Quantitative assessment of Bell's palsy-related facial thermal asymmetry using infrared thermography: A preliminary study

Facial Paralysis Detection in Infrared Thermal Images Using Asymmetry Analysis of Temperature and Texture Features

“…Based on previous studies [ 14 ] and the symptoms of facial paralysis, this study divided facial thermal images into 10 ROIs that were distributed in pairs on the left and right sides on the face, as shown in Figure 2 . Automatic segmentation of facial ROIs is helpful to improve the reliability of thermal asymmetry analysis [ 36 ].…”

“…According to previous research [ 14 , 30 , 31 ], the parameters in the GLCM in this study are as follows: the offset distance Δ is set to 2 and 5, the offset direction θ is set to 0°, 45°, 90°, 135°, and the maximum gray level G is set to 16 and 32. After calculating all 16 GLCMs and based on the given parameters, each GLCM uses 4 s-order statistics to represent the texture features of the thermal image.…”

“…For a healthy person, the temperature distribution on the face is symmetrical [ 13 ]. Previous studies [ 14 ] show that there is a difference in the symmetry of facial temperature distribution between facial paralysis patients and healthy people. Further, facial paralysis can be diagnosed early by measuring the degree of facial thermal asymmetry in patients.…”

“…Infrared thermal imaging (IRT) is a noncontact, nonradiation, and easy-to-use temperature measurement tool [ 15 ]. It involves temperature measurement and thermal asymmetry analysis of the facial regions of interest (ROIs) and has been widely used for diagnosis and assessment of some facial diseases [ 16 , 17 , 18 , 19 ], such as dry eye [ 20 ], temporomandibular disorders [ 21 , 22 ], chronic migraine [ 23 ], and facial paralysis [ 14 , 24 , 25 ]. Existing studies [ 26 , 27 ] used IRT to measure the temperature difference between the left and right sides of the face of facial paralysis patients; this difference was related to the severity of facial paralysis [ 28 , 29 ].…”

“…The temperature difference in facial paralysis patients was significantly greater than that in healthy people [ 12 ]. Moreover, the feasibility of IRT assisted early diagnosis of facial paralysis was evaluated [ 14 ]; the results showed that facial paralysis on thermal images can be preliminarily detected by thermal asymmetry analysis; however, more effective temperature distribution representation is required. Texture features are used to represent the spatial distribution of temperature on thermal images, and they have been used in the detection of diabetic foot ulcers [ 30 ] and breast cancer [ 31 , 32 , 33 ].…”