image showing the
other regions; Figure 2(e) indicates the segmented hot
spot regions (absent in case of normal) and Figure 2(f)
represents the gray segmented output image. Figure 3
represents the dorsal view of normal thermal image,
where Figure 3(a) shows the input image; Figure 3(b)
represents the cluster 1 image displaying the hot spot

Downloaded from pih.sagepub.com by guest on May 1, 2015

326

Proc IMechE Part H: J Engineering in Medicine 229(4)

Figure 3. Normal subjectdorsal view of the various stages of segmented output hand thermal image: (a) input hand thermal
image, (b) cluster 1 image depicts absence of hot spot region, (c) cluster 2 image represents the background, (d) the cluster 3 image
depicts the other region, (e) segmented output image depicts the absence of hot spot region and (f) segmented gray output image
without any hot spot region.

regions; Figure 3(c) illustrates the cluster 2 images

depicting the background; Figure 3(d) displays the cluster 3 images showing the other regions; Figure 3(e) indicates the segmented hot spot regions and Figure 3(f)
represents the gray segmented output image. Similarly,
the ventral view of thermal images (Figures 4 and 5)
demonstrates the segmented hot spot region if any present in RA and normal groups.
In the RA group (n = 30), the feature extracted
parameters were at significantly higher values

compared to normal group (p \ 0.05) as mentioned in

Table 2. In comparison of RA and normal, the feature
extraction parameters such as mean, skewness, entropy,
kurtosis, contrast, correlation, energy and homogeneity
show high significance (p \ 0.01), but moderate significance (p \ 0.05), achieved in features such as smoothness and variance.
A computer-aided diagnostic tool using MATLAB
coding was developed for automated image segmentation of hot spot regions, feature extraction and

Downloaded from pih.sagepub.com by guest on May 1, 2015

Snekhalatha et al.

327

Figure 4. RA patientventral view of various stages of the segmented output hand thermal image: (a) input hand thermal image, (b)
cluster 1 image depicts presence of hot spot region, (c) cluster 2 image depicts the background, (d) cluster 3 image depicts the other
regions, (e) segmented output image depicts presence of hot spot region and (f) segmented gray output image with hot spot region.

classification of subjects into RA group and normal

group based on the segmented hot spot region of the
hand thermal image. Figure 6(a) and (b) shows
computer-aided diagnostic tool of thermal image analysis of hand-segmented image of RA patient and normal
subject, respectively.

Discussion
In this study, the mean average skin surface temperature was measured at DIP, PIP and MCP of digits 15

which were greater for the RA group compared to the

normal group. Frize et al.31 in their thermal infrared
imaging study predicted that second MCP, third MCP
and the knee joints showed the greatest statistical difference between the control subjects and the patients.
In our study, among all the joints, the third MCP, third
PIP and third DIP show statistically significant difference in temperature between RA group and normal
group. Hence, it was observed that in the total population studied, middle fingers have been affected with
RA severely compared to other fingers in hand region.

Downloaded from pih.sagepub.com by guest on May 1, 2015

328

Proc IMechE Part H: J Engineering in Medicine 229(4)

Figure 5. Normal subjectventral view of the various stages of segmented output hand thermal image: (a) input hand thermal
image, (b) cluster 1 image depicts absence of hot spot region, (c) cluster 2 image represents the background, (d) cluster 3 image
depicts the other region, (e) segmented output image depicts the absence of hot spot region and (f) segmented gray output image
without any hot spot region.

The difference in temperature between RA and normal

group in the palm region (ventral side) is less compared
to the dorsal side of the hand thermal image. Also, in
this study, from the dorsal view of the hand thermal
image, it was observed that hot spot region was found
at both wrist and joint regions. These findings had an

agreement with the result of thermal imaging study in

RA conducted by Spading et al.25
Borojevic et al.32 found in their study that the RA
patients had a mean surface temperature on the ventral
side of the hand varying from 24.8 C to 36.5 C with
that of the healthy controls ranging from 24.8 C to

Downloaded from pih.sagepub.com by guest on May 1, 2015

Snekhalatha et al.

329

Figure 6. A computer-aided diagnostic tool for thermal image analysis of hand-segmented image: (a) sample RA patient and (b)
sample normal subject.

35.5 C. In our study, RA patients had the mean average surface temperature in the ventral side of the hand
as 35.39 C compared with that of the healthy controls
as 34.43 C. The percentage temperature difference
obtained between RA and normal groups was higher
(2.71%) in our study compared to the study conducted
by Borojevic et al. (1.63%). Hence, the reason for
increased temperature in the hands of RA was due to
the increased vascularity caused by synovitis and the
proliferation of synovial cells. Also, the blood flow was
prevented from deep circulation to subcutaneous

circulation which causes an increase in temperature at

the surface of joints.
Collins et al. calculated the TI for quantitative analysis of inflammation using thermography in arthritis
using multi-isothermal analysis. They had obtained the
mean TI of 3.77 in the rheumatoid knees and compared
the isothermal pattern of inflamed knee with the
derived TI before and after treatment.19 In our study,
the mean TI attained was 4.37 in the rheumatoid
hands, which was higher by 35.6% than the normal
group. Also, there was a statistically significant

Downloaded from pih.sagepub.com by guest on May 1, 2015

330

Proc IMechE Part H: J Engineering in Medicine 229(4)

(p \ 0.01) positive correlation observed between TI

and feature extracted parameters in the total population studied. Salisbury et al.20 suggested the HDI as a
valid temperature parameter for interpreting the RA
information to detect abnormalities in the knee, wrist,
elbow and ankle joints. Ilowite et al.33 validated and
confirmed that thermographic HDI is one of the best
methods to assess inflammation of the synovial membrane for RA diagnosis. Spading et al. evaluated the
RA patients with standardized thermographic methods
and calculated the HDI as twice the SD of all the temperatures in the pre-defined ROI. In addition to that,
they suggested that thermal imaging could be used to
improve the assessment of disease activity in arthritis
and could quantify clinically reasonable changes in
arthritic joints in response to therapy.25 In our study,
we calculated the HDI in the hand thermal image of
RA group and normal group as suggested by Spading
et al.; the results revealed that the percentage difference
of HDI was found to be higher by 51.8% in the RA
group compared to normal group. To the best of our
knowledge, the correlation studies between temperature
indices and feature extracted parameters were limited.
But in our study, the correlation between HDI and feature extracted parameters was found to be positive and
statistically significant (p \ 0.05) in the total studied
population.
Zhou et al. investigated the utilization of level set
methods to extract the boundary on thermal IR
images. An initial contour is grown adaptively using a
speed function based on edge/direction map. They also
indicated that level set active contour method showed
promising results but are still unable to recover full
contours for weak edges and poor contrast images.34
Frize et al. used image processing techniques such as
histogram and thresholding for analysis of hand thermal images in RA. But thresholding techniques do not
perform well and often missed out important regions of
the object of interest or segment larger regions than the
expected regions where fine details are lost. They also
reported that features such as mode/max, variance and
maxmin calculations best discriminate between the
temperature measurement of control and patient
groups.35 In our study, k-means algorithm with three
clusters was used for effective segmentation of hot spot
regions of hand thermal images in RA group. Also, 10
statistical features were extracted from a segmented
output image to distinguish between RA group and
normal group. These features depict higher values for
RA patients compared to normal group and had an
agreement with findings of Frize et al.
The uniqueness of this study includes a computeraided diagnostic tool using MATLAB coding developed for automated image segmentation of hot spot
regions, feature extraction and classification of subjects
into RA group and normal group based on the segmented hot spot region of the hand thermal image. The
potential limitation of this study includes requirement

to validate the results for the largest number of sample

size.

Conclusion
In conclusion, in the population studied, the measured
mean average skin temperature of MCP3 joint highly
correlated with most of the extracted features of the
hand. Thermal imaging parameters such as skin temperature measurements, TI and HDI correlated significantly with feature extracted parameters in the
population studied. Hence, the developed computeraided diagnostic tool using MATLAB could be used as
a reliable method in diagnosing and analyzing the
arthritis in hand thermal images.
Declaration of conflicting interests
The authors declare that there is no conflict of interest.
Disclosure
This article has been read and approved by all the
authors. This article is not under consideration by any
other publications and has not been published
elsewhere.
Funding
This research received no specific grant from any funding agency in the public, commercial or not-for-profit
sectors.
References
1. Silman AJ and Hochberg MC. Epidemiology of the rheumatic diseases. 2nd ed. New York: Oxford University
Press, 2001.
2. Milind P and Sushila K. How to live with rheumatoid
arthritis? Int Res J Pharm 2012; 3(3): 115121.
3. Pincus T. Long term outcomes in rheumatoid arthritis.
Br J Rheumatol 1995; 34(Suppl. 2): 5973.
4. Sokko T. Work disability in early rheumatoid arthritis.
Clin Exp Rheumatol 2003; 5(Suppl. 31): 7174.
5. Symmons DP and Gabriel SE. Epidemiology of CVD in
rheumatic disease, with a focus on RA and SLE. Nat
Rev Rheumatol 2011; 7(7): 399408.
6. Devereaux MD, Parr GR, Lachmann SM, et al. Thermographic diagnosis in athletes with patellofemoral arthralgia. J Bone Joint Surg Br 1986; 68(1): 4244.
7. Maeda M, Kachi H, Ichihashi N, et al. The effect of electrical acupuncture stimulation therapy using thermography and plasma endothelin levels in patients with
progressive systemic sclerosis. J Dermatol Sci 1998;
17(2): 151155.
8. Devereaux MD, Parr GR, Page Thomas DP, et al. Disease activity indexes in rheumatoid arthritis: a prospective comparative study with thermography. Ann Rheu
Dis 1985; 44: 434437.
9. Viitanen SM and Laaksonen AL. Thermography in juvenile rheumatoid arthritis. Scand J Rheumatol 1987; 16(1):
9198.

Downloaded from pih.sagepub.com by guest on May 1, 2015

Snekhalatha et al.

331

10. Varju G, Pieper CF, Renner JB, et al. Assessment of hand

osteoarthritis: correlation between thermographic and
radiographic methods. Rheumatology 2004; 43(7): 915
919.
11. Warashina H, Hasegawa Y, Tsuchiya H, et al. Clinical
radiographic and thermographic assessment of osteoarthritis in the knee joints. Ann Rheum Dis 2002; 61(9): 852854.
12. Jiang LJ, Ng EY, Yeo AC, et al. A perspective on medical infrared imaging. J Med Eng and Technol 2005; 29(6):
257267.
13. Glehr M, Stibior A, Sadoghi P, et al. Thermal imaging as
a non-invasive diagnostic tool for anterior knee pain following implantation of artificial knee joints. Int J Therm
2011; 14(2): 7178.
14. Trafarski A, Rozanski L, Straburzynska-lupa A, et al.
The quality of diagnosis by IR thermography as a function of thermal stimulation in choosing medical applications. In: Proceedings of the 9th international conference
on quantitative infrared thermography, Krakow - Poland,
25 July 2008. Available at: qirt.gel.ulaval.ca/archives/
qirt2008/papers/03_13_17.pdf
15. Ring EF, Collins AJ, Bacon PA, et al. Quantitation of
thermography in arthritis using multi-isothermal analysis.
II. Effect of nonsteroidal anti-inflammatory therapy on the
thermographic index. Ann Rheum Dis 1974; 33(4): 353356.
16. Boas NF. Thermography in rheumatoid arthritis. Ann
New York Acad Sci 1964; 121(1): 223234.
17. Vardasca R. Hand thermogram standardisation with
barycentric warp model. In: Proceedings of the 4th
research student workshop, University of Glamorgan,
South Wales, March 2009, pp.7375.
18. Vardasca R, Ring EFJ, Plassmann P, et al. Thermal symmetry of the upper and lower extremities in healthy subjects. Thermol Int 2012; 22(2): 5360.
19. Collins AJ, Ring EF, Cosh JA, et al. Quantitation of thermography in arthritis using multi-isothermal analysis. I.
The thermographic index. Ann Rheum Dis 1974; 33(2):
113115.
20. Salisbury RS, Parr G, De Silva M, et al. Heat distribution over normal and abnormal joints: thermal pattern
and quantification. Ann Rheum Dis 1983; 42(5): 494499.
21. Bajwa U, Vardasca R, Ring EFJ, et al. Comparison of
boundary detection techniques to improve image analysis
in medical thermography. Imag Sci J 2010; 58(1): 1219.
22. Selvarasu N, Nachiappan A and Nandhitha NM. Euclidean distance based color image segmentation of
abnormality detection from pseudo color thermographs.
Int J Comp Theory Eng 2010; 2(4): 514516.

23. Mishra R, Sharma BL, Gupta R, et al. Indian Rheumatology Association consensus statement on the management of adults with rheumatoid arthritis. Indian J
Rheumatol 2008; 93(3): S1S16.
24. http://www.iact-org.org (accessed on December 2014).
25. Spading SJ, Kwoh CK, Boudreau R, et al. Three dimensional and thermal surface imaging produces reliable
measures of joint shape and temperature: a potential tool
for quantifying arthritis. Arthritis Res Ther 2008; 10(1):
R10.
26. Rovensky J, Clague R and Payer J. Thermographic
index. In: Dictionary of rheumatology. Vienna: Springerverlag, 2009, p. 212. DOI: 10.1007/978-3-211-79280-3_1104.
27. Kanungo T and Mount DM, Netanyahu NS, et al. An
efficient k means clustering algorithm: analysis and implementation. IEEE Trans Pattern Anal Mach Intell 2002;
24(7): 881892.
28. Haralick RM, Shanmugam K and Dinstein I. Texture
features for image classification. IEEE Trans Syst Man
Cybern Syst 1973; 3(6): 610621.
29. Salvatore D and Reagle D. Schaums outline series of theory and problem of statistics and econometrics. New York:
McGraw-Hill, 2002.
30. Girisha AB, Chandrashekhar MC and Kurian MZ. Texture feature extraction of video frames using GLCM. Int
J Eng Trends Tech 2013; 4(6): 27182721.
31. Frize M, Karsh J, Herry C, et al. Preliminary results of
severity of illness measures of rheumatoid arthritis using
infrared imaging. In: Proceedings of the international
workshop on medical measurements and applications,
Cetraro, 2930 May 2009, pp. 187192. New York:
IEEE.
32. Borojevic N, Kolaric D, Grazio S, et al. Thermography
hand temperature distribution in rheumatoid arthritis
and osteoarthritis. Period Biol 2011; 113(4): 445448.
33. Ilowite NT, Walco GA and Pochaczevsky R. Assessment
of pain in patients with juvenile rheumatoid arthritis:
relation between pain intensity and degree of joint inflammation. Ann Rheum Dis 1992; 51: 343346.
34. Zhou Q, Li Z and Aggarwal JK. Boundary extraction in
thermal images by edge map. In: Proceedings of the ACM
symposium on applied computing, Nicosia, 1417 March
2004, pp.254258. New York: ACM.
35. Frize M, Adea C, Payeur P, et al. Detection of rheumatoid arthritis using infrared imaging. In: Proceedings of
the SPIE: medical imaging, vol. 79620M, 10 March 2011.
Bellingham, WA: SPIE.

Downloaded from pih.sagepub.com by guest on May 1, 2015