Multi-parameter Photoacoustic Mesoscopy For In Vivo Small Animal Tumor Imaging

Medical functional imaging technology has become the key development of modern medical researches,to reveal the formation mechanism of diseases and explore the diagnosis as well as treatment of diseases(such as tumors).Due to the complexity and diversity of tumor cell pathways,it’s necessary to conduct more comprehensive imaging studies and quantitative analyses of tumor angiogenesis and drug effects based on in vivo small animal in pre-clinical experiments.Photoacoustic mesoscopy(PAMe)method based on acoustic focusing efficiently combines the high optical contrast of optical imaging with high spatial resolution of acoustic imaging.PAMe can achieve high resolution imaging(10-100μm)with the imaging depth of1-10 mm,suitable for the biomedical researches of whole subcutaneous solid tumors,including structural imaging of vessels,functional imaging of oxygen saturation and molecular imaging of targeted probes or drugs.Due to the requirements of imaging sensitivity,quantification and practicability in the research of small animal tumor imaging,PAMe method still has limitations in optical modeling,acoustic reconstruction,imaging quality and measuring system.The existed problems include: lack of a fast and accurate photon transport modeling method with multi-angle wide-field illuminations in photoacoustic imaging to effectively provide a reliable optical mathematical model;lack of an efficient reconstruction method of initial acoustic pressure for large datasets PAMe that fully considers the impulse response characteristics of transducer to improve acoustic quantification;lack of a highly applicable limited-view photoacoustic imaging quality optimization method to eliminate artifacts and distortions in limited-view PAMe images;lack of a multi-spectral PAMe measurement system with high signal-to-noise ratio and multi-angle measurement data to achieve high-sensitivity and multi-angle information measurement.On account of the existed problems,this paper aims to establish a set of PAMe approaches on in vivo small animal tumor imaging,improving the sensitivity,accuracy and feasibility of PAMe and effectively revealing multi-parameter information of tumors such as the structural and functional information.A fast and accurate photon transport modeling method for photoacoustic imaging with arbitrary multi-angle wide-field illuminations is developed.This method can provide the distribution of photon density in biological tissues,which is helpful to the establishment of the mathematical model of initial acoustic pressure.This method simultaneously and randomly initializes the incident angles and positions of emitted photons,realizing the rapid and precise photon transport modeling with arbitrary wide-field illuminations by parallel acceleration.Compared with traditional photon transport modeling methods with wide-field illuminations,this method shows higher computing efficiency and universal applicability.An initial pressure reconstruction method that incorporates the impulse response of high-frequency focused transducer is developed,which can effectively remove the effect of the impulse response characteristics of the commonly used focused transducers in PAMe system on the accuracy and fidelity of reconstructed images of initial acoustic pressure.The influence of electrical impulse response of transducer depending on the electrical property of the piezoelectric crystal can be firstly removed by a deconvolution method.After that,the correlation between the spatial impulse response of transducer and the space as well as ultrasound frequency is fully considered.As the spatial impulse response is only related to the size and position of the transducer,a time-domain back-projection reconstruction method combined with the space-variant filter is constructed.Experiments are performed on a large datasets PAMe system based on translation and rotation scanning.The proposed method with parallel computing can effectively correct the image distortion caused by spatial impulse response and improve the accuracy of reconstructed PAMe images.This method overcomes the limitations of traditional PAMe algorithms,which cannot manage the large PAMe datasets due to the difficulty of large-scale matrix construction and cannot fully consider the effect of spatial impulse response.The proposed method meets the requirements for the accuracy and efficiency of image reconstruction in the PAMe method on in vivo small animal tumor imaging.A deep learning approach to optimize limited-view photoacoustic imaging quality is proposed,solving the problem of imaging quality degradation when the full-view detection cannot be achieved in PAMe systems.This method can overcome the problems of prior knowledge and long running time required by traditional iterative methods.By designing reasonable network structure,optimization method and loss function,the method can effectively utilize the generative and adversarial characteristics to achieve the optimization of limited-view photoacoustic imaging quality.To reduce the need for large experimental datasets and improve the applicability of the method in different situations,hybrid datasets have been established including rich simulation datasets and relevant experimental datasets.Simulations and experiments have proved the ability of proposed method to effectively improve the quality of photoacoustic images in different limited-view cases,and its reliability in practical applications.Without additional training,this method can be effectively transplanted to another limited-view PAMe system to improve the quality of photoacoustic images.A set of multi-spectral multi-angle PAMe experimental system is designed for the experimental researches on in vivo small animal tumor imaging.The combined strategy of single focused transducer and multi-angle scanning is established to make up for the inability of existed technologies,achieving both high signal-to-noise ratio and multi-angle data acquisition.The high resolution of the system(less than 100μm),multi-spectral imaging capability and molecular imaging capability are verified by phantom and in vivo experiments.In the PAMe experiments of pancreas cancer and liver cancer of in vivo mouse tumors,the tumor vasculature can be observed.According to the distribution of multi-parameter information such as the structure and function of the tumor revealed by PAMe,these two types of tumors can be preliminarily distinguished.The therapeutic efficacy of a targeted drug is explored by the PAMe system based on in vivo mouse tumors of human lung cancer.By monitoring the changes in body weight,tumor volume,tumor hemoglobin and blood oxygen concentration of mice in the control and treated groups during the experimental period,the monitor of tumor growth and evaluation of treatment efficacy are preliminarily realized.In view of the limitations of reconstruction methods and measurement systems of PAMe technology for the research of in vivo small animal tumor imaging,the effective modeling of photon transport,the accurate reconstruction of initial acoustic pressure,and the optimization of limited-view imaging quality have been developed.What’s more,a multi-spectral multi-angle PAMe system has been designed and the system performance has been verified by experiments.Finally,in vivo small animal tumor experiments for tumor type identification and drug efficacy evaluation have been carried out.The multi-parameter PAMe method developed in this paper preliminarily realizes the overall high-resolution imaging and quantitative evaluation of structural and functional information of subcutaneous solid tumors.It can be used to quantitatively analyze the differences and changes in tumor internal texture and functional parameters of different types or different treatment time points,and promote the application process of photoacoustic imaging technology in tumor disease detection and quantitative evaluation of therapeutic efficacy.