Research On Technology And System Of Multi-parameter Photoacoustic Microscopy

Photoacoustic tomography is a new hybrid imaging mode based on optical excitation and ultrasonic detection.It has been developed rapidly in recent years and has become a research hotspot in biomedical imaging field.Multi-parameter photoacoustic microscopy is an important technology belonging to photoacoustic imaging.Through this technology,multiple parameters of microvascular in living biological tissue can be measured simultaneously,and then the oxygen metabolism of the measured tissue can be analyzed.During the growth of many kinds of nervous system diseases and tumors,the abnormal oxygen metabolism is often accompanied,therefore multi-parameter photoacoustic microscopy plays an important role in the research of Neurology and Oncology,and has great potential in clinical application.However,due to the influence of many factors,such as bulky imaging probe,single imaging scanning method,and low efficiency of imaging data calculation,the application of multi-parameter photoacoustic microscopy has been greatly limited and couldn't be used in clinical application.Based on the international cooperation project and the biomedical research fund of the University of Virginia,aiming at the main problems existing in multi-parameter photoacoustic microscopy and its application requirements in biomedical research and clinical diagnosis,this paper has carried out research on the technology and system of multi-parameter photoacoustic microscopy.The research is carried out in four aspects,ranging from the hardware structure,scanning method,data calculation to real-time imaging.The functional verification and performance evaluation of the design content are carried out through a number of simulation experiments and live animal imaging experiments.The main work of this paper includes the following aspects:(1)According to different imaging requirements,the hardware structure of desk and hand-held multi-parameter photoacoustic microscopy is studied.In the research of hardware structure of desktop imaging system,the structure of coaxial reflection probe and confocal detection mode are realized by using annular ultrasonic detector,and the mechanical scanning of probe is realized by using linear displacement platform fixed on optical platform,so as to meet the needs of desktop multi-parameter photoacoustic microscopy in large area of biological tissue.In the research of hardware structure of hand-held imaging system,the scanning light path based on MEMS lens and the ultrasonic coupling water tank based on cage cube are used to realize the hand-held probe structure,and the confocal detection mode of probe head is realized by using weak focus ultrasonic detector and making the focus spot conduct optical scanning in the ultrasonic focus area,so as to meet the requirements of hand-held multi-parameter photoacoustic microscopy in small area of biological tissue.The signal-to-noise ratio and resolution of the above two imaging systems are tested by confocal adjustment experiment and resolution test experiment.(2)Aiming at the signal detection and scanning requirements of multi-parameter photoacoustic microscopy,a raster scanning method and a real-time contour scanning method are proposed.In the raster scanning method,the surface of the measured tissue is scanned point by point,and at the same time,multiple parameter values are simultaneously measured at the same resolution by using dual wavelength pulse laser intensive excitation.On this basis,a real-time contour scanning method is proposed to solve the problem that the raster scanning method cannot image the curved tissue.In this method,the real-time calculation of the measured tissue surface profile and the dynamic adjustment of the imaging probe height are carried out at the same time of raster scanning,so as to ensure that all imaging areas are within the confocal range of the probe.On the desktop imaging system,the above two scanning methods were used to conduct large-area imaging experiments on the cerebral cortex of living mice.The results of the two groups were processed and compared in various ways,which showed the effect and mechanism of defocusing on the measurement results of multiple parameters during raster scanning,and the effectiveness of real-time contour scanning.(3)In order to solve the problem of low efficiency of data calculation in imaging system,high-speed parallel computing is studied based on the data characteristics and computing requirements of multi-parameter photoacoustic microscopy.Aiming at the same calculation process of photoacoustic signals at dense sampling points,a highspeed parallel calculation method of imaging data based on GPU is proposed.Aiming at the problem of repeated calculation of the algorithm,the optimal calculation method of calculating correlation coefficient table first and then extracting data to construct correlation coefficient curve is adopted.Aiming at the processing flow and dependency of imaging data,the cache space and calculation source are partitioned in hierarchical structure,and a parallel algorithm structure with multiple parallel granularities is designed.In the CUDA 9.1 software environment,NVIDIA GTX 1080 Ti computing hardware is used to test the proposed high-speed parallel computing method.The acceleration ratio of each computing step is tested through a variety of comparative experiments,and the performance of the overall design is evaluated through a number of test data.(4)Research on real-time imaging of hand-held multi-parameter photoacoustic microscopy.In order to improve the scanning speed of hand-held probe,a fast raster scanning method based on data reorganization is proposed,which combines the driving characteristics of dual-axis MEMS mirror and the principle of multi-parameter measurement.The real-time imaging software is designed by integrating multiple steps from imaging scanning,data transmission,data calculation to result display in a unified development environment.In order to realize the non-blocking transmission of sampling data,the data transmission mode of multi-cache space rotation is proposed.For the serial and parallel computing requirements of imaging data in real-time imaging,the data processing method based on CPU-GPU cooperation is proposed.The real-time imaging function of the hand-held imaging system was verified through the imaging experiments of microvessels in the ears and cerebral cortex of mice in vivo.