| CHAPTER ONE:Support vector machine based MRI radiomics to identify primary hepatic carcinomas with different pathological typesObjectiveTo investigate the value of support vector machine based MRI-radiomics in the differential diagnosis of primary hepatic carcinomas(PHCs).MethodsPHCs patients were retrospectively collected from July 2013 to February 2017 in the First Affiliated Hospital of Zhejiang University.All patients underwent unenhanced and enhanced MRI liver scan before surgery,and confirmed by pathology.A total of 294 PHCs patients(305 lesions),including 96 cases(97 lesions)of massive type cholangiocarcinoma(MCC),107(107 lesions)of hepatocellular carcinoma(HCC),and 91(101 lesions)of mixed hepatocellular and cholangiocellular carcinomas(HCC-CC).All patients underwent unenhanced and dynamic enhanced MRI liver scan including arterial,portal venous and equilibrium phases.Two hundred and three lesions(65 MCC,71 HCC,67 HCC-CC)were assigned into the training set,the remaining 102 lesions(32 MCC,36 HCC,34 HCC-CC)into the validation set,according to a ratio of 2:1.The entire lesions were delineated manually using a region of interest on equilibrium phase of enhanced MRI by using a home-made dedicated software(Analysis Kit,AK,General Electrics,US).Then the least absolute shrinkage and selection operator(LASSO)regression was used to select image features with a method of 10 fold cross-validation,and to reduce the dimensionality.The spearman method was used afterwards to condense the image features by removing redundant.A predictive model of diagnosing the different types of PHCs was established based on support vector machines(SVM),and the accuracy of applying the model in the data sets was used to evaluate the diagnostic efficacy of the model.ResultsA total of 280 quantitative imaging features were extracted in the training set.Thirty one imaging features were selected after LASSO regression and dimensionality reduction,and 21 features were remained after redundancy removing.The SVM showed the best generalization ability when the first 11 imaging features were used due to the Hughes effect.The 11 imaging features include 4 parameters of histogram,2 of textures,4 of gray-level co-occurrence matrix and lof gray-level run length matrix.A predictive model for PHCs was established after the study of the 11 imaging features and a regression analysis using SVM.The accuracy of the predictive model was 80.3%(163/203)in differentiating PHCs in the training set.The accuracy of the model was 75.5%(77/102)after applying it in the validation set.The diagnostic accuracy for HCC-CC,HCC and MCC was 85.3%(29/34),77.8%(28/36)and 62.5%(20/32),respectively,in the validation set.For HCC-CC,3 cases were misdiagnosed as MCC and 2 cases as HCC.For HCC,3 cases were misdiagnosed as HCC-CC and 5 cases as MCC.For MCC,9 cases were misdiagnosed as HCC-CC and 3 cases as HCC.The model showed the highest accuracy in predicting HCC-CC.ConclusionRadiomics method based on SVM may have a high accuracy in predicting different pathologic types of PHC,with the highest accuracy for HCC-CC.CHAPTER TWO:MR radiomics combined with clinicopathological features to predict postoperative survival of primary hepatic carcinomas with different pathological typesObjective:This study was to investigate the factors that affect the prognosis of patients with primary hepatic carcinomas with different pathological types.Methods:We retrospectively collected 210 cases of primary hepatic carcinomas undergoing surgical resection at the First Affiliated Hospital of Zhejiang University from July 2013 to November 2015.All patients underwent preoperative liver MRI including plain and contrast enhanced scans,and confirmed by pathology.Seventy-four cases of primary liver cancers were excluded(68 cases lost,1 case of tumor recurrence,3 cases of lack of preoperative tumor indicators,1 case of liver transplantation,1 case of sarcoma).In total,136 cases of primary liver cancers were finally included in the study.Out of which,45 were mass-type cholangiocarcinoma(MCC)with a patient having 2 lesions;59 were hepatocellular carcinomas(HCC);36 were combined hepatocellular-cholangiocarcinoma(CHCC)patients,with a patient having 2 lesions,and a patient having 3 lesions.The largest tumor lesion was selected and the lesion was manually drawn using dedicated AK software(Analysis Kit,GE)to extract radiomics histological characteristics on the MRI diffusion image and the enhanced balance period image.If there was a missing value in the image feature column,fill in the missing value with the median of the column.Spearman method and lasso regression were applied for removing redundancy and screening feature parameters respectively.In the MR diffusion scan(DWI),385 image features were acquired,and 36 features remained after redundancy removing,and 6 characters remained after dimension reduction by using LASSO method.In contrast enhanced MR equilibrium phase(EP),385 image features wer eacquired,remaining 18 after redundancy removing,and 11 remaining after dimension reduction of LASSO.There were 18 clinical pathological features.All patients were followed up by telephone and the patient's postoperative survival time was recorded and divided into three groups for analysis of Cox regression models:group A(DWI and clinicopathological feature group),group B(contrast enhanced equilibrium phase and clinicopathological feature group),Group C(DWI,contrast enhanced equilibrium phase,and clinicopathological feature group).Results:Cox multivariate regression analysis showed that the radiomics characteristics and clinicopathological characteristics were significantly correlated with the survival of the three groups of primary liver cancers.In group A,preoperative AFP positive and Cluster Shade were negative prognostic factors affecting postoperative survival time of liver cancer(all p<0.05),p(AFP)=0.00581<0.05,(HR:3.23;95%confidence interval[CI]:1.4,7.43);p(Cluster Shade)=0.02392<0.05,(HR:1.0;95%confidence interval[CI]:1.0,1.0).In group B,the maximum diameter of the lesion,the number of preoperative chemotherapy,preoperative ferritin positive and preoperative AFP positive were the poor prognostic factors affecting the survival time of liver cancers(all p<0.05),p(maximal lesion diameter)= 0.000368<0.05,(HR:1.61;95%confidence interval[CI]:1.24,2.09);p(preoperative chemotherapy times)= 0.0458<0.05,(HR:0.16;95%confidence interval[CI]:0.0264,0.966);p(ferritin)= 0.0422<0.05,(HR:2.58;95%confidence interval[CI]:1.03,6.43);p(AFP)=0.0251<0.05,(HR:2.74;95%confidence interval[CI]:1.14,6.64);In group C,preoperative AFP positive and diffusion image Cluster Shade were the poor prognostic factors affecting the survival time of liver cancer(all P<0.05),p(AFP)=0.0147<0.05,(HR:3.03;95%confidence interval[CI]:1.24,7.39);p(DWI-Cluster Shade)=0.0372<0.05,(HR:1.0;95%confidence interval[CI]:1.0,1.0).The median overall survival time of 44 patients with mass cholangiocarcinoma was 29.8 months.The overall survival rates at 6 months,1 year,2 years,and 3 years were 93.2%,91.5%,79.8%,and 68.1%,respectively.The median overall survival time of 59 patients with hepatocellular carcinoma was 30.8 months.The overall survival rates at 6 months,1 year,2 years,and 3 years were 97.2%,94.1%,91.9%,and 78.3%,respectively.The median overall survival time of 33 patients with mixed hepatocellular and cholangiocellular carcinomas was 25.5 months.The overall survival rates at 6 months,1 year,2 years,and 3 years were 98.8%,89.6%,66.1%,and 51.6%,respectively.Conclusion:The largest diameter of the lesion,a large number of preoperative chemotherapy,preoperative ferritin positive and preoperative AFP positive were independent factors affecting the postoperative survival time of primary liver cancer patients with different pathological types.Cluster Shade was a promising biomarker and may be used for survival assessment of primary postoperative liver cancers. |
PDF Full Text Request