Rapid vascularization identification using adaptive Gamma correction and support vector machine based on simulated annealing

Microscopic hyperspectral imaging technology of biological material is the forefront of biological spectroscopy study. It is important to make sure whether the dermal substitute transplanted in patient's wounds gets into normal vascularization process when burned or deeply traumatic patients are treated. This is the key to evaluating the quality of repair material and is also an important index of patient's wounds recovery. This paper proposes and realizes a method of rapid vascularization identification based on G-SA-SVM. This method is based on the microscopic hyperspectral imaging. First, the blank correction is used in hyperspectral data. Second, an adaptive Gamma correction model is employed to take advantage of the spectral and spatial features. Finally, simulated annealing is used to optimize the parameters of support vector machine (SA-SVM). SA-SVM is applied to locating the red blood cells effectively and then locating the blood vessels quickly. The experimental results confirm that the proposed method called G-SA-SVM has higher classification accuracy. Hence, it can be applied to evaluating the vascularization process.