Research On In Vivo Computational Technology For Early Tumor Detection

The detection of early tumor lesions is of great significance to improve the survival rate of patients,but it has been a great challenge in the field of biomedicine due to the difficulty of its realization.The nanoparticles used in the current medical contrast imaging methods rely on the human circulatory system for large-area delivery as contrast agents,and lack the external force drive and guidance required for local targeted transmission,which easily leads to toxic side effects,low targeting and low efficacy,etc.Although a nanorobot based tumor detection system can improve efficiency,it requires specific location information of the tumor as prior knowledge.Due to the limitation of current medical imaging technology,the location information of early tumor lesions is difficult to obtain.Therefore,this thesis intends to propose a brand new tumor detection system that can combine the advantages of the above two technologies,and use nanorobots to achieve the purpose of direct targeted detection without the need for tumor location information as a priori knowledge.This project aims to design a manipulable smart nano system based on nanorobots from the perspective of computation.The system can control the nanorobots to search regularly for the high-risk tissue by using the changes in the tissue environment induced by early tumor lesions,and finally realize the detection of tumor lesions.The use of this method for early tumor detection does not require prior knowledge of the location of the tumor lesions,and can greatly improve the detection efficiency.The main research contents of the thesis are as follows:The specific changes of the biological tissue environment induced by early tumor lesions are analyzed,and the concept of “ biological gradient field ” is proposed.Externally manipulable magnetic nanorobot systems that can be used for tumor detection are introduced,and the computing theory of swarm intelligence algorithms is briefly analyzed.Based on this,the concept of “in vivo computation” is proposed under the consideration of the in vivo constraints of the nanorobots during the tumor detection process.With the ideal control of nanorobots,the mathematical optimization model of early tumor detection is established under the consideration of the speed decay of the nanorobotas a computing agent during the tumor detection process,and the tumor vascular network is modeled by using invasion percolation technology.The tumor vascular network is used as the main constraint for in vivo computation.Inspired by the theory of swarm intelligence algorithms,the algorithms for unifocal and multifocal tumor detection are designed respectively.The control problem of swarm nanorobots under the current magnetic field control technology is studied.Aiming at this constraint condition,an in vivo computing strategy(the weak priority evolution strategy(WP-ES))that can effectively realize tumor detection is proposed.Under the action of this strategy,the nanorobot swarm can approach the tumor lesion location along a relatively ideal path under the control of an external uniform magnetic field,and finally reach the target location to complete the tumor detection.Simulation experiments have been performed to verify the effectiveness of the computing strategy in a variety of representative biological gradient field landscapes.According to the movement direction of the nanorobots calculated by WP-ES during the iterative calculation,this thesis next studied the step size of the nanorobots during the iterative calculation.The “Tension-Relaxation(T-R)” strategy was designed to balance the control and tracking periods of the nanorobots during the calculation process under a fixed step size;an exponential evolution strategy was designed to adjust the “exploration” and“exploitation” performances of in vivo computation under varying step sizes.The results of simulation experiments proved that the design of the T-R strategy and exponential evolution strategy is of great significance to improve the efficiency of tumor detection.Based on the research results of unifocal tumor detection and the characteristics of multifocal tumors,in vivo computational algorithms based on a sequential targeting strategy(Se-TS)were proposed for multifocal tumor detection.The nanorobot swarms injected at the same injection position at different times could respectively detect their targeted tumor lesions along different motion paths under the effect of the algorithms,and finally achieve multifocal tumor detection.By comparing with the traditional brute-force search algorithm,we could know the superiority of the in vivo computational algorithm in tumor detection,and the comparison with the traditional swarm intelligence algorithm proved its speciality.According to the proposed in vivo computation theory,a corresponding in vitroexperimental platform was set up to achieve in vitro verification of in vivo computation.A two-dimensional microchannel network was fabricated using photolithography to simulate the vascular network structure of high-risk tissues.Janus particles were fabricated as magnetic micro/nanorobots to realize motion control under the action of Helmholtz coil system.The experimental results showed that the nanorobot swarms can realize the targeted detection of corresponding tumor lesions under the action of in vivo computational algorithms.In vitro experiments have proved the effectiveness of the in vivo computation proposed in this thesis for the detection of unifocal tumor and multifocal tumors,laying a foundation for its early clinical application.