Application Of ADC Image-based Histogram Analysis And Radiomics In Evaluating The Efficacy Of Neoadjuvant Chemoradiotherapy For Locally Advanced Rectal Cancer

Part One Application of whole tumor ADC histogram analysis in evaluating the efficacy of neoadjuvant therapy for locally advanced rectal cancerObjective: To explore the application of ADC histogram analysis of whole tumor in the evaluation of neoadjuvant curative effect of locally advanced rectal cancer.Methods: A retrospective analysis of 43 non-metastatic locally advanced rectal cancer(LARC)cases from March 2017 to March 2019.All patients underwent neoadjuvant chemoradiotherapy and 3T magnetic resonance imaging(MAGNETOM Skyra,Siemens Medical,Germany,before and after treatment).Erlangen)MR examination,the scanning sequence includes T2 WI,T1-DIXON,DWI sequence.Before and after chemoradiotherapy,the tumor maximum diameter lines Lbefore and Lafter,the maximum thickness Dbefore and Dafter were measured,and the differences ?L,?D and the change rates% ?L,% ?D were calculated.The open-source software of ITK-SNAP(https://itk.org/)was used to manually draw the ROI on T2 WI to obtain the tumor volume before and after chemoradiation and calculate the tumor volume reduction percentage(TVR).The ITK-SNAP was used to manually draw the ROI on the ADC map to obtain the first-order parameters of the ADC histogram after chemoradiotherapy,and calculate the difference and change rate of each parameter.The efficacy of rectal cancer was evaluated with reference to RECIST 1.1,TVR evaluation criteria,and TRG(tumor regression classification)-MRI diagnostic criteria.With the postoperative pathology as the gold standard,TRG(0,1,2)was defined as the effective group,and TRG3 was regarded as the ineffective group.The accuracy,sensitivity,specificity,positive predictive value and negative predictive value of tumor regression response were calculated.A paired t test was used to assess whether there were statistical differences in the first-order parameters of the ADC histogram before and after treatment and the rate of change.Two-sample t test and t' test were used to analyze the difference and change rate of the first-order parameters of the ADC histogram between the treatment-effective group and the ineffective group,and to check whether there was a statistical difference in the values of the parameters between the two groups at baseline.Finally,the ROC curve and AUC were used to compare the differences in histogram analysis parameters with statistical differences,the rate of change,and baseline diagnostic parameters to compare the diagnostic efficacy.Results: The accuracy of the RECIST1.1 standard thickness assessment of the efficacy of rectal cancer n CRT was 69.8%,the sensitivity was 69.7%,the specificity was 30%,the PPV was 71.9%,and the NPV was 23.1%.The accuracy of the long-range evaluation of the efficacy of rectal cancer n CRT was 53.5%,the sensitivity was 57.6%,the specificity was 60%,the PPV was 82.6%,and the NPV was 30%.The accuracy of volumetric evaluation of rectal cancer n CRT was 30.2%,sensitivity was 36.4%,specificity was 90%,PPV was 92.3%,and NPV was 30%.The accuracy of TRG-MRI for evaluating the efficacy of rectal cancer n CRT was 60.4%,the sensitivity was 54.5%,the specificity was 80%,the PPV was 90%,and the NPV was 34.8%.ADC 10 th percentile,90 th percentile,Energy,Entropy,Maximum,Mean,Median,Minimum,Extreme Difference(Range),Root Mean Squared,Skewness,Total Energy,Uniformity and the change rate of energy and total energy before and after neoadjuvant chemoradiation Statistical significance(all P <0.05).The rates of change in Energy and Total Energy before and after treatment in the effective group were greater than those in the ineffective group,and the differences were statistically significant.Using the ROC curve,when the energy and total energy change rate were greater than 38.28%,the accuracy of predicting the effectiveness of neoadjuvant chemoradiotherapy for rectal cancer was 72.1%,the sensitivity was 81.8%,the specificity was 60%,and the PPV was 87.1% and the NPV was 50%.Part two Application of ADC imaging-based different Radiomics models in predicting the efficacy of neoadjuvant chemoradiotherapy for locally advanced rectal cancerObjective: To investigate the effectiveness of predicting the efficacy of neoadjuvant chemoradiotherapy for locally advanced rectal cancer based on different Radiomics models based on ADC map.Methods: A retrospective analysis of 43 non-metastatic locally advanced rectal cancer(LARC)cases from March 2017 to March 2019.All patients underwent neoadjuvant chemoradiotherapy.3T magnetic resonance imaging(MAGNETOM Skyra,Siemens Medical)was used before and after treatment.Erlangen,Germany)for MR inspection.ITK-SNAP was used to manually outline the ROI on the ADC map before treatment,and 109 imaging omics features were extracted by the imaging omics software Pyradiomics(https://pyradiomics.readthedocs.io/en/latest/index.html).Forty-three patients used paired-difference analysis(PDA)to increase the sample size,and a total of 378 sample pairs were obtained,which were randomly divided into training and test groups according to 7: 3.Support Vector Machine(SVM),Auto-Encoder(AE),Linear Discriminant Analysis(LDA),Random Forest(Random Forest),and Logistic Regression(LR)And LR-Lasso(Logistic Regression via Lasso).According to the accuracy,sensitivity,and specificity of the model on the test set,the optimal combination of a model is determined,and the receiver's working characteristic curve analysis is used to evaluate the diagnostic performance of different models.The analysis model was developed based on Sklearn(https://scikit-learn.org/)and Fe Ature Explorer(FAE,v0.2.5,https://github.com/salan668/FAE)developed by East China Normal University.Results: Finally,the three more stable models were screened: SVM,RF,and LR-Lasso models.The SVM model has an AUC,accuracy of 0.934 and 98.4%,sensitivity and specificity of 80% and 100%,negative predictive value and positive The predicted values are 98.3% and 100%.The RF model has AUC,accuracy of 0.998 and 98.4%,sensitivity and specificity of 100% and 98.3%,negative predictive value and positive predictive value of 100% and 83.2%.The LR-Lasso model has AUC,accuracy of 0.997 and 98.4%,sensitivity and specificity of 100% and 98.3%,negative predictive value and positive predictive value of 100% and 83.3%.Conclusions:1.Compared with the conventional MR morphology and volume change rate before and after tumor treatment,the energy and total energy change rate based on the first-order parameters of the whole tumor ADC histogram can predict the efficacy of neoadjuvant therapy for rectal cancer and improve the diagnostic efficiency.2.The Radiomics model has a higher accuracy in predicting the efficacy of locally advanced rectal cancer.The Radiomics model established by the Random Forest(RF)method has higher diagnostic efficiency than other omics models.