Imaging Diagnosis And Image Feature Analysis Of Liver Cancer Based On Phatoacoustic Molecular Imaging

The survival rate of patients with melanoma liver metastasis is very low,and melanoma has a poor prognosis,high morbidity and high mortality to lead aggressive progression and high risk of metastasis.Therefore,early and accurate diagnosis of liver micrometastasis is essential to improve the prognosis and survival rate.Traditional imaging techniques have some limitations such as low sensitivity,long scanning time,and high inspection costs,and ionizing radiation.Therefore,a new medical imaging technology needs to be proposed to solve these problems.The advantages of photoacoustic imaging technology with high optical contrast,high spatial resolution,low cost,real-time imaging,and no ionizing radiation can well complement the existing imaging technology to detect and characterize early tumors.Moreover,computer-aided diagnosis of liver tumors based on photoacoustic images is helpful for rapid and accurate diagnosis of clinical tumor samples.The main contents of this paper are summarized as follows:In Chapter 1,we summarized the current research status of melanoma liver metastasis,including the current status,metastasis mechanism,and existing treatment methods.Subsequently,the current status of molecular imaging diagnosis for liver metastases of melanoma was summarized,including ultrasound imaging,MRI,CT,PET,and PAI.Then,Radomics research in liver tumors is summarized.Finally,the basis for the topic selection and research content of this thesis are proposed.In Chapter 2,we evaluated the ability of photoacoustic imaging technology for early diagnosis and intraoperative photoacoustic image-guided tumor resection by constructing mice model of melanoma liver metastasis.,The detection limit of PAI is about 219 cells per microliter in vitro.PAI can detect melanomas as small as 400 μm at a depth of 7 mm in vivo.At the same time,we proved that PAI has high sensitivity and specificity as a new imaging method.It can detect small tumors at the submillimeter level that cannot be detected by traditional imaging techniques,and realize early diagnosis of tumors and help guide the doctor develops a treatment plan as soon as possible.In addition,under high-resolution and high-sensitivity detection conditions,Photoacoustic images guiding intraoperative navigation can further guide tumor resection by reducing residuals.In Chapter 3,we used label-free method of photoacoustic imaging to study the feasibility of liver tumor oxygenation and liver ICG metabolism using mice models of melanoma liver metastasis and primary liver cancer.Our research shows that photoacoustic imaging technology can well evaluate the oxygenation of mice liver and generate high-resolution structural information,which can be verified by immunohistochemistry and pathology methods.Subsequently,we used this system to evaluate ICG metabolism for mice with primary liver cancer.In short,photoacoustic functional imaging can well evaluate liver tumor oxygenation kinetics and pharmacokinetics,and it also provides important value for clearly formulating treatment plans and determining prognosis evaluation.In Chapter 4,we use photoacoustic imaging technology to image mice with liver metastases from melanoma,and combine the imaging feature analysis method with photoacoustic imaging technology to construct a computer-aided diagnosis model for liver metastases of melanoma,which can achieve rapid,non-destructive,and accurate diagnosis.Based on the Radiomics feature extraction method,106 features were extracted from the photoacoustic images of liver tumor tissues,and the extracted data were analyzed and processed through the mRMR feature selection algorithm and TPOT automatic machine learning tools.Finally,the classification model of normal liver tissue and tumor tissue was successfully constructed,and the AUC value was 0.96,which showed that the model had high accuracy.This method avoids tedious experimental steps,saves time,and extremely accurate compared with pathological staining,which can greatly assist clinicians in diagnosis and treatment.In Chapter 5,we propose a tumor computer-aided diagnosis model based on photoacoustic images to distinguish between good and bad tumor samples.Our results show that the boundary between normal tissue and tumor tissue can be distinguished by extracting the contour information of photoacoustic image,and the degree of tumor invasion can be displayed quickly.ROC analysis of poorly differentiated and moderately differentiated tumors showed the highest kurtosis of AUC histogram parameters.Therefore,the histogram parameter kurtosis can be used as a potential tool to identify the degree of tumor heterogeneity.This means that photoacoustic image histogram analysis is a fast and accurate discrimination method compared with the traditional pathological diagnosis method,which is helpful to the pathological analysis of tumor.