Role Of CT Texture Analysis In The Early Assessment Of The Response Of Gastric Cancers To Chemotherapy

Objective:To explore the role of computed tomography（CT）texture analysis parameters in the early assessment of the treatment response during the chemotherapy of gastric cancers.Methods:This retrospective study was approved by the local ethics committee.Twenty-four patients with gastric carcinoma underwent contrast-enhanced CT before chemotherapy and at the end of the 2nd and 4th cycles during chemotherapy.Polygonal regions of interest（ROIs）within the identified lesions in CT imagesbefore chemotherapy and at the end of the 2nd cycles during chemotherapy were manually drawn along the margin of the tumour on the section that depicted the largest area of the lesion.CT textureanalysis was carried out on those ROIs.CT texture analysis generated the following parameters:（1-3）max,min and mean,（4-9）the 5th,10th,25th,50th,75th and 90th percentiles,（10）skewness,（11）kurtosis,（12）entropy,and（13）inertia.After four cycles of treatment,the therapeutic responses were assessed according to the Response Evaluation Criteria in Solid Tumours.Specifically,patients with a complete response（CR）or a partial response（PR）were considered as the response group,while those with stable disease（SD）or progressive disease（PD）were classified as the non-response group.Furthermore,patients with CR,PR and SD were considered as the non-progress（non-PD）group.The parameters were compared among different responses to chemotherapy using the Mann-Whitney U-test.Additionally,the diagnostic performances of the parameters in differentiating different responses to chemotherapy were analysed by receiver operating characteristic（ROC）curve analysis.Results:The CT texture parameters of the 75th percentileP2,90th percentileP2 andΔinertia₁2V differed significantly between the progressive disease（PD）and non-PD groups while the CT texture parameters of maxA1,skewnessV1 and ΔskewnessV differed significantly between the non-response（stable disease or progressive disease）and response（complete response or partial response）groups.The above parameters performed well in distinguishing different responses to chemotherapy with areas under the ROC curve（AUCs）ranging from 0.754 to 0.832（all P<0.05）.The multivariate model showed better diagnostic efficiency with AUCs of 0.882 and 0.781,respectively（all P<0.05）.Conclusions:CT texture analysis has the potential to monitor the early tumour response in patients with gastric cancer undergoing chemotherapy.