Endocrine

DOI 10.1007/s12020-012-9645-8

ORIGINAL ARTICLE

Medical thermography: a diagnostic approach for type 2 diabetes

based on non-contact infrared thermal imaging
S. Sivanandam • M. Anburajan • B. Venkatraman •

M. Menaka • D. Sharath

Received: 15 December 2011 / Accepted: 24 February 2012

Ó Springer Science+Business Media, LLC 2012

Abstract To test the potential of Infrared (IR) thermog- and specificity (56 %) was obtained in determining the
raphy in diagnosing as well as predicting type 2 diabetes undiagnosed diabetes with positive predictive value of
and its complications compared with biochemical assay of 65 %, negative predictive value of 85 % and accuracy of
HbA1c as standard. As per American Diabetes Association 73 %. As HbA1c increases, skin temperature decreases.
criteria, threshold for diagnosis of diabetes was set as Skin temperature enables early detection of diabetes as
HbA1c C 6.5 % (7.7 mmol L-1). The total subjects compared to HbA1c. The decrease in skin temperature may
(n = 62) were studied out of which control (n = 32) and be due to the decrease in the basal metabolic rate, poor
diabetic subjects (n = 30). IR camera was used to capture blood perfusion and high insulin resistance. Thermography
the thermal images of the skin for diagnosis of the disease; can be used as a diagnostic as well as prognostic tool for
receiver operating characteristic (ROC) curve was used to the diabetes.
set temperature (°C) as threshold for statistically significant
body regions under t test. In diabetic group, HbA1c showed Keywords Infrared  Thermography  HbA1c  Diabetes 
negative correlation with carotid region (r = -0.471, Skin temperature  Insulin resistance
p \ 0.01) and the mean skin temperature was lower than
the normal group at body regions namely knee
(p = 0.002), tibia (p = 0.003), forehead (p = 0.014), and Introduction
palm (p = 0.019). The palm region showed highest area
under the curve of 0.711 (95 % CI: 0.581–0.842) and the Diabetes mellitus (DM) is a chronic disease characterized
threshold was set as B33.85 °C, thereby sensitivity (90 %) by hyperglycemia with disturbances of carbohydrate, fat
and protein metabolism resulting from defects in insulin
secretion, insulin action, or both [1]. There is a global
predominance of the disease powered by ageing popula-
S. Sivanandam (&)  M. Anburajan
tions, poor diet, and the concurrent epidemic of obesity,
Department of Biomedical Engineering, SRM University,
Kattankulathur 603203, Tamilnadu, India physical inactivity and unhygienic environment [2, 3]. The
e-mail: [email protected]; [email protected] International Diabetes Federation (IDF) [4] estimated the
M. Anburajan total number of people in India with diabetes to be around
e-mail: [email protected]; [email protected] 61.3 million in 2011, rising to 101.2 million by 2030.
The American Diabetes Association (ADA), European
B. Venkatraman  M. Menaka  D. Sharath
Association for the Study of Diabetes (EASD), and IDF
Reactor Safety Engineering Group, Indira Gandhi Centre for
Atomic Research, Kalpakkam 603102, Tamilnadu, India commissioned an international expert committee, which
e-mail: [email protected] recommended that HbA1c should be the preferred diag-
M. Menaka nostic test for diabetes with diagnostic criteria for the
e-mail: [email protected] disease was set as HbA1c C 6.5 % (7.7 mmol L-1) [5, 6].
D. Sharath Temperature is a vital and useful indicator of various
e-mail: [email protected] diseases [7]; there is a resurgence of interest in the
 Endocrine

application of infrared (IR) imaging in medicine with Physical and physiological measurements
improvements in camera technology, which captures the
natural thermal radiation emitted by any object above The anthropometrical measurements namely height (cm),
absolute zero [8, 9]. IR imaging allows the representation weight (Kg), waist and hip (cm) measurements were
of the body surface thermal distribution, where it depends obtained using standardized techniques. The blood pressure
on the complex relationships defining the heat exchange (mmHg) was recorded, in the sitting position, in the right
processes between skin tissue, inner tissue, local vascula- arm using sphygmomanometer.
ture, and metabolic activity [10].
This imaging modality provides functional information Biochemical analysis
not easily measured by any other methods [11], which is
non-invasive, as it passively captures the radiant heat The blood sample collected from an individual was used to
emitted from the human body. Hence, it is risk-free, rela- obtain the various biochemical parameters namely HbA1c,
tively inexpensive and can be readily used as an adjunct to which was determined from the whole blood using
the other standard imaging methods [12]. Complications Olympus auto analyzer, Germany. Estimated average glu-
affecting the lower limbs are the most common manifes- cose (eAG), a derived parameter from HbA1c is measured
tations of diabetes [13, 14]; IR imaging provides the pre- in mg/dl. The total cholesterol and high density lipoprotein
cise level of amputation to be performed and thus prevents (HDL) were obtained by enzymatic method using Olympus
re-amputation as seen in conventional methods. auto analyzer, Germany.
The objectives of the study were: (i) to examine the
discriminatory performance of various risk factors and Thermal measurements and analysis
anthropometric indices between the control and diabetic
group of subjects as segregated by HbA1c and (ii) to The IR thermal imaging procedure does not impose any
evaluate the potential of thermograph in diagnosis as well restrictions on diet, exercise or any other physical activity.
as prognosis of the disease from the study group. When compared with other imaging modalities, the
requirement for thermal imaging is affordable. It requires
an air conditioner of 1.5 ton capacity, waiting and imaging
room with a dimension (150 9 100 ), 5A socket outlet with
Materials and methods 100–240 V AC, 50/60 Hz line frequency, an AC adaptor
with 12 V DC output to recharge the Lithium-ion battery.
Study population The thermal camera can be operated continuously for 4 h
on battery without the need of external power, the imaging
The screening camp for the diagnosis of the disease was room requires no high intensity discharge lamps as thermal
conducted in the month of April 2010 at SRM Medical camera picks up the naturally emitted IR rays from the
College Hospital and Research Centre, Kattankulathur, body, no harmful radiation is involved, no waiting time in
Tamilnadu, India. The institutional ethical clearance was obtaining the report as the captured images can be trans-
obtained in performing the study. The informed consent ferred to the computer through USB (universal serial bus)
form and risk assessment questionnaire was obtained from port, where the report will be generated within few minutes
the total subjects (n = 75) and the questionnaire was after the image capture.
carefully scrutinized for the subjects with confounding In a temperature controlled room, the selected body
factors namely, fever (n = 0), cardiovascular diseases regions to be imaged were disrobed, made free of metallic
(n = 13), and thyroid dysfunction (n = 0), from which the ornaments and other objects to get equilibrated with the
subjects reported with known cardiovascular complications room temperature at about 23 °C for 15 min, where the
were excluded from the study to avoid confounding of the temperature, humidity, and air circulation was kept under
results. Thereby, our study group had n = 62 subjects aged control [15], which allows to capture the skin temperature
between 20 and 80 years with male (n = 27) and female variations in the range of 0.05–0.1 °C from the affected
(n = 35). region to the surrounding body region or the contralateral
The completed questionnaire revealed that 29 % (n = 18) region of the body, this ensures that the naturally emitted
of the study group were reported with known type 2 diabetes thermal radiation from the human body is neither absorbed
with a mean duration of 4 years in taking the medications. nor scattered by any of the objects [16].
Further, 94 % (n = 17) of the known type 2 diabetic subjects The thermal image was captured using ThermaCam
were reported with no physical activity during the leisure (FLIR T400) manufactured by FLIR Systems, Inc., USA.
time or at the work place, who were following a sedentary The camera has Focal Plane Array (FPA), Uncooled mic-
type of lifestyle. robolometer detector, with a thermal sensitivity of 0.05 °C
Endocrine

at 30 °C. The thermal images were captured at a distance curve for the study group was computed to set the optimal
of three feet from the various body regions namely face, threshold in temperature for the statistically significant
neck, hand, and leg regions and the images were analyzed body regions.
using the Quick Report, a thermal software with version
1.2 supported by FLIR T400 camera (www.flir.com). The
software provides the settings for object parameters like Results
emissivity (0.98), reflected apparent temperature (23 °C),
atmospheric temperature (23 °C), relative humidity (50 %), The thermogram obtained from the human subject may
and distance of measure (3 feet), which can be manually vary due to multiple factors such as fever, cardiovascular
entered in accordance with one’s measuring environment. diseases, and thyroid dysfunction. In order to set the
The temperature scale can be set for the various body threshold in temperature for diagnosis of DM, the risk
regions without disturbing the measured temperature, assessment questionnaire was carefully analyzed and the
which was fixed as standard for all the subjects and the subjects with confounding complications were excluded
rainbow palette was chosen to analyze the thermogram in from the analysis to increase the diagnostic accuracy, to
our study. study the severity of the disease and its consequences and
to envisage the disease at the earlier stage.
Measurement Figure 1 illustrates the negative correlation between
HbA1c and skin temperature of the carotid region (r =
The thermal software provides various measurement tools -0.471, p \ 0.01), this can be due the carotid stenosis
namely flying spot meter, spot meter, line and area tool. In become prominent when the blood glucose level increases
order to allow repeatability and to minimize the deviations in an uncontrolled manner leading to cardiovascular
in the measurement of skin temperature, we have chosen diseases.
the area tool for all the regions of interest (ROI) except at The ROC curve for the body regions where illustrated in
fingers and anterior region of the toes, where the line tool Fig. 2, where the curve trajectories of the knee and tibial
was more compatible when compared to an area tool. From regions pass closer to the upper left corner indicates the
head to toe, the skin temperature for the desired ROI was higher accuracy of the test. The vascular disease in the
measured separately for left and right side at the same diabetic subjects can occur at the popliteal and tibial
region approximately. In the measurement panel of the region, at which the atherosclerosis tends to be multiseg-
software, the temperature for the measured regions indicate mental rather than involving in a particular region of the
minimum, maximum, and average value, where the aver- artery wall.
age temperature was noted down for all the subjects con- The thermogram of the palm region of a control and
sidered under study. diabetic subject was illustrated in Fig. 3a, b, respectively.
As per the ADA diagnostic criteria, the threshold for In a control subject, the skin temperature of the palm
DM was set as HbA1c C 6.5 % (7.7 mmol L-1) based on indicates 35.2 °C, this can be due to the normal metabolic
which the subjects were classified into control (n = 32) function of the body and good perfusion of the blood,
and diabetic subjects (n = 30). The completed question- whereas in a diabetic subject, the palm temperature indi-
naire revealed that only 60 % (n = 18) of the diabetic cates 33.0 °C, which was lower when compared to the
subjects were previously diagnosed for the disease whereas control subject, this may be due to poor blood perfusion
40 % (n = 12) of the diseased group were unaware of and decreased metabolism.
being developed into type 2 diabetic individual.

Statistical analysis

A computerized database was created for all the records.

The statistical package for the social sciences (SPSS)
(Version 17.0) was used for the analysis. All the data were
expressed as mean ± SD. The statistical significance was
assumed at p \ 0.05. The independent samples t test for
the clinical, bio-chemical parameters and the skin tem-
perature of the body regions were computed for the entire
study group. The Pearson correlation was studied for the
DM group to obtain the relation between the measured Fig. 1 Pearson correlation between HbA1c and skin temperature of
parameters. The receiver operating characteristic (ROC) the carotid region in the DM group
 Endocrine

thermal distribution of a control subject in which the

measured temperature lies within the mean ± SD of the
control group, whereas Fig. 4b, illustrates the foot tem-
perature of a diabetic subject without the complication of
ulceration, where a decrease in temperature of 2 °C was
observed in comparison with the control subject, this could
be due to the decrease in the basal metabolic rate and
the obstruction of blood flow in the minor blood vessels.
Figure 4c, illustrates the skin temperature of the diabetic
foot with early signs of ulceration, where the average
temperature of the base foot was decreased by 0.5 °C when
compared with the control subject, whereas the toe regions
illustrate higher skin temperature of 3.9 °C and 2.2 °C in
comparison with the right and left foot, respectively. Fur-
ther, the right toe indicated an increase in temperature of
1.7 °C in comparison with contralateral region, which may
Fig. 2 ROC curve for the statistically significant body regions
lead to amputation, if necessary corrective measures are
not taken.
Table 1, describes the Pearson correlation coefficient
for the DM group, in which the age of the diabetic sub-
jects were positively correlated with HbA1c (r = 0.379,
p \ 0.01), this could be due to the sedentary lifestyle and
poor diet. The systolic blood pressure (SBP) (r = 0.555,
p \ 0.01) of the DM group was positively correlated with
age and negatively correlated with skin temperature of the
body regions namely forehead (r = -0.481, p \ 0.01), eye
(r = -0.574, p \ 0.01), and ear (r = -0.403, p \ 0.05),
this could be due to the decrease in the metabolic rate and
endothelial dysfunction, respectively. In contrast, the age
parameter was positively correlated with skin temperature
of the posterior side of the toes (r = 0.390, p \ 0.05),
where this could be the early sign of foot ulceration. The
HbA1c was negatively correlated with skin temperature of
the ear lobe (r = -0.440, p \ 0.05) and the carotid region
(r = -0.470, p \ 0.01), these could be due the athero-
sclerosis of the arteries. There also exists a positive corre-
lation in the skin temperature between the lower extremities
and the upper extremities in the DM group, where most of
them are affected; this could be due to the dependency of
blood perfusion at the extremities.
Table 2 describes the independent samples t test for the
whole study group, in which the level of significance and
the 95 % confidence interval (CI) measures for the clinical,
anthropometrical indices and skin temperature of the body
regions were analyzed. Table 2 indicates that the age
parameter was significant in control (40 ± 12 years) and
DM group (51 ± 15 years) (p = 0.003), but the deviation
Fig. 3 The thermogram of the palm region in a control subject
(a) and diabetic subject (b)
in both the groups were high. The higher range of the
control group may be due to the awareness of the disease
and lower range of the DM group may be due to the family
The thermogram of the foot region was analyzed by history, poor diet and sedentary lifestyle. The waist and the
measuring the mean skin temperature of the metatarsal, hip circumference in the diabetic group was larger than
mid foot and the heel regions. Figure 4a, illustrates the foot control group by 10 and 9 cm (p = 0.010), respectively,
Endocrine

the skin temperature at the knee (p = 0.002, 95 % CI:

0.27–1.10) and tibia region (p = 0.003, 95 % CI: 0.22–
1.01) indicate significance among the study group, this can
be due to the atherosclerosis of the arteries in the lower
limb.
Table 3 describes the area under curve (AUC) for the
statistically significant body regions of the study group.
The skin temperature at the knee, tibia, and palm region
has an AUC of (0.727, 0.722, and 0.711 with 95 % CI:
0.598–0.855, 0.594–0.851, and 0.581–0.842), respectively.
Table 4 presents the statistical measures of performance
of the binary classification test for the study group, where
temperature (°C) was set as a threshold point for the body
regions, by choosing an optimal sensitivity and specificity
from the coordinates of the ROC curve. The palm region
with threshold value B 33.85 °C obtains the sensitivity of
90 %, this indicates the percentage of diabetic subjects who
are correctly identified as diseased and specificity of 56 %
indicates the percentage of healthy subjects identified as
control. Also, positive predictive value (PPV) of 65 %
indicates proportion of subjects with positive test results
who are correctly diagnosed, negative predictive value
(NPV) of 85 % indicates proportion of subjects with neg-
ative test who are correctly diagnosed and an accuracy of
the test for the palm region was obtained as 73 %.

Discussion

This study reported a high prevalence of type 2 DM

(48 %), which was consistent with the study reported by
Ramachandran et al. [17]. Our study was a preliminary
research on the application of thermography in the diag-
nosis and prognosis of the disease.
The negative correlation between HbA1c and the skin
temperature of the carotid region indicate decreased blood
flow in the carotid artery of DM subjects; the thermogram
of the popliteal (knee) and tibial region of the lower limb
indicate a decrease in the skin temperature, these can be
due to atherosclerosis, the effect of high blood sugar on the
cells lining the blood vessels cause increased production of
free radicals, the highly reactive molecules tear the sensi-
tive cell components like deoxy ribo nucleic acid (DNA)
Fig. 4 The thermogram of the foot region in a control subject (a), which lead to apoptosis. This process also reduces the
diabetic subject (b), and diabetic subject with early signs of ulceration availability of nitric oxide (NO) which is responsible for
(c) dilation of blood vessels [18, 19].
The abnormal increase in the temperature of the specific
this can be due to poor diet leading to central obesity and sites of the foot was observed in comparison with the
lack of regular exercises. Among the body regions con- surrounding regions and with the contralateral region of
sidered for IR imaging, there observed a lower skin tem- few of the diabetic individuals, which indicate the early
perature in the DM group at forehead, palm, knee, tibia, stages of ulceration, this could be due to the elevated levels
toes, and foot regions, this can be due to decreased meta- of glycated hemoglobin; the red blood cells (RBC) become
bolic rate. Among the statistically significant body regions, more rigid to pass through the capillaries, thus poor blood
Table 1 Pearson correlation between various variables studied in DM group (n = 30)
Parameters Age HbA1c SBP Body skin temperature (°C)
(Years) (%) (mmHg)
Fore Nose Neck Palm Forearm Finger Finger Knee Tibia Ankle- Toes Toes
head tip whole region toe anterior posterior

Age (Years) 1
HbA1c (%) 0.37** 1
a
SBP (mmHg) 0.55** 0.21 1
Body skin Forehead -0.48** -0.32 -0.43* 1
temperature (°C) Eye -0.57** -0.33 -0.38* 0.75**
Nose 0.02 -0.20 -0.18 0.38* 1
Neck -0.15 -0.25 -0.41* 0.32 0.16
Ear -0.40* -0.44* -0.12 0.57** 0.06
Carotid -0.27 -0.47** -0.29 0.04 0.10 0.23
Palm 0.16 -0.05 -0.20 0.15 0.61** 0.29 1
Forearm 0.08 -0.25 -0.28 0.34 0.61** 0.51** 0.80** 1
Fingers tip 0.27 -0.11 -0.06 0.04 0.58** 0.17 0.92** 0.73** 1
Fingers whole 0.25 -0.09 -0.15 0.07 0.61** 0.23 0.95** 0.78** 0.98** 1
Knee 0.31 0.02 -0.12 0.00 0.37* 0.15 0.46** 0.48** 0.48** 0.54** 1
Tibia region -0.09 0.01 -0.34 0.17 0.34 0.30 0.59** 0.66** 0.51** 0.56** 0.59** 1
Ankle-toe 0.18 0.04 0.01 -0.03 0.38* -0.04 0.57** 0.51** 0.56** 0.57** 0.57** 0.74** 1
Toes anterior 0.34 0.13 0.08 -0.03 0.50** 0.00 0.64** 0.58** 0.62** 0.64** 0.45* 0.48** 0.81** 1
Toes posterior 0.39* 0.22 0.09 -0.00 0.47** 0.09 0.63** 0.58** 0.59** 0.62** 0.47** 0.48** 0.75** 0.95** 1
Foot 0.25 0.15 0.01 0.00 0.54** 0.04 0.61** 0.57** 0.55** 0.59** 0.47** 0.60** 0.82** 0.91** 0.91**
* Correlation is significant at the 0.05 level (2-tailed)
** Correlation is significant at the 0.01 level (2-tailed)
a
Systolic blood pressure
Endocrine
Endocrine

Table 2 Principal characteristics of the study subjects

Parameters Normal (n = 32) mean ± SD Diabetic (n = 30) mean ± SD Sig. (2-tailed) 95 % CI of the measure
Lower bound Upper bound

Age (years) 40 ± 12 51 ± 15 0.003** -18 -4

HbA1c (%) 5.7 ± 0.4 8.5 ± 1.8 0.000*** -3.5 -2.1
a
eAG (mmol L-1) 6.54 ± 0.64 11 ± 2.93 0.000*** -5.52 -3.34
b -1
TC (mmol L ) 4.69 ± 1.07 4.83 ± 0.97 0.616 -0.67 0.41
c
HDL(mmol L-1) 1.04 ± 0.24 1.02 ± 0.18 0.554 -0.08 0.15
d
SBP (mmHg) 123 ± 17 131 ± 16 0.088 -16 1
e
DBP (mmHg) 79 ± 9 84 ± 12 0.155 -9 1
f
BMI (kg m-2) 26 ± 6 28 ± 4 0.157 -4 0.7
g
WC (cm) 87 ± 13 97 ± 13 0.010* -17 -2
h
HC (cm) 96 ± 11 105 ± 11 0.010* -15 -2
Body regions (°C)
Forehead 34.98 ± 0.48 34.70 ± 0.40 0.014* 0.06 0.51
Eye 34.88 ± 0.62 34.72 ± 0.40 0.223 -0.1 0.42
Nose 32.55 ± 2.07 31.94 ± 1.76 0.216 -0.37 1.59
Neck 35.16 ± 0.39 34.98 ± 0.45 0.095 -0.03 0.4
Ear 36.50 ± 0.41 36.38 ± 0.40 0.227 -0.08 0.33
Carotid region 35.10 ± 0.51 34.82 ± 0.97 0.166 -0.12 0.66
Palm 33.64 ± 1.42 32.86 ± 1.09 0.019* 0.13 1.43
Forearm 33.33 ± 2.76 33.39 ± 0.64 0.9 -1.1 0.97
Fingers tip 32.40 ± 2.44 31.38 ± 1.98 0.079 -0.12 2.15
Fingers whole 32.71 ± 2.20 31.76 ± 1.74 0.066 -0.07 1.96
Knee 32.34 ± 0.94 31.66 ± 0.65 0.002** 0.27 1.1
Tibia region 32.82 ± 0.75 32.20 ± 0.81 0.003** 0.22 1.01
Ankle-toe 30.80 ± 1.57 30.23 ± 1.26 0.12 -0.15 1.3
Toes anterior 27.27 ± 2.72 26.12 ± 1.59 0.047* 0.02 2.27
Toes posterior 26.59 ± 2.86 25.05 ± 1.74 0.013* 0.33 2.76
Foot 28.43 ± 2.20 27.34 ± 1.65 0.033* 0.09 2.08
* Correlation is significant at the 0.05 level (2-tailed)
** Correlation is significant at the 0.01 level (2-tailed)
*** Correlation is significant at the 0.001 level (2-tailed)
a
Estimated average glucose
b
Total cholesterol
c
High density lipoprotein
d
Systolic blood pressure
e
Diastolic blood pressure
f
Body mass index
g
Waist circumference
h
Hip Circumference

perfusion lead to starvation of the tissues from the essential The SBP was negatively correlated with forehead, eye
nutrients and oxygen and thus the tissue breakdown, and neck regions, this can be due to increased insulin
leading to ulceration [20]. The ulceration is a condition of resistance resulting in decreased skin temperature as blood
inflammation of the foot region. The typical regions of the flow reduces. This can also be due to endothelial dys-
foot which are more prone to ulceration are the metatarsal function, a condition in which the inner lining of the blood
region, heel and lower digits of the foot [21, 22], these can vessels does not function normally, which develops ath-
be due to the excessive foot pressure resulting from foot erosclerosis [23–25].
deformities like hammer toes, bony prominences and The HbA1c indicates negative correlation with upper
muscle atrophy. extremities, whereas with lower extremities it indicates
 Endocrine

Table 3 Area under curve (AUC) for the statistically significant body regions
Test result variable(s) Area Standard errora Asymptotic sig.b Asymptotic 95 % CI
Lower bound Upper bound

Knee 0.727 0.065 0.002 0.598 0.855

Tibia region 0.722 0.066 0.003 0.594 0.851
Palm 0.711 0.067 0.004 0.581 0.842
Forehead 0.669 0.069 0.022 0.534 0.805
Toes posterior 0.655 0.07 0.036 0.518 0.792
Foot 0.64 0.07 0.058 0.503 0.778
Toes anterior 0.624 0.071 0.092 0.485 0.764
a
Under the nonparametric assumption
b
Null hypothesis: true area = 0.5

Table 4 Binary classification test for the statistically significant body regions
a b
Sl. No Body regions Threshold for diabetes (°C) Sensitivity (%) Specificity (%) PPV (%) NPV (%) Accuracy (%)

1 Palm B33.85 90 56 65 85 73
2 Knee B32.40 86 53 63 81 69
3 Tibia region B32.78 80 62 66 77 71
4 Foot B28.41 76 50 59 69 63
5 Toes anterior B26.83 73 40 53 62 56
6 Forehead B34.93 70 59 61 67 65
7 Toes posterior B26.14 70 50 56 64 60
a
Positive predictive value
b
Negative predictive value

positive correlation. This seems to be anonymous, may this hypertension, dyslipidemia, and elevated fasting glucose
relationship was due to the sedentary lifestyle of the par- [28].
ticipants involved in the study [26]. This can also be due to The merits of the IR imaging procedure are non-inva-
the mechanism of lower limb involved in the ambulation, sive, risk-free; no diet, age invariant, less time, and inex-
where the exercising muscle allows the blood glucose to pensive when compared to other imaging modalities. The
enter the tissue without the need of insulin whereas in the measurement procedure had a demerit, that the temperature
upper extremities insulin resistance predominates due to coefficient was measured by manual selection of ROI,
lack of physical activity. As usage of mixer grinders, which may result in slight deviation in the skin tempera-
washing machines and electric motor for pumping water ture. To overcome this, an advanced image processing
has become part of the modern lifestyle, which reduces the tools may be incorporated to obtain a better result with
upper limb activity and the skin temperature of the upper higher accuracy.
extremities were low in comparison with the lower
extremities, due to poor metabolism.
The body mass index (BMI) of the study population was Conclusion
not statistically significant whereas the waist circumference
and hip circumference indicate significance among the The prevalence of diabetes continues to increase on a global
study group, this indicate that South Indians are prone to level where most of the people were undiagnosed for the
central obesity as compared to the western countries where disease. A non-invasive diagnostic method is necessary to
total obesity is the major risk factor for the disease [27]; eliminate the subject discomfort, to enable early detection
this can be due to insulin resistance syndrome, a group of and to reduce the cost involved. The IR imaging procedure
risk factors that occur together and increase the risk for the provided enormous information about the physiological
disease. The risk factors include, abdominal obesity, processes through examining the skin temperature distribu-
insulin resistance, the sedentary lifestyle of the subjects as tions, which can be related to the core body temperature and
revealed in the risk assessment questionnaire of this study, blood perfusion. The test results indicated better sensitivity,
Endocrine

specificity, accuracy in detecting the disease as compared 12. R.J. Harding, Investigating deep venous thrombosis with infrared
with HbA1c. IR thermograph has achieved the most success imaging. IEEE Eng. Med. Biol. 17(4), 43–46 (1998)
13. F. Al-Maskari, M. El-Sadig, Prevalence of risk factors for dia-
in examining the extremities of the body where the minimal betic foot complications. BMC Fam. Pract. 8, 59 (2007)
temperature changes are more readily detected, because they 14. R.M. Anjana, M.K. Ali, M. Deepa, M. Datta, R. Unnikrishnan,
were separated from the thermal reservoir of the body’s core. M. Rema, V. Mohan, The need for obtaining accurate nationwide
The early signs provided by the thermogram of the foot can estimates of diabetes prevalence in India—Rationale for a
national study on diabetes. Indian J. Med. Res. 133, 369–380
be used as a valid prognostic tool in detecting the ulcera- (2011)
tion among the individuals who were being developed into 15. S. Bagavathiappan, T. Saravanan, J. Philip, T. Jayakumar, Baldev
diabetic subject and individual who were in the earlier stages Raj, R. Karunanithi et al., Investigation of peripheral vascular
of the disease, if unchecked, which may lead to amputation disorders using thermal imaging. Br. J. Diabetes. Vasc. Dis. 8(2),
102–104 (2008)
and other series of complications may result in the near 16. E.F.J. Ring, K. Ammer, The technique of infrared imaging in
future. The thermograph can be used as a diagnostic and medicine. Thermol. Int. 10(1), 7–14 (2000)
predictive tool for the DM and which is the need of the hour 17. A. Ramachandran, A.K. Das, S.R. Joshi, S. Shah, K.M. Prasanna
to control the severity of the disease globally. Kumar, Current status of diabetes in India and need for novel
therapeutic agents. J. Assoc. Physicians India 58, 7–9 (2010)
18. M.S. Amer, M.M. Maher, O.H. Omar, R.A. Reda, A.E. Elawam,
Acknowledgments We were grateful for the thermal camera sup- H.S. Sweed, Carotid intima-media thickness can predict coronary
port, kindly provided by the Indira Gandhi Centre for Atomic atherosclerosis in diabetic elderly patients. Eur. J. Gen. Med.
Research (IGCAR), Kalpakkam, Tamilnadu, India. Also, we thank 7(3), 245–249 (2010)
the management of SRM University, Kattankulathur, Tamilnadu, 19. T. Yoshimura, E. Suzuki, K. Egawa, Y. Nishio, H. Maegawa, S.
India, for conducting the cost free screening camp for the welfare and Morikawa et al., Low blood flow estimates in lower-leg arteries
the betterment of the society. predict cardiovascular events in Japanese patients with type 2
diabetes with normal ankle-brachial indexes. Diabetes Care 29,
Conflict of interest The authors declare that there is no conflict of 1884–1890 (2006)
interest associated with this manuscript. 20. S. Shin, Y. Ku, N. Babu, M. Singh, Erythrocyte deformability and
its variation in diabetes mellitus. Indian J. Exp. Biol. 45, 121–128
(2007)
21. K. Roback, An overview of temperature monitoring devices for
References early detection of diabetic ulcers. Expert Rev. Med. Devices 7(5),
711–718 (2010)
22. M. Bharara, J.E. Cobb, D.J. Claremont, Thermography and
1. American Diabetes Association, Diagnosis and classification of
thermometry in the assessment of diabetic neuropathic foot: a
diabetes mellitus. Diabetes Care 33(1), S62–S69 (2010)
case for furthering the role of thermal techniques. Int. J. Extrem
2. A. Boutayeb, E.H. Twizell, K. Achouayb, K. Chetouani, A
Wounds 5(4), 250–260 (2006)
mathematical model for the burden of diabetes and its compli-
23. A.D. Baron, G. Brechtel-Hook, A. Johnson, D. Hardind, Skeletal
cations. Biomed. Eng. Online 3, 20 (2004)
muscle blood flow. A possible link between insulin resistance and
3. B.K. Barbara, S.F. Jeffrey, Obesity and insulin resistance. J. Clin.
blood pressure. Hypertension 21(2), 129–135 (1993)
Invest. 106, 473–481 (2000)
24. M. Juonala, J.S.A. Viikari, T. Ronnemaa, H. Helenius, L.
4. International Diabetes Federation, The global burden (2011),
Taittonen, O.T. Raitakari, Elevated blood pressure in adolescent
Available from, http://www.idf.org/diabetesatlas/5e/the-global-
boys predicts endothelial dysfunction. Hypertension 48, 424–430
burden. Accessed 14 Dec 2011
(2006)
5. American Diabetes Association, Standards of medical care in
25. Y. Chen, Y. Huang, X. Li, M. Xu, Y. Bi, Y. Zhang et al.,
diabetes. Diabetes Care 34(1), S11–S61 (2011)
Association of arterial stiffness with HbA1c in 1,000 type 2 dia-
6. V. Mohan, V. Vijayachandrika, V. Gokulakrishnan, R.M. Anjana,
betic patients with or without hypertension. Endocrine 36,
A. Ganesan, M.B. Weber et al., A1c cut points to define various
262–267 (2009)
glucose intolerance groups in Asian Indians. Diabetes Care 33(3),
26. D. Umpierre, P. Ribiero, C.K. Cramer, C.B. Leitao, T.N. Zucatti,
515–519 (2010)
M.J. Azevedo et al., Physical activity advice only or structured
7. E.F.J. Ring, Thermal imaging today and its relevance to diabetes.
exercise training and association with HbA1c levels in type 2
J. Diabetes Sci. Technol. 4(4), 857–862 (2010)
diabetes. JAMA 305(17), 1790–1799 (2011)
8. B.F. Jones, P. Plassmann, Digital infrared thermal imaging of
27. M.P. Bahillo-Curieses, F. Hermoso-Lopez, M.J. Martinez-
human skin. IEEE Eng. Med. Biol. 21(6), 41–48 (2002)
Sopena, P. Cobreros-Garcia, P. Garcia-Saeta, M. Triguez-Garcia
9. YuV Gulyaev, A.G. Markov, G. Koreneva, P.V. Zakharov,
et al., Prevalence of insulin resistance and impaired glucose tol-
Dynamic infrared thermography in humans. IEEE Eng. Med.
erance in a sample of obese Spanish children and adolescents.
Biol. 14(6), 766–771 (1995)
Endocrine (2011). doi:10.1007/s12020-011-9540-8
10. N. Kakuta, S. Yokoyama, K. Mabuchi, Human thermal models
28. F. Amati, J.J. Dube, P.M. Coen, M. Stefanovic-Racic, F.G.S.
for evaluating infrared images. IEEE Eng. Med. Biol. 21(6),
Toledo, B.H. Goodpaster, Physical inactivity and obesity underlie
65–72 (2002)
the insulin resistance of aging. Diabetes Care 32(8), 1547–1549
11. N.A. Diakides, Infrared Imaging: an emerging technology in
(2009)
medicine. IEEE Eng. Med. Biol. 17(4), 17–18 (1998)