Research Of Cherenkov Radiation For In Vivo Tumor Detection And Image-guided Tumor Surgery

Malignant tumor has become a great threat to human health.In China,estimated more than 10,000 people are diagnosed with tumor every day.Precise diagnosis and treatment are believed crucial for tumor patients.However,the high complexity of tumor causes high difficulty to improve the detection accuracy.In recent years,a promising molecular imaging technique,Cherenkov luminescence imaging(CLI)is expected to help scientists and clinicians to fight against tumor.CLI is an optical molecular imaging modality that could utilize clinical radiopharmaceuticals as imaging agents.Therefore,CLI inherits the merits from both optical imaging and nuclear imaging.In the past several years,many advantages of CLI has been validated,such as,high spatial resolution,high sensitivity,high throughput,etc.Currently,CLI has been widely used in tumor studies and even in some clinical scenarios.In order to further improve the performance of CLI,this dissertation proposed novel ideas from the perspectives of imaging methods,imaging systems and biomedical applications.Regarding imaging methods,this dissertation designed a multilayer fully connected neural network(MFCNN)to perform Cherenkov luminescence tomography(CLT).This MFCNN CLT method could efficiently analyze the complicated non-linear relationship between the inner Cherenkov luminescence(CL)sources and the detected surficial signals.Comparing with the radiative transfer equation(RTE)based CLT,MFCNN could avoid the computing errors during mathematical modeling and lead to the increase of imaging accuracy.The experimental data showed that MFCNN CLT outperformed the RTE method.In simulation evaluation,the localization error of CLT could be reduced to lower than 0.20 mm,while the reconstructed result was highly consistent to the true sources in shape(Dice coefficient>0.90).The high performance of MFCNN CLT could also be observed in vivo.The gliomas in mice were effectively detected by MFCNN and the accuracy was proved by PET,MRI and pathological results.All the experimental data supported the idea that MFCNN method is effective to improve the performance of tumor optical tomography.To explore the value of CL image-guided tumor surgery,an endoscopic CLI(ECLI)system and an intraoperative CLI system for open surgery were developed.A clinical laparoscope was used as the main ECLI component,and the optical path was improved to enhance the light transmission efficiency.Meanwhile,the dark current and the thermal noise during ECLI acquisition were reduced by integrating liquid-cooling.The ECLI system was characterized as 62.5 ?m for spatial resolution and 62.9 kBq/mL '8F-FDG for sensitivity.In liver tumor surgery on mice,the ECLI system showed high efficacy to delineate the tumor margin and guide the tumor resection.On the other hand,another intraoperative CLI system was designed to in vivo detect the tumor lesions during open surgery.The system integrated up to three different imaging channels.The surgeons could obtain color light anatomic information and high-specificity CLI results,which were beneficial to determine the resection areas.Characterization demonstrated that the intraoperative in vivo CLI system owned high spatial resolution(55.68 ?m)and desirable sensitivity(19.61 kBq/mL 18F-FDG).The experiments on mice and swine colorectal tumor models further validated the practicability of the system.The intraoperative in vivo results showed that the primary and metastatic colorectal tumors could be efficiently detected and resected.Radiation measurements of the related staffs also verified the safely to perform in vivo CL image-guided open surgery.With support from the clinical surgeons,this dissertation presented two novel strategies for image-guided liver tumor surgery.The first one combined PET,Cherenkov radiation energy transfer(CRET)and confocal laser endomicroscopy(CLE).This strategy applied ?+radiopharmaceuticals and fluorescein sodium(FS)as imaging tracers.PET provided whole-body tumor detection,CRET localized the tumor lesions on the specific liver areas with high contrast ratio,then CLE revealed the tumor margin in a microscopic view.This PET-CRET-CLE strategy imaged the tumor lesions from different perspectives,wchich could strongly help the surgeons in tumor staging and surgery.One significant advantage of PET-CRET-CLE imaging is that all the imaging tracers are clinical approved,and therefore brings high potential for clinical translation.In the other surgical strategy,visible light imaging,the first(NIR-?)and second(NIR-?)near-infrared window imaging were integrated to guide liver tumor surgery.Clinical approved indocyanine green(ICG)was injected as the imaging probe Cooperating with a surgical team,this multispectral image-guided surgical strategy was performed in human.In the trial involved 23 liver tumor patients,the visible-NIR-?/?strategy showed strengthened ability to detect small tumors,compared with the current clinical imaging modalities(CT,MRI,PET and ultrasound).The statistical analysis also revealed that the multispectral imaging,especially the NIR-? imaging could provide desirable tumor detection sensitivity(100%),specificity(88.46%)and accuracy(91.43%).These encouraging results were expected to promote the application of CLI and multispectral imaging in tumor surgery.Overall,this dissertation provided several novel thoughts to enhance the tumor detection ability using promosing optical imaging techniques.Innovations could be seen in imaging methods,imaging systems and biomedical applications,respectively.Experimental data in different scenarios highlighted the great potential of CLI and multimodality imaging to improve the precision of tumor detection.