Spatially resolved laser doppler based diffused beam reflectance measurement of blood glucose diagnosis in real time by noninvasive technique

A method provides for the design of Fast Haar wavelet with the supervised neural network for diagnosis of diabetes mellitus with nearer accuracy with in short time application non-invasively is proposed. Our proposed work consists of two parts. First, analysis bank of haar wavelet is modified by using polyphase structure. Second, supervised back propagation neural network in the diagnosis of diabetes mellitus with the output of fast haar wavelet as its input is formulated. Finally, a fast haar wavelet based neural network is designed and it satisfies alias free and gives perfect nearer blood glucose value non-invasively which correlates the value of 99.69% of clinical diagnosis. Number of arithmetic calculations (convolution and addition) is reduced in fast haar wavelet transform with neural network. This makes the less computational complexity and less power consumption by reducing in time. Wavelet based supervised back propagation network is used for diagnosing blood glucose which extract from the blood flow doppler signals non-invasively.