Research on Method of Ultrasonic Noninvasive Body Temperature Monitoring
Abstract：Generally human noninvasive temperature measurement methods refer as Ultrasonic, nuclear magnetic resonance, residence measurement and so on, the last two methods have the disadvantages of high cost, low resolution or can’t measure regional temperature. Ultrasonic is the most common used method, which includes theories such as transmission attenuation, transmission time shift, emission attenuation, reflection time shift, reflection frequency shift and so on. This paper proposed a new monitoring method based on reflection attenuation, which computes temperatures directly from the B-ultrasonic images and needs no more other equipments. After contrast dada of temperature and B-ultrasonic images is measured under limited experiments conditions, the least-squares method is used to get gray-temperature equation. Other tissue parameters are added to amend the relation equation between temperature and the image gray value. Image segmentation flowchart was designed to partition the different tissue organ area in B-ultrasonic images automatically. The overall program was designed to analyze temperature distribution of different tissues by extracting texture characteristics in ultrasonic images and training artificial neural network. This method has characteristics of low cost, simple operability, high safeness, regional temperature monitoring and has good application prospect.
Keywords: noninvasive temperature monitoring, least-squares method, B-ultrasonic images
