| Part I Analysis of texture features of MR-T2 WI prostate cancerObjective:The quantitative parameters of PCa and benign prostatic hyperplasia(BPH)was compared by texture analysis of Prostate MRI Axial and high resolution T2-weighted imaging to analyze the difference of the texture characteristics of T2 WI prostate cancer and to explore the value of T2 WI texture analysis in prostate to differentiate PCa and BPH.Materials and Methods: To collect the patients of the Second Affiliated Hospital of Suzhou University diagnosed the clinical prostate disease from January 2015 to December 2016 and gave them multiparameter magnetic resonance imaging(mp-MRI)examination,and obtained the results of the diagnosis of prostate biopsy Case data of 300 cases,including 140 cases of prostate cancer,benign prostatic hyperplasia in 160 cases.According to each case of puncture pathological results,the prostate cancer zone and the benign proliferative region were predefined in their corresponding Axial and high resolution T2 WI images by 2 urological sub-professional imaging diagnostic physicians(both by PI-RADS v2 application training,prostate mp-MRI diagnosis experience were 2 years and 1 year respectively),and the region of interest(ROI)was set up.By delineating the cancer group along the edge of the cancer layer(as far as possible including the entire foci,but not beyond the scope of cancer)and benign hyperplasia group along the edge of the proliferative noduleand(as far as possible including the entire proliferative nodules,but not beyond the scope of proliferative nodules),the three-dimensional volum of interest(VOI)of the whole lesion of the cancer group and the benign hyperplasia group were obtained.The texture analysis quantitative parameters such as entropy,mean value,standard deviation,nonuniformity,peak value and skewness were obtained by texture analysis of VOI by Fire Voxel software.The quantitative parameters of the texture analysis between prostate cancer group and benign hyperplasia group were compared by independent sample t test while multivariate parameter analysis of T2 WI texture parameters were performed by Logistic regression analysis,and the receiver operator characteristic(ROC)was used to obtain the area under the curve(AUC),to determine the quantitative parameters of the texture analysis and threshold and diagnostic efficacy of related parameters diagnosing prostate cancer and benign proliferation.Results:1.The entropy,mean value,standard deviation,peak value,skewness and nonuniformity of prostate cancer group were 3.97 ± 0.28,89.95 ± 24.5,14.83 ± 5.41,0.42 ± 0.52,1.44 ± 2.71,0.17 ± 0.04 in the MR-T2WI(Axial,high resolution)image texture analysis parameters included in the study case.Meanwhile,the above parameters of benign proliferative group were 4.13 ± 0.23,114.31 ± 43.93,17.65 ± 7.29 and 0.59 ± 0.41,1.38 ± 1.57,0.15 ± 0.02.There were statistical differences in entropy,mean value,standard deviation,nonuniformity and peak value between the prostate cancer group and the benign hyperplasia group(t = 5.466,6.023,3.767,3.182,-2.818,P <0.05)).There was no statistical difference in the quantitative parameter skewness between the two groups(t =-0.239,P> 0.05).2.The AUC values and diagnostic efficiencies of the MR-T2WI(Axial,high resolution)image texture analysis included in the study group were: the AUC value of the cancer group and benign hyperplasia group identified by entropy was 0.696,and when the threshold was 4.05,the sensitivity of the entropy was 66.4%,the specificity was 69.4% and the accuracy was 68.0%.the AUC value of the cancer group and benign hyperplasia group identified by the mean value was 0.804,and when the threshold was 94.34,the diagnostic sensitivity was 73.6 %,the specificity was 80.0% and the accuracy was 77.0%.the AUC value of the cancer group and benign hyperplasia group identified by standard deviation was 0.674,and when the threshold was 13.92,the diagnostic sensitivity was 56.4%,the specificity was 76.2% and the accuracy was 70.0%.the AUC value of the cancer group and benign hyperplasia group identified by peak value was 0.631,,and when the threshold was 0.37,the sensitivity was 44.3%,the specificity was 69.4% and the accuracy was 57.7%.the AUC value of the cancer group and benign hyperplasia group identified by nonuniformity was 0.565,and when the threshold was 0.17,the sensitivity was 35.7%,the specificity was83.1% and the accuracy was 61.0%.3.Logistic regression analysis was used to analyze the quantitative parameters of T2 WI texture,The results of multi-parameter analysis of AUC and diagnostic efficacy of cancer group and benign hyperplasia group were as follows: the AUC value,sensitivity,specificity and accuracy of the two groups identified by entropy and mean value were: 0.799,67.9%,86.3% and 77.7% respectively.The AUC values,sensitivity,specificity and accuracy of the two groups identified by entropy and standard deviation were 0.696,67.9% 68.7% and 68.3% respectively.The AUC values,sensitivity,specificity and accuracy of the two groups identified by entropy and peak value were: 0.709,57.9%,79.4% and 69.4% respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by entropy and nonuniformity were 0.848,70.0%,83.8% and 77.4% respectively.The AUC values,sensitivity,specificity and accuracy of the mean value and standard deviation of the two groups identified by mean value and standard deviation were 0.802,57.9%,79.4% and 69.4% respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by mean value and peak value were: 0.795,77.1%,73.1% and 75.0% respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by standard deviation and peak value were: 0.682,50.0%,81.9% and 67.0% respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by peak value and nonuniformity were: 0.665,49.3%,82.5% and 67.0% respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by entropy and mean value and standard deviation were: 0.848,68.6%,93.7% and 82.0% respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by entropy and mean value and peak value were: 0.791,64.3%,85.0% and 75.3% respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by entropy and mean value and peak value were: 0.791,64.3%,85.0% and 75.3% respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by entropy and mean value and nonuniformity were: 0.849?74.3%?86.3%?80.7%respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by entropy and standard deviation and peak value were:0.710?57.9%?79.4%?69.4%respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by entropy,mean value and standard deviation and nonuniformity were:0.851?70.7%?90.0%?81.0%respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by entropy and peak value and nonuniformity were:0.862?87.9%?73.8%?80.4%respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by mean value and standard deviation and peak value were:0.795?67.1%?84.4%?76.3%respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by mean value and standard deviation and nonuniformity were:0.808?70.7%?82.5%?77.0%respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by mean value and peak value and nonuniformity were:0.794?67.9%?84.4%?76.7%respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by standard deviation and peak value and nonuniformity were:0.785 ? 63.6 % ? 87.5 % ? 76.3 % respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by entropy and mean value,and standard deviation and peak value were:0.861?80.0%?81.9%?81.0%respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by entropy and mean value,and standard deviation and nonuniformity were:0.850?68.6%?92.5%?81.3%respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by entropy and standard deviation and peak valueand nonuniformity were:0.798?70.7%?82.5%?77.0%respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by mean value and standard deviation and peak value and nonuniformity were:0.867?84.3%?76.2%?80.0%respectively.The AUC value,sensitivity,specificity and accuracy of the two groups identified by entropy and mean value and standard deviation and peak value and nonuniformity were:0.867?87.9%?73.8%?80.4%respectively.Conclusions:The results of quantitative analysis of MR-T2WI(Axial,high resolution)image analysis of prostate cancer and benign hyperplasia patients show that: 1.The entropy,mean value and standard deviation of prostate cancer group are lower than those of benign hyperplasia group while the peak value and unevenness are higher which indicate that there are statistical difference between the two groups.2.MR-T2 WI image texture single quantitative parameter analysis shows that the mean value to identify prostate cancer group and benign hyperplasia group has the highest efficiency,and the AUC value was 0.804 and when the threshold was 94.34,the sensitivity is 73.65%,the specificity is 80.8% and the accuracy is 77.0%.3.The results of MR-T2 WI image texture several quantitative parameters analysis show that:(mean + standard deviation + peak value + nonuniformity)and(entropy + mean value + standard deviation + peak value + nonuniformity)have the highest efficacy to identify the cancer group and the benign proliferative group.AUG are all 0.867,the sensitivity are 84.3%,87.9% respectively,the specificity are 76.2%,73.8% respectively,and the accuracy are 80.0%,80.4% respectively.Thus,prostate cancer group and benign hyperplasia group MR-T2WI(Axial,high resolution)image texture analysis quantitative parameters are different which help to identify prostate cancer and benign hyperplasia.Part ? MR-T2 WI image texture analysis quantitative parameters for identifying different GS score group of prostate cancerObjective:The difference of the quantitative parameters between the low-GS group and the medium-high GS group was analyzed by texture analysis of MR-T2 WI images of prostate cancer,and to explore the value of T2 WI texture quantitative parameters analysis to differentiate low-GS and the medium-high GS groups of prostate cancer.Materials and Methods: The puncture pathological results of total 140 patients from the second hospital of the Second Affiliated Hospital of Suzhou University from January 2015 to December 2016 with clinical diagnosis of prostatic diseases were collected and then gave them multiparameter magnetic resonance imaging(mp-MRI)examination.The results of the diagnosis were as follows: GS ? 3 + 3 were included in the low GS group,the diagnosis of puncture pathology GS> 3 + 3 were included in the high GS group,including 40 cases of low GS group,100 cases of high GS group.According to each case of puncture pathological results,the prostate cancer zone and the benign proliferative region were predefined in their corresponding Axial and high resolution T2 WI images by 2 urological sub-professional imaging diagnostic physicians(both by PI-RADS v2 application training,prostate mp-MRI diagnosis experience were 2 years and 1 year respectively),and the region of interest(ROI)was set up.By delineating the cancer group along the edge of the cancer layer(as far as possible including the entire foci,but not beyond the scope of cancer)and benign hyperplasia group along the edge of the proliferative noduleand(as far as possible including the entire proliferative nodules,but not beyond the scope of proliferative nodules),the three-dimensional volum of interest(VOI)of the whole lesion of the cancer group and the benign hyperplasia group were obtained.The texture analysis quantitative parameters such as entropy,mean value,standard deviation,nonuniformity,peak value and skewness were obtained by texture analysis of VOI by Fire Voxel software.The difference of the quantitative parameters between the low-GS group and the medium-high GS group were compared by independent sample t test and the receiver operator characteristic(ROC)was used to obtain the area under the curve(AUC),to determine the threshold and diagnostic efficacy of quantitative parameters of the texture analysis to differentiate low-GS and the medium-high GS groups of prostate cancer.Results:1.The research results of the value of T2WI(Axial,high resolution)texture quantitative parameters analysis to differentiate low-GS and the medium-high GS groups showed that the entropy,mean value,standard deviation,peak value,skewness and nonuniformity of low-GS group were 4.03±0.29?100.63±28.51?16.79±6.56?0.51±0.57?1.16±2.91?0.17±0.04 respectively.The entropy,mean value,standard deviation,peak value,skewness and nonuniformity of the medium-high GS group were 3.95 ± 0.27,85.68 ± 21.41,14.04 ± 4.69 and 0.38 ± 0.49 1.56 ± 2.63,0.17 ± 0.04 respectively.The mean value and standard deviation had statistically significant(P = 2.994,2.413,P <0.05)in differentiating the low GS group and the middle and high GS group.There was no statistically significant difference for entropy,peak,skewness and nonuniformity to differentiate the low GS group and the middle and high GS group(t = 1.612,1.36,-0.772,0.105,P> 0.05).2.The AUC values and diagnostic efficiencies of the function of Part2 MR-T2WI(Axial,high resolution)image texture analysis quantitative parameters for identifing different GS score group of prostate cancer were: the AUC value of identifing different GS score group identified by mean value was 0.704,and when the threshold was 93.86,the sensitivity was 55.0%,the specificity was 82.0% and the accuracy was 73.0%.the AUC value of identifing different GS score group identified by standard deviation was 0.648,and when the threshold was 18.07,the sensitivity was 40.0%,the specificity was 88.0% and the accuracy was 73.3%.Conclusions:The research results of the value of T2WI(Axial,high resolution)texture quantitative parameters analysis to differentiate low-GS and the medium-high GS groups showed that mean value and standard deviation those of low-GS group are higher than those of medium-high GS group,and the statistics difference is significant and diagnostic efficacy of the mean value identifing different GS score group is higher than those of standard deviation.The AUC value is 0.704.When the threshold is 93.86,the diagnostic sensitivity is 55.0% and the specificity is 82.0% and the accuracy is 73.8%.It can be seen that the mean value and standard deviation of quantitative parameters in MR-T2WI(Axial,high resolution)image analysis are different between low GS group and middle and high GS group in prostate cancer,which is helpful to the differential diagnosis of low GS group and high GS group in prostate cancer. |
PDF Full Text Request