Value Of Intratumoral Susceptibility Signal Intensities Automatic Extraction Combined With R₂^* In Quantitatively Evaluating Pathological Grade Of Hepatocellular Carcinoma Using Enhanced T2 Star-weighted Angiography And R₂<

Part 1 Value of intratumoral susceptibility signal intensities automatic extraction combined with R₂^* in quantitatively evaluating pathological grade of hepatocellular carcinoma using enhanced T2 star-weighted angiographyObjective To explore the value of of intratumoral susceptibility signal intensities（ITSS）automatic extraction combined with R₂^* in quantitatively evaluating pathological grade ofhepatocellular carcinoma（HCC）by using enhanced T2 star-weighted angiography（ESWAN）.Materials and Methods From October 2011 to December 2019,a total of 58 HCC patients who underwent liver 1.5T MR scanner preoperatively in our hospital were enrolled in this retrospective study.All the patients enrolled were pathologically confirmed as HCC post-operatively,including well-moderately differentiated HCC group（n = 37）and poorly-differentiated HCC group（n = 21）.The MRI scanning sequences included T1 WI,T2WI,liver acquisition with volume acceleratio（LAVA）enhanced scanning and ESWAN sequence.All images on ESWAN sequence were transferred to the GE AW 4.6 workstation for post-processing and magnitude maps,phase maps,R₂^* maps and T2* maps were obtained.The solid part of HCC lesions were measured by two radiologists（observer 1 with 3 years of MRI reading experience and observer 2 with 10 years of MRI reading experience）respectively on R₂^* maps to obtain R₂^* values.Phase maps after using batch program to remove artifacts were exported as NII format,and transmitted to Anatomy Sketch（A.S.）software,which is programmed using C++ based on Qt and VTK libraries（Dalian University of Technology）.Referencing to T2 WI images,regions of interest（ROI）were delineated from the head side,around the edge of the first,the last,and the any layers between the two layers of the tumor（excluding tumor pseudo-capsule）on phase maps by the two radiologists who measured R₂^* values above.Maximum cross section of tumor can be obtained without layer-by-layer annotation through the interpolation and annotation tools of AS.After that,the AS software automatically calculated the ITSS ratio of the phase maps.All statistic analyses were analyzed by SPSS 25.0 software and Medcalc software 19.5.Intra-class correlation coefficients（ICC）were used to evaluate the consistency of the two radiologists’ measurements.Normal distribution was tested by Shapiro-Wilks test.ITSS ratios and R₂^*values between well-moderately and poorly-differentiated HCC were compared by two independent-sample t test or Mann-Whitney U test.Spearman’s rank correlation test was used to evaluate ITSS ratios,R₂^* values and HCC pathological grade.Receiver operating characteristic（ROC）analysis was performed to evaluate diagnostic performance of ITSS ratios,R₂^* values and the combination of the two.Delong test was used to compare the area under the curve（AUC）of ITSS ratios,R₂^* values and the combination of the two.Results ITSS ratios and R₂^* values by two radiologists were in great agreement（ICC > 0.80）.There were significant differences of ITSS ratios and R₂^* values between two groups（P =0.002,0.000,respectively）.The ITSS ratios and R₂^* values of poorly-differentiated HCC were higher than those of well-moderately differentiated HCC（ITSS ratios:（0.192±0.107）>（0.116±0.073）;R₂^* values:（29.820±7.488）Hz >（19.754±6.408）Hz）.Spearman’s rank correlation coefficient of ITSS ratios,R₂^* values and HCC pathological grade was 0.396,0.583（P < 0.05）,the correlation coefficient between R₂^* values and ITSS ratios was 0.327（P< 0.05）.The AUC of ITSS ratios,R₂^* values and the two combined were respectively 0.708,0.862 and 0.885,sensitivity and specificity of（71.4%,62.2%）,（90.5%,70.3%）,and（81.0%,86.5%）.By Delong test,there was significant difference between the AUC of the combination of the two and the AUC of ITSS ratios（Z = 2.043,P = 0.041）,there was no significant difference between the AUC of the combination of the two and the AUC of R₂^* values（Z =1.227,P = 0.220）.Conclusion ITSS automatic extraction based on ESWAN sequence can simply and conveniently evaluate the pathological grade of HCC,and combined with quantitative R₂^*values can improve the accuracy of evaluation greatly.Part 2 Value of R₂^* maps-based texture analysis for evaluating the pathological grade of hepatocellular carcinomaObjective To explore the value of R₂^* maps derived from enhanced T2 star-weighted angiography（ESWAN）sequence-based texture analysis in evaluating the pathological grade of hepatocellular carcinoma（HCC）.Materials and Methods From October 2011 to December 2019,a total of 58 HCC patients（58 HCC lesions）who underwent liver MR examinations（including ESWAN sequence）with liver 1.5T MR scanner preoperatively in our hospital were enrolled in this retrospective study.All the patients enrolled were pathologically confirmed as HCC post-operatively,including well-moderately differentiated HCC group（n = 37）and poorly-differentiated HCC group（n =21）.By using ITK-SNAP software,two radiologists delineated the edges of HCC lesions（including tumor pseudo-capsule）manually,and volume of interests（VOIs）were saved.R₂^*maps were saved as DICOM format,and transmitted to A.K.software（GE health care）to extract and analyses gray level run length matrix（GLRLM）features.Intra-class correlation coefficients（ICC）were used to evaluate the consistency of the two radiologists’ measurements.Normal distribution was tested by Shapiro-Wilks test.The texture features were tested by two independent-sample t test or Mann-Whitney U test.Spearman’s rank correlation test was used to remove redundancy,and features with correlation coefficient greater than 0.90 were removed.Receiver operating characteristic（ROC）analysis was performed to evaluate the diagnostic performance of the texture features which were statistically significant and of the combination of these features.Delong test was used to compare the AUC of any single texture feature and the combination of these features.All statistic analyses were analyzed by SPSS 25.0 software and Medcalc software 19.5.Results The consistency of lesions delineating by two radiologists were in great agreement（ICC > 0.80）.Eight GLRLM features were selected and were statistically significant（P<0.05）.The values of Long Run Emphasis All Direction_offset8_SD,Long Run Low Grey Level Emphasis_All Direction_offset8_SD,Long Run Low Grey Level Emphasis_All Direction_offset4_SD,Long Run Low Grey Level Emphasis_All Direction_offset2_SD,Long Run Low Grey Level Emphasis_All Direction_offset8,Long Run Low Grey Level Emphasis_angle135_offset8,Long Run Low Grey Level Emphasis_angle 90_offset2 and Short Run Low Grey Level Emphasis_angle45_offset8 in poorly-differentiated HCC group were greater than those in well-moderately differentiated group.The AUC of combination of 7 long run-length matrix features features was 0.834,sensitivity of 71.1% and specificity of 76.2%.Conclusion R₂^* values derived from enhanced T2 star-weighted angiography（ESWAN）sequence-based texture analysis can evaluates the heterogeneity of HCC at the pixel level,and has a promising value in evaluation of the pathological grade of HCC.