Clinical Research On Predicting Microvascular Invasion And Treatment Prognosis In Hepatocellular Carcinoma By Intelligent Medical Image Analysis

Part Ⅰ A Comparative Study for CT-and MRI-based Prediction of Microvascular Invasion in Solitary Hepatocellular Carcinoma Background and objective: Computed tomography(CT)and magnetic resonance imaging(MRI)are both capable of predicting microvascular invasion(MVI)in hepatocellular carcinoma(HCC).However,which modality is better is unknown.The study aimed to intraindividually compare CT and MRI for predicting MVI in solitary HCC and investigate the added value of radiomics analyses.Methods: This bi-institutional retrospective study enrolled 402 consecutive patients with pathologically confirmed HCC underdoing preoperative CT and MRI scans.Patients were randomly split into training(n = 300)and validation sets(n =100)according to a ratio of 3:1.CT-and MR-based radiomics signatures(RS)were constructed using the least absolute shrinkage and selection operator regression.CT-and MR-based radiologic(R)and radiologicradiomics(RR)models were developed by univariate and multivariate logistic regression.The performance of the RS/models was compared between two modalities.To investigate the added value of RS,the performance of the R models was compared with the RR models in HCC of all sizes and 2–5 cm in size.Model performance was quantified by the area under the receiver operating characteristic curve(AUROC)and compared using the Delong test.Results: Histopathologic MVI was identified in 161 patients(training set: validation set =130:31).MRI-based RS/models tended to have a marginally higher AUROC than CT-based RS/models(AUROCs of CT vs.MRI,P: RS,0.801 vs.0.804,0.96;R model,0.809 vs.0.832,0.09;RR model,0.835 vs.0.872,0.54).The improvement of RR models over R models in all sizes was not significant(P = 0.21 at CT and 0.09 at MRI),whereas the improvement in 2–5cm was significant at MRI(P < 0.05)but not at CT(P = 0.16).Conclusions: CT and MRI had a comparable predictive performance for MVI in solitary HCC.The RS of MRI only had significant added value for predicting MVI in HCC of 2–5 cm.Part Ⅱ CT Radiomic-based Microvascular Invasion Prediction in Assistance of Selecting Hepatectomy Types in Hepatocellular CarcinomaBackground and objective: Prediction models with or without radiomic analysis for microvascular invasion(MVI)in hepatocellular carcinoma(HCC)have been reported,but the potential for model-predicted MVI in surgical planning is unclear.Therefore,we aimed to explore the effect of predicted MVI on early recurrence after anatomic resection(AR)and nonanatomic resection(NAR)to assist surgical strategies.Methods: Patients with a single HCC of 2–5 cm receiving curative resection were enrolled from 2 centers.Their data were used to develop(n = 230)and test(n = 219)two prediction models for MVI using clinical factors and preoperative computed tomography images.The two prediction models,clinico-radiologic model and clinico-radiologic-radiomic(CRR)model(clinico-radiologic variables + radiomic signature),were compared using the Delong test.Early recurrence based on model-predicted high-risk MVI was evaluated between AR(n = 118)and NAR(n = 85)via propensity score matching using patient data from another 2 centers for external validation.Results: The CRR model showed higher area under the curve values(0.835–0.864 across development,test,and external validation)but no statistically significant improvement over the clinico-radiologic model(0.796–0.828).After propensity score matching,difference in 2-year recurrence between AR and NAR was found in the CRR model predicted high-risk MVI group(P = 0.005)but not in the clinico-radiologic model predicted high-risk MVI group(P = 0.31).Conclusions: The prediction model incorporating radiomics provided an accurate preoperative estimation of MVI,showing the potential for choosing the more appropriate surgical procedure between AR and NAR.Part Ⅲ A CT Radiomic-based Model for Prediction of Overall Survival in Patients with Hepatocellular Carcinoma Undergoing Transarterial ChemoembolizationBackground and objective: Patients with hepatocellular carcinoma(HCC)receiving transarterial chemoembolization(TACE)have various clinical outcomes.The study aimed to develop a radiomic signature(Rad-signature)using pretreatment CT scans to establish a combined radiomics-clinic(CRC)model for estimating overall survival(OS)in these patients.Additionally,the performance of the proposed CRC model was compared with the existing prognostic models for predicting patient survival.Materials and methods: This retrospective study included multicenter data from 162 treatment-na?ve patients with unresectable HCC undergoing TACE as an initial treatment from January 2007 and March 2017.Patients were split into a training cohort(n = 108)and a testing cohort(n = 54)according to enrolled time.Radiomics features were extracted from intra-and peritumoral regions on both the arterial and portal venous phases CT images.A Rad-signature for predicting OS was constructed using the least absolute shrinkage and selection operator Cox regression method in the training cohort.With univariate and multivariate Cox regression analyses,a CRC model was developed using the Rad-signature and the clinical factors with a potential association with OS.Model performance,discrimination,and calibration were measured by concordance-index(C-index),the time dependent area under receiver operating characteristic curve(AUROC),and a calibration curve,respectively.Finally,the C-index of CRC model was compared with that of the seven well-recognized models,including four-and seven criteria,six-and-12 score,HAP score,mHAP score,mHAP-Ⅱ score,mHAP-Ⅲ score,and ALBI grade.Results: The radiomics signature,consisting of two features from arterial phase from tumor VOI and peritumoral VOI and four features from tumor VOI on portal phase,achieved C-index of 0.68(95% confidence interval [CI]:0.62,0.74)and 0.67(95% CI:0.56,0.79)in the training and testing cohorts,respectively.Univariate and multivariate Cox analysis identified the radiomic signature and tumor numbers were associated with OS.The CRC model combining the Rad-signature and tumor number performed better than the other seven well-recognized prognostic models,with C-indices of 0.73(95% CI: 0.68–0.79)and 0.70(95% CI: 0.62–0.82)in the training and testing cohorts,respectively.Among the seven models tested,the six-and-12 score and four-and-seven criteria performed better than the other models,with Cindices of 0.64(95% CI: 0.58–0.70)and 0.65(95% CI: 0.55–0.75)in the testing cohort,respectively.Conclusion: The CT radiomics signature represents an independent biomarker of survival in patients with HCC undergoing TACE,and the CRC model displayed improved predictive performance.