RESUMO
Traditional computed tomography (CT) delivers a relatively high dose of radiation to the patient and cannot be used as a method for screening of pathologies. Instead, infrared thermography (IRT) might help in the detection of pathologies, but interpreting thermal imaging (TI) is difficult even for the expert. The main objective of this work is to present a new, automated IRT method capable to discern the absence or presence of tumor in the orofacial/maxillofacial region of patients. We evaluated the use of a special feature vector extracted from face and mouth cavity thermograms in classifying TIs against the absence/presence of tumor (n = 23 patients per group). Eight statistical features extracted from TI were used in a k-nearest neighbor (kNN) classifier. Classification accuracy of kNN was evaluated by CT, and by creating a vector with the true class labels for TIs. The presented algorithm, constructed from a training data set, gives good results of classification accuracy of kNN: sensitivity of 77.9%, specificity of 94.9%, and accuracy of 94.1%. The new algorithm exhibited almost the same accuracy in detecting the absence/presence of tumor as CT, and is a proof-of-principle that IRT could be useful as an additional reliable screening tool for detecting orofacial/maxillofacial tumors.
Assuntos
Algoritmos , Neoplasias , Análise por Conglomerados , Humanos , Termografia , Tomografia Computadorizada por Raios XRESUMO
OBJECTIVES: To evaluate thermal images (TIs) by using an algorithm for optimized region of interest (ROI) and image segmentation, in order to find zones of the facial skin surface with asymmetrical temperature, and to test consistency with CT findings, to detect maxillofacial pathologies (i.e. tumours). METHODS: The following steps for the TI evaluation were applied: data acquisition/pre-processing of frontal face and mouth projection, detection of face and mouth external contour, finding face and mouth symmetry axis, calculation of differences in average and maximal temperatures between left and right face and mouth sides, image segmentation of the selected ROI, and evaluation of diagnostic accuracy by comparing the TI results with CT findings. RESULTS: In healthy subjects, the average temperature difference between left/right sides of facial and mouth ROI was negligible (0.02 ± 0.21 °C and 0.05 ± 0.19 °C, respectively; n = 23). In the presence of tumour, the average temperature difference was higher in corresponding TIs (0.47 ± 0.1 °C and 0.66 ± 0.1 °C, for facial and mouth ROI, respectively; n = 19, p < 0.05). For large tumours, thermal asymmetry in the corresponding TI is easily detected, and image segmentation is optional for finding the affected zone. For small or deeply localized tumours, segmentation of the mouth cavity of the ROI was required for the detection of hot and cold spots. CONCLUSIONS: Asymmetrical temperature zones and their location as detected from thermal images coincide well with the presence and localization of maxillofacial pathologies (i.e. tumours) established by CT. However, accurate information could often be obtained only after application of image segmentation algorithm to the selected ROI.