A Preliminary Study Of CT/MRI-based Radiomics In Predicting The Efficacy Of TACE With Arsenic Trioxide In The Treatment Of Primary Hepatocellular Carcinoma

Objective: As an important part of precision medicine,radiomics plays an important role in the diagnosis and differential diagnosis of primary liver cancer,case classification and efficacy prediction.At present,the application of imaging omics has not been found to predict the efficacy of TACE treatment with arsenic trioxide.Therefore,this study aims to analyze CT and MR images,obtain relevant texture features,and construct corresponding radiomics models to evaluate whether it can predict the efficacy of TACE in the treatment of primary liver cancer with arsenic trioxide.Methods: A retrospective collection of 60 patients in the experimental group of the First Affiliated Hospital of China Medical University from 2014 to 2016 participated in the national “863” project “Clinical Application of Transcatheter Arterial Chemoembolization with Arsenic Trioxide in the Treatment of Primary Liver Cancer”.Preoperative enhanced CT images of 32 patients with primary liver cancer who underwent TACE with arsenic trioxide(ATO)regimen and 55 patients with primary liver cancer who received TACE with arsenic trioxide regimen were enrolled in the study.Images,according to the treatment effect after 6 months,the patients were divided into the remission group and the non-remission group by using the RECIST standard.The VOI images of each group of patients were obtained by layer-by-layer segmentation of images,and the texture features in the images were extensively extracted,and dimension reduction model was established by LASSO.Dimensionality reduction and model establishment of image features are performed to evaluate whether the acquired image features can effectively distinguish the patient’s curative effect.Results: The CT-based image omics model obtained five texture features,namely kurtosis,histogram energy,high-density compensation,high-density long-range compensation and total entropy.The area under the receiver operating characteristic curve(AUC)reached 0.90.The sensitivity is 63.63% and the specificity is 90.91%.The MR-based imaging omics model obtains three texture features,namely mean deviation,long run compensation and clustering,and the area under the receiver operating characteristic curve(AUC).It reached 0.80,the sensitivity was 61.90%,and the specificity was 82.35%.Conclusion: The CT-based imaging ensemble model predicts the efficacy of TACE in the treatment of primary liver cancer with arsenic trioxide-based regimen.The MR-based radiomics model predicts the efficacy of TACE in the treatment of primary liver cancer with arsenic trioxide.All of them can provide reference for clinical rational diagnosis,treatment and prognosis prediction.