(PDF) Establishing Differences in Thermographic Patterns between the Various Complications in Diabetic Foot Disease

Hindawi International Journal of Endocrinology Volume 2018, Article ID 9808295, 7 pages https://doi.org/10.1155/2018/9808295 Research Article Establishing Differences in Thermographic Patterns between the Various Complications in Diabetic Foot Disease Alfred Gatt ,1 Owen Falzon ,2 Kevin Cassar,3 Christian Ellul,1 Kenneth P. Camilleri ,2,4 Jean Gauci,2 Stephen Mizzi ,1 Anabelle Mizzi,1 Cassandra Sturgeon,3 Liberato Camilleri,5 Nachiappan Chockalingam ,6 and Cynthia Formosa 1 1 Faculty of Health Sciences, University of Malta, Msida, Malta 2 Centre for Biomedical Cybernetics, Faculty of Engineering, University of Malta, Msida, Malta 3 Faculty of Medicine and Surgery, University of Malta, Msida, Malta 4 Department of Systems & Control Engineering, University of Malta, Msida, Malta 5 Department of Statistics and Operations Research, Faculty of Science, University of Malta, Msida, Malta 6 Faculty of Health Science, Staffordshire University, Stoke-on-Trent, UK Correspondence should be addressed to Alfred Gatt;

Received 29 September 2017; Accepted 3 January 2018; Published 12 March 2018 Academic Editor: Nikolaos Papanas Copyright © 2018 Alfred Gatt 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. Aim. To evaluate the potential of thermography as an assessment tool for the detection of foot complications by understanding the variations in temperature that occur in type 2 diabetes mellitus (DM). Methods. Participants were categorized according to a medical examination, ankle brachial index, doppler waveform analysis, and 10-gram monofilament testing into five groups: healthy adult, DM with no complications, DM with peripheral neuropathy, DM with neuroischaemia, and DM with peripheral arterial disease (PAD) groups. Thermographic imaging of the toes and forefeet was performed. Results. 43 neuroischaemic feet, 41 neuropathic feet, 58 PAD feet, 21 DM feet without complications, and 126 healthy feet were analyzed. The temperatures of the feet and toes were significantly higher in the complications group when compared to the healthy adult and DM healthy groups. The higher the temperatures of the foot in DM, the higher the probability that it is affected by neuropathy, neuroischaemia, or PAD. Conclusions. Significant differences in mean temperatures exist between participants who were healthy and those with DM with no known complications when compared to participants with neuroischaemia, neuropathy, or PAD. As foot temperature rises, so does the probability of the presence of complications of neuropathy, neuroischaemia, or peripheral arterial disease. 1. Introduction early stage of the disease. Prophylactic foot care has been shown to decrease patient morbidity [1]. The diabetic foot is characterized by the presence of various Common diabetic foot complications such as ischae- complications that typically tend to develop as a result of mia and neuropathy have an effect on the temperature poor glycaemic control. The main complications include of the foot [2]. It is postulated that changes in tempera- neuropathy, peripheral arterial disease (PAD), and neu- ture of the foot may be indicative of the presence of roischaemia, amongst others. These complications are such complications. amongst the most serious and costly as they often lead to Thermography or medical infrared imaging may be used amputation. In many cases, development of diabetic foot to detect temperature changes. This technique is deemed safe complications can be avoided or substantially delayed with since it is noncontact, noninvasive, and nonirradiant and has timely assessment, diagnosis, and treatment provided at an been utilized in a number of medical applications including 2 International Journal of Endocrinology imaging of the breast [3], skin [4] and foot vascular compli- Participants who presented with active ulceration or cations, and ulceration in diabetes [5–7]. other significant comorbidities that could affect the distribu- The current clinical practice of temperature assessment is tion of thermographic patterns, such as rheumatoid arthritis mainly by manual palpation of foot temperature. However, a or Raynaud’s phenomenon, were excluded. gradual increase in foot temperature may be too subtle to be Following a 15-minute acclimatization period, all partic- detected only by the hand, making the timely and early ipants rested in a supine position on a couch in a room which detection of underlying pathology difficult. It has been was temperature controlled at 23°C. established that increased temperatures in the foot may be An ABPI was obtained [14, 15], according to standard present up to a week before a foot ulcer occurs [8], thus clinical practice utilizing a Huntleigh (Cardiff, Wales) Dop- making it important for any variations in temperature to be plex Assist. A cuff was applied proximal to the ankle, and detected promptly. an 8 Mhz doppler probe was applied at the posterior tibial Thermal imaging offers an excellent means of making a artery and the dorsalis paedis artery. The probe was held at quantitative determination of surface temperature and can an angle of 45° against blood flow while the cuff was inflated offer an alternative mode of detection of major foot compli- until the doppler signal was cut off. Then the cuff pressure cations since it has been reported that monitoring of skin was slowly released. Once the signal was reobtained, the temperature reduces the risk of diabetic foot ulceration in systolic pressure of the particular artery was noted. high-risk patients [9]. This process was repeated for the brachial artery; the cuff Although Nagase et al. had looked at variations in plantar was applied above the elbow and the doppler probe was held thermographic patterns in normal controls and nonulcer dia- in order to obtain its systolic pressure. The ABPI was calcu- betic patients, it has not been fully elucidated to what extent lated with the higher of the posterior tibial and dorsalis pae- the individual variations of the plantar thermographic pat- dis pressures being taken into consideration. Normal ABPI terns show different trends between these two cohorts [10]. values ranged from 0.9 to 1.3. Presently, little is known about the range of abnormal Spectral Doppler waveform analysis was also employed thermoregulation in those patients with diabetes present- to classify the recorded waveform as being triphasic, biphasic, ing both for screening and management [8]. The aim of or monophasic [16]. A triphasic waveform is indicative of this study was to evaluate the potential of thermography normal arterial perfusion, whilst the other two classifications as an assessment tool for the detection of type 2 diabetic are indicative of PAD, with the monophasic waveform foot disease by assessing the variations in temperature that denoting a more severe form of the condition. Only those occur in the diabetic foot before ulceration appears. Objec- participants with monophasic waveforms and an ABPI < 0.6 tives were to find a possible correlation between temperature were included to ensure an unequivocal diagnosis of PAD. readings of the plantar foot and toes in diabetic foot compli- Testing for neuropathy involved the use of a 10 g Semmes cations to reduce the risk of foot ulceration by early detection Weinstein monofilament administered at 10 sites on each of pathologies. foot. In this validated method, exactly 10 g of force was applied before the monofilament bent, thus ensuring that exactly the same amount of force is applied. Neuropathy 2. Method was diagnosed if at least one site was not felt by the partici- pant. All the above measurements were carried out by the This study employed medical infrared imaging to visualize same experienced clinician in order to ensure consistency. the temperature distribution of the feet of participants living with DM with or without complications and healthy 3. Image Acquisition, Segmentation, Data controls. Following ethical approval from the University Extraction, and Analysis Ethics Committee, initially, healthy adult participants were recruited, medically examined, and imaged as reported else- A FLIR SC7200 infrared camera with a spatial resolution of where [11] whilst participants with type 2 diabetes mellitus 320 × 256 pixels and a temperature resolution of 20 mK was were recruited from the patient list of a vascular surgeon. used for the acquisition of thermal images. The protocol for All participants underwent a thorough clinical examination obtaining thermal images followed the recommendations of that included validated tests for neuropathy [12] and periph- the American Academy of Thermology [17]. The camera eral arterial disease [13]. was placed on a tripod 1.5 m from the subject and perpen- Participants with DM were categorized into 4 groups dicular to the body part that was being photographed [11]. based on the medical examination and testing: a heathy Images of the plantar aspect of the feet were recorded for DM group (i.e., presenting with DM but no significant later analysis. medical comorbidities and/or complications), a PAD group Thermal images obtained of the feet were divided into (presenting with ABPI < 0.6 and monophasic Doppler regions so that temperature data could be extracted spectral waveforms at the ankles, but no neuropathy), a neu- (Figures 1(a) and 1(b)) [11]. ropathic group (presenting with positive 10-gram monofila- ment at any one of 10 tested sites and/or reduced vibration 4. Results perception threshold as measured with a tuning fork and an ABPI between 0.9 and 1.3), and a neuroischaemic group Thermographic images from 43 neuroischaemic limbs (from (presenting with ABPI < 0.9 and neuropathy). 30 subjects), 41 neuropathic limbs (from 32 subjects), International Journal of Endocrinology 3 2 1 6 7 3 8 4 9 5 11 19 17 10 (a) (b) Figure 1: (a) Diagram showing the foot regions considered for temperature extraction. (b) An actual thermal image and the corresponding regions of interest. The temperatures from the toe regions and forefoot regions were considered for further analysis. Figure 2: Toe temperature distribution. Figure 3: Forefoot temperature distribution. 58 PAD limbs (from 42 subjects), 21 DM healthy limbs When analyzing the mean temperature data in all 5 toes (15 subjects,), and 126 healthy limbs (from 63 subjects) and 3 plantar regions of the forefoot (medial, central, and were analyzed. lateral regions) (Figures 2 and 3), there are significant 4 International Journal of Endocrinology Table 1: Tukey test to compare mean temperatures between the two groups. Sample size Mean Std. deviation p value Neuropathic, neuroischaemic, and PAD 121 27.83 2.637 Toe 1 0.000 Healthy and DM healthy 123 25.52 3.473 Neuropathic, neuroischaemic, and PAD 128 27.38 2.899 Toe 2 0.000 Healthy and DM healthy 122 25.12 3.452 Neuropathic, neuroischaemic, and PAD 131 27.33 2.886 Toe 3 0.000 Healthy and DM healthy 123 25.19 3.476 Neuropathic, neuroischaemic, and PAD 125 27.27 2.926 Toe 4 0.000 Healthy and DM healthy 123 25.21 3.317 Neuropathic, neuroischaemic, and PAD 125 27.29 2.852 Toe 5 0.000 Healthy and DM healthy 123 25.38 3.238 Neuropathic, neuroischaemic, and PAD 135 28.67 2.454 Medial forefoot 0.000 Healthy and DM healthy 123 26.88 2.874 Neuropathic, neuroischaemic, and PAD 134 28.35 2.474 Central forefoot 0.000 Healthy and DM healthy 123 26.68 2.835 Neuropathic, neuroischaemic, and PAD 130 28.58 2.374 Lateral forefoot 0.000 Healthy and DM healthy 123 26.80 2.861 differences in mean temperatures between the five groups of Table 2: Likelihood ratio tests (model 1). patients, as demonstrated by the Tukey test (Table 1). This test clusters these five groups into two groups where the Likelihood ratio tests mean temperatures of diabetic participants with peripheral Model fitting criteria Likelihood ratio tests Effect arterial disease, diabetes patients with neuropathy, and dia- −2 log likelihood Chi-square df p value betes patients with neuroischaemia are significantly higher Intercept 1495.659 0.000 0 . than the mean temperatures of healthy adults and diabetes Temperature 1634.676 139.02 1 0.000 patients with no known complication. Thus, for further Toe location 1496.714 1.055 4 0.901 statistical comparison, these five categories were divided into a “healthy group” (comprised of healthy adults and DM participants with no known complications) and a “complica- regression model that yields the probability that a patient tions group” (comprised of neuropathic, neuroischaemic, has neuropathy, neuroischaemia, or PAD given the toe tem- and PAD participants). perature is given by p 5. Logistic Regression Analysis loge = −5 786 + 0 220 temperature, 1 1−p In the first logistic regression model fit, we relate the health status (neuropathic, neuroischaemic, or PAD; healthy or where p is the probability that the participant has neurop- DM healthy) to two predictors, which include toe tempera- athy, neuroischaemia, or PAD and 1 − p is the probability ture and toe location. As indicated in Table 2, a binomial that the patient is healthy. The probability curves dis- distribution is assumed since the dependent variable has played in Figure 4 clearly show that the likelihood of neu- two categories, while a logit link function is used since this ropathy, neuroischaemia, or PAD increases as the toe is the canonical link for the binomial distribution. temperature increases. Toe location was not found to be a significant predictor In the second logistic regression model fit, we relate since the p value (0.901) exceeds the 0.05 level of significance. the health status (neuropathic, neuroischaemic, or PAD; A backward procedure was used to fit the parsimonious model, healthy or DM healthy) to two predictors, which include which identified temperature as a sole significant predictor. temperature and forefoot location, whether medial, central, As shown in Table 3, the regression coefficient of temper- or lateral. A binomial distribution and a logit link function ature (0.220) is positive indicating that the toe temperature of are again assumed. neuropathic, neuroischaemic, or PAD participants is Plantar location was not found to be a significant predic- expected to be higher than that of healthy or DM healthy tor since the p value (0.912) exceeds the 0.05 level of patients. The odds ratio indicates that the odds that the significance. A backward procedure was used to fit the parsi- participant has neuropathy, neuroischaemia, or PAD rather monious model, which identified temperature as a sole than being healthy increases by 24.7% for every 1°C increase significant predictor (Table 4). in toe temperature. This odds ratio ranges from 19.8% to The results in Table 5 indicate that the regression coeffi- 29.7% assuming a 95% confidence level. The logistic cient for the plantar forefoot temperature (0.254) is positive International Journal of Endocrinology 5 Table 3: Parameter estimates of the logistic regression model for toe temperatures. 95% CI for odds ratio Effect B Std. error Wald p value Odds ratio Lower bound Upper bound Intercept −5.786 0.534 117.363 0.000 Temperature 0.220 0.020 119.693 0.000 1.247 1.198 1.297 the patient is healthy. The probability curves displayed in 1.0 Figure 5 clearly show that the likelihood of neuropathy, 0.9 neuroischaemia, or PAD increases as the plantar forefoot 0.8 temperature increases. 0.7 0.6 6. Discussion 0.5 The results of this thermographic study demonstrate three 0.4 main inferences: (i) that there are no significant differences 0.3 in mean temperatures of the toes and forefoot between 0.2 healthy subjects and patients with diabetes showing no com- 0.1 plications; (ii) that there are no significant differences in mean temperatures of these same areas between participants 0.0 with complications of neuropathy, neuroischaemia, and 25 26 27 28 29 30 31 32 33 34 35 PAD; (iii) that there are significantly higher mean tempera- Temperature (degree celsius) tures in these latter group of subjects when compared to both Probability that a patient healthy and DM participants with no complications. has neuroischaemia, neuropathy, or PAD This increase in temperature is further confirmed by the Probability that a patient logistic regression models of both toe and forefoot areas, is healthy which establish temperature as being the sole significant pre- dictor of complications. These models demonstrate that the Figure 4: Logistical regression curves of toe temperatures. probability of complications of PAD, neuropathy, and/or neuroischaemia being present increases as the temperature of these regions rises. Table 4: Likelihood ratio tests (model 2). This study is the first of its kind to report temperature dif- Likelihood ratio tests ferences between possible categories of complications of DM Model fitting criteria Likelihood ratio tests relative to healthy adults. The authors recommend that these Effect findings and thermographic techniques should be considered −2 log likelihood Chi-square df p value Intercept 924.056 0.000 0 . for further clinical investigations of the DM patient. These results imply that should a rise in temperature be detected Temperature 1004.520 80.464 1 0.000 in the diabetic foot, there is a higher likelihood that diabetic Plantar forefoot foot complications have set in, as further reported by Sun 924.240 0.184 2 0.912 location et al. [8] who state that thermographic patterns may change as early as one week prior to ulceration. The findings of the study indicate that an increase in temperature may not indicating that the plantar forefoot temperature of neuro- necessarily imply impending ulceration, but simply the pathic, neuroischaemic, or PAD patients is expected to development of peripheral neuropathy, ischaemia, or both. be higher than that of healthy or DM healthy patients. The Further research is warranted to establish whether the odds ratio indicates that the odds that the participant has inclusion of thermography into screening protocols could neuropathy, neuroischaemia, or PAD rather than being help detect the development of diabetic foot complications healthy increases by 28.9% for every 1°C increase in plantar earlier so that appropriate prompt preventative measures temperature. This odds ratio ranges from 19.8% to 29.7% may be taken to avoid unnecessary complications. assuming a 95% confidence level. The logistic regression Whilst neuropathic feet have been previously reported as model that yields the probability that a patient has neuropa- being warmer than healthy feet, we can now confirm that thy, neuroischaemia, or PAD given the plantar temperature even neuroischaemic and ischaemic feet exhibit the same is given by trend. This may be due to altered thermoregulatory mecha- p nisms of the feet, which can be affected by both neuropathy loge = −6 949 + 0 254 temperature, 2 and PAD. Local ischaemia may lead to disruption of sympa- 1−p thetically mediated noradrenergic vasoconstriction which where p is the probability that the patient has neuropathy, leads to increased flow to the cutaneous vessels rather than neuroischaemia, or PAD and 1 − p is the probability that through the deeper nutritive vessels which in turn leads to 6 International Journal of Endocrinology Table 5: Parameter estimates of the logistic regression model for forefoot temperatures. 95% CI for odds ratio Effect B Std. error Wald p value Odds ratio Lower bound Upper bound Intercept −6.949 0.846 67.395 0.000 Temperature 0.254 0.030 69.481 0.000 1.289 1.214 1.368 1.0 Conflicts of Interest 0.9 The authors declare no conflict of interest. 0.8 0.7 Acknowledgments 0.6 The authors would like to thank all the participants 0.5 who consented to take part in this trial. This research 0.4 project was financed by the Malta Council for Science and 0.3 Technology through the National Research & Innovation Programme 2013. 0.2 0.1 References 0.0 [1] M. S. Pinzur, M. P. Slovenkai, E. Trepman, N. N. Shields, and 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Diabetes Committee of American Orthopaedic Foot and Temperature (degree celsius) Ankle Society, “Guidelines for diabetic foot care: recommen- Probability that a patient dations endorsed by the Diabetes Committee of the American has neuroischaemia, neuropathy, or PAD Orthopaedic Foot and Ankle Society,” Foot & Ankle Interna- Probability that a patient tional, vol. 26, no. 1, pp. 113–119, 2005. is healthy [2] S. Bagavathiappan, J. Philip, T. Jayakumar et al., “Correlation between plantar foot temperature and diabetic neuropathy: a Figure 5: Logistical regression curves of forefoot temperatures. case study by using an infrared thermal imaging technique,” Journal of Diabetes Science and Technology, vol. 4, no. 6, pp. 1386–1392, 2010. higher heat emissivity. Arteriovenous anastamoses (AVA) [3] B. B. Lahiri, S. Bagavathiappan, T. Jayakumar, and J. Philip, which are thick-walled, low resistance conduits allow high- “Medical applications of infrared thermography: a review,” flow rates directly from arterioles to venules. AVA are Infrared Physics & Technology, vol. 55, no. 4, pp. 221–235, numerous and richly innervated by sympathetic vasocon- 2012. strictor nerves. Substantial changes occur in blood flow [4] K. Otsuka, S. Okada, M. Hassan, and T. Togawa, “Imaging of depending on whether AVAs are closed or open [18]. skin thermal properties with estimation of ambient radiation It is reported that the application of clinical examina- temperature,” IEEE Engineering in Medicine and Biology tion or nerve conduction studies alone is not adequate in Magazine, vol. 21, no. 6, pp. 49–55, 2002. screening diabetic at-risk feet at early stage [8]; thus, the [5] H. Wang, D. R. Wade Jr., and J. Kam, “IR imaging of blood use of an adjunct method such as thermography may circulation of patients with vascular disease,” Proc. SPIE, D. prove useful. D. Burleigh, K. E. Cramer and G. R. Peacock, Eds., vol. 5405, Further research into thermographic patterns of patients pp. 115–123, 2004. with diabetes and with active ulcers may help elucidate the [6] E. Ring and K. Ammer, “The technique of infrared imaging natural history of development of ulceration. in medicine,” Thermology International, vol. 10, pp. 7–14, 2000. [7] D. G. Armstrong and L. A. Lavery, “Monitoring healing of acute Charcot’s arthropathy with infrared dermal thermome- 7. Conclusions try,” Journal of Rehabilitation Research and Development, vol. 34, no. 3, pp. 317–321, 1997. This study has confirmed that the mean temperatures of [8] P.-C. Sun, H.-D. Lin, S.-H. E. Jao, Y.-C. Ku, R.-C. Chan, the toes and forefeet of the complications group exhibit and C.-K. Cheng, “Relationship of skin temperature to significantly higher temperatures than those of the healthy sympathetic dysfunction in diabetic at-risk feet,” Diabetes group, whilst each group presents with comparable tem- Research and Clinical Practice, vol. 73, no. 1, pp. 41–46, peratures within themselves. These results indicate that 2006. thermography demonstrates potential as a screening or [9] D. G. Armstrong, K. Holtz-Neiderer, C. Wendel, M. J. Mohler, clinical investigation tool, although more research in the H. R. Kimbriel, and L. A. Lavery, “Skin temperature monitor- area is warranted. ing reduces the risk for diabetic foot ulceration in high-risk International Journal of Endocrinology 7 patients,” The American Journal of Medicine, vol. 120, no. 12, pp. 1042–1046, 2007. [10] T. Nagase, H. Sanada, K. Takehara et al., “Variations of plantar thermographic patterns in normal controls and non-ulcer dia- betic patients: novel classification using angiosome concept,” Journal of Plastic, Reconstructive & Aesthetic Surgery, vol. 64, no. 7, pp. 860–866, 2011. [11] A. Gatt, C. Formosa, K. Cassar et al., “Thermographic patterns of the upper and lower limbs: baseline data,” International Journal of Vascular Medicine, vol. 2015, Article ID 831369, 9 pages, 2015. [12] S. Baraz, K. Zarea, H. B. Shahbazian, and S. M. Latifi, “Com- parison of the accuracy of monofilament testing at various points of feet in peripheral diabetic neuropathy screening,” Journal of Diabetes and Metabolic Disorders, vol. 13, no. 1, p. 19, 2014. [13] C. Formosa, A. Gatt, and N. Chockalingam, “Screening for peripheral vascular disease in patients with type 2 diabetes in Malta in a primary care setting,” Quality in Primary Care, vol. 20, no. 6, pp. 409–414, 2013. [14] C. Formosa, K. Cassar, A. Gatt et al., “Hidden dangers revealed by misdiagnosed peripheral arterial disease using ABPI mea- surement,” Diabetes Research and Clinical Practice, vol. 102, no. 2, pp. 112–116, 2013. [15] A. T. Hirsch, Z. J. Haskal, N. R. Hertzer et al., “ACC/AHA 2005 practice guidelines for the management of patients with peripheral arterial disease (lower extremity, renal, mesenteric, and abdominal aortic),” Circulation, vol. 113, no. 11, pp. e463–e465, 2006. [16] L. Norgren, W. R. Hiatt, J. A. Dormandy et al., “Inter-society consensus for the management of peripheral arterial disease (TASC II),” International Angiology, vol. 45, no. 1, pp. S5–S67, 2007. [17] M. Bharara, J. E. Cobb, and D. J. Claremont, “Thermography and thermometry in the assessment of diabetic neuropathic foot: a case for furthering the role of thermal techniques,” The International Journal of Lower Extremity Wounds, vol. 5, no. 4, pp. 250–260, 2006. [18] K. Lossius, M. Eriksen, and L. Walløe, “Fluctuations in blood flow to acral skin in humans: connection with heart rate and blood pressure variability,” The Journal of Physiology, vol. 460, no. 1, pp. 641–655, 1993. MEDIATORS of INFLAMMATION The Scientific Gastroenterology Journal of World Journal Hindawi Publishing Corporation Research and Practice Hindawi Hindawi Diabetes Research Hindawi Disease Markers Hindawi www.hindawi.com Volume 2018 http://www.hindawi.com www.hindawi.com Volume 2018 2013 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Journal of International Journal of Immunology Research Hindawi Endocrinology Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Submit your manuscripts at www.hindawi.com BioMed PPAR Research Hindawi Research International Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Journal of Obesity Evidence-Based Journal of Stem Cells Complementary and Journal of Ophthalmology Hindawi International Hindawi Alternative Medicine Hindawi Hindawi Oncology Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2013 Parkinson’s Disease Computational and Mathematical Methods in Medicine Behavioural Neurology AIDS Research and Treatment Oxidative Medicine and Cellular Longevity Hindawi Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018