Preliminary Study Of Kurtosis Model For DWI And T1 Relaxation Time In The Rotating Frame Of Prostate Cancer And Benign Prostate Hyperplasia

Objective:Diffusion weighted imaging based on kurtosis model and T1 Relaxation Time in The Rotating Frame findings of Prostate Cancer(PCa) and Benign Prostate Hyperplasia (BPH) with pathologically confirmed were analyzed retrospectively. To explore the value of three parameters named Dapp, Kapp and T1rho in the diagnosis of prostate cancer and their correlation with Gleason score of prostate cancer.Material and methods:Collect imaging data of patients performed prostate DKI or T1rho imaging from December 2014 to December 2015 in Southern Medical University Nanfang Hospital, including a total of 60 patients.60 patients were divided into two groups. The first group included 30 cases of prostate cancer, among them,19 cases were confirmed by pathology after radical prostatectomy,11 PCa cases occurred in peripheral zone were confirmed by TRUS-guided biopsy. Gleason score was 6-9 points, including GS=3+3:7 cases, GS=3+4:10 cases, GS=4+3:6 cases, GS>4+3:7 cases. The second group included 30 cases of BPH, among them, TRUS-guided biopsy after MRI scan prompted BPH, and both clinical diagnosis and MRI diagnosis were BPH. Prostate cancer cases divided into two types of groups according to the pathological Gleason score. Here were two ways of grouping:(1) divided into three groups:3+3 (7 cases),3+4 (10 cases),>3+4 (13 cases); (2) divided into four groups:3+3 (7 cases),3+4 (10 cases),4+3 (6 cases),>4+3 (7 cases). Prostate cancer of Gleason score= 3+3 were defined as low grade PCa; Prostate cancer of Gleason score larger than 3+4 is defined as intermediate-high grade PCa.The patients underwent prostate examination using the DKI and T1rho sequences on a 3.0 Tesla Philips MR scanner (Philips Healthcare, Best, Netherland). Conventional MRI sequences included axial T1WI, axial and sagittal T2WI, axial DWI, and axial e-THRIVE DCE sequences. DKI adopted single-shot echo planar imaging (SS-EPI) sequence.TR 2000 ms, TE 67 ms, field of view (FOV) 240mmx240mm, flip angle= 90 degrees, matrix= 120 x 160, slice thickness 3.5mm,excitation frequency (NEX) 4, slices 20, diffusion sensitive gradient pulse in X, Y, Z axis direction applied with weighting factors of b= 0,500,1000,1500s/mm2. The whole scanning was about 6 minutes 44 seconds. T1rho imaging adopted turbo field echo imaging (TFE) sequence. TR 3.5ms, TE1.5ms, field of view (FOV) 240mrn×240mm, flip angle= 40 degrees, matrix=118 x 192, slice thickness 3.5mm,excitation frequency (NEX) 1, slices 20, Spin lock frequency= 500 Hz; spin lock time is set to 0,10,20,40,60 ms.The whole scanning was about 5 minutes.The kurtosis model from a DKI sequence was expressed by equation Sb/So=exp (-bDapp+1/6b2Dapp2Kapp). DKI data uploaded to the local postprocessing workstation, and used sofeware named IDL Virtual Machine to automatically generate corresponding Dapp map and Kapp map.Then selected and measured the region of interest (ROI) in the parameter maps directly. T1rho image signal strength meet the formula:S (TSL)= S0*exp (-TSL/T1rho), five spin lock time image data uploaded to the machine processing workstation. Then used the IDL virtual machine software based on the single exponential decay model to generate T1rho-mapping. At last, used ImageJ software to open T1rho-mapping, and draw ROIs to measure T1rho values. Combined with the puncture point and postoperative pathological results describing the fosi location, ROIs were drawn and measureed in all patients to get DapP、Kapp and T1rho values. The positioning and delineation of ROI were perfromed by two radiologists at the same time and reach an agreement. Each lesion in each case did three measurements, and averaged records. Only drew a ROI in parameter map of either Dapp or Kapp, then copied ROI in another parameter map. Try to choose and measure the same ROI as DKI parameter map did in T1rho-mapping using the ImageJ software.Selection of ROIs:combined of postoperative pathological results and puncture point, with reference to signal characteristics of conventional T2WI, DWI and DCE, and tried to avoid transitional zone and peripheral zone at the junction, the seminal root, blood vessels, hemorrhage or calcification etc. Placed a ROI in transitional zone and peripheral zone respectively in each case from BPH group. ROI should be chosen in stromal hyperplasia nodule (often showed nodular or patchy low signal intensity inT2WI). Tried to choose a slice that displayed peripheral zone well, then hand-painted ROI on it, the scope is not exceed peripheral zone boundary, the ROI contains about 20 to 30 pixels, and ROI area is about 5.6*6.0 mm2. Prostate cancer group:delineated ROI in the cancer focus according to the pathological result, the scope is located in the central location of the lesion, and can not exceed the lesion.ROI contains about 20 to 30 pixels, and ROI area was about 5.0*5.6mm2.All statistical analysis was conducted by SPSS software (version 22.0) and Medcalc software. Dapp Kapp and T1rho values were represented by means±standard deviation. Shapiro-Wilk test was used for Normality tests, and Levene test for homogeneity of variance test. Two independent samples t test were in prior carried out between the different disease groups, as well as peripheral zone and transitional zone in BPH group, to campare Dapp、 Kapp and T1rho values. When the data fit the normal distribution but the variance were not equel, Satterthwaite approximation t test were conducted. Nonparametric test of Mann Whitney U test were used when the data fail to meet the normal distribution. Dapp、Kapp and T1rho values in different Gleason score groups of prostate cancer were compared using single factor variance analysis (one way ANOVA).Post hoc multiple comparisons were conducted to campare each two Gleason score groups of prostate cancer using the Bonferroni method and conducted P value correction. Spearman correlation analysis was used to assess the correlation between parameters and Gleason score groups, and observed the magnitude as well as direction of correlation coefficient. ROC (receiver operating characteristic curve) curve was performed to evaluate the predictive ability of the above-mentioned parameters on differentiation between low grade and intermediate-high grade prostate cancer. Estimated the optimal cut-off according to the Youden Index (sensitivity+specificity-1) and the area under the ROC curve (AUC). The confidence interval of average values was set to 95%, P value less than 0.05 was considered to indicate a significant difference in all statistical tests.Results:3.1 DKI features3.1.1 Fitting curve of kurtosis model:At the low values of b (b<500s/mm2), the signal attenuation is quick and the slope of curve is steep; While at the high values of b, the signal attenuation is slow and the slope of curve goes fainter. The curve can acquired good fitting effect for all b values.3.1.2 Two parameters of Dapp、Kapp in PCa and BPHThe prostate cancer group:Kapp was (1.17+0.20)* 10-3, Dapp was (1.32+0.32) * 10-3mm2/s. At b= 1500s/mm2 DWI, tumors showed nodular or flake high signal intensity. Tumors lacated in darker areas compared with contralateral relatively normal tissue in Kapp color map. Tumors lacated in lighter areas compared with contralateral relatively normal tissue in Dapp color map.Kapp and Dapp values of peripheral zone in BPH cases were (0.63±0.087) x 10’3, (2.44±0.24) × 10’3mm2/s, respectively. Kapp and Dapp values of stromal hyperplasia nodules were (0.92±0.19) ×10-3, (1.85±0.17) ×10-3mm2/s, respectively. Two independent samples t test shown that Kapp value in peripheral zone was lower than stromal hyperplasia nodules in BPH and Dapp value in peripheral zone was high er than stromal hyperplasia nodules in BPH. The difference of the two parameters in the two different anatomical regions was statistical significant (P=0.000). Inhomogeneous slightly low signal intensity in stromal hyperplasia nodules were showed in Kapp color map and intermediate signal intensity in Dapp color map. Uniform low signal intensity in the peripheral zone was showed in Kapp color map and uniform high signal intensity in Dapp color map.Satterthwaite approximate t-test shown Kapp value in tumor was higher than KaPP value in peripheral zone of BPH, and the difference between them was statistical significant (P= 0.000). Mann Whitney U test shown KaPP value in tumor was higher than Kapp value in stromal hyperplasia nodules of BPH, and the difference between them was statistical significant (P= 0.000).Two independent samples t test shown Dapp value in tumor was lower than Dapp value in peripheral zone of BPH, and the difference between them was statistical significant (P= 0.000); Dapp value in tumor was lower than KaPP value in stromal hyperplasia nodules of BPH, and the difference between them was statistical significant (P= 0.000);3.1.3 The comparison of Dapp and KaPP in different prostate cancer Gleason score groups.To divide prostate cancer cases into 3 or 4 Gleason score groups. Single factor variance analysis showed either three or four groups, Dapp and KaPP in Gleason score groups accord with normal distribution and homogeneity of variance. One-way ANOVA showed that the difference of Dapp and Kapp among different Gleason score groups was statistical significant (P= 0.000). When Gleason score groups were three: GS= 3+3,3+4,>3+4, multiple comparisons between groups with Bonferroni method showed the difference of Dapp and Kapp between each two Gleason score groups was statistical significant (P= 0.000-0.015); When Gleason score groups were four:GS= 3+3,3+4 and 4+3,>4+3. The difference of Dapp and KaPP between GS= 3+4/4+3 groups was statistical insignificant (P=0.261 and P=0.740, respectively). The difference of Dapp and Kapp between GS= 4+3/>4+3 groups was statistical insignificant (P= 1.000 and P= 0.180,respectively), besides, the difference of Dapp between GS= 3+4/>4+3 groups was statistical insignificant (P=0.049).3.1.4 The Correlation between Kapp (or Dapp) and Gleason score groupsSpearman correlation analysis showed that Kapp increased gradually when prostate cancer Gleason score increased, and a positive correlation between them. When Gleason score groups were 3, ρ=0.815 (P=0.000), when Gleason score groups were 4, ρ= 0.841 (P=0.000); Spearman correlation analysis showed that Dapp decreased gradually when prostate cancer Gleason score increased, and a negative correlation between them. When Gleason score groups were 3, ρ=-0.803 (P= 0.000), when Gleason score groups were 4, p=-0.771 (P= 0.000).3.1.5 ROC curve analyze the performance of Kapp, Dapp in the discrimination between low-grade and intermediate/high-grade prostate cancerROC curve analysis obtained the cut-off of Kapp value discriminating GS=3+3 from GS≥3+4 prostate cancer was 1.08 (10-3),as well as the area under the ROC curve (AUC) was 0.988 (Z=32.293, P< 0.0001), sensitivity was 91.3%, specificity was 100%.The cut-off Dapp values discriminating GS=3+3 from GS≥3+4 prostate cancer was 1.44 (10-3 mm2/sec), AUC was 0.947 (Z=11.007, P< 0.0001), sensitivity was 91.3%, specificity was 85.7%.3.2 T1rho imging features3.2.1 Signal intensity of T1rho imagesFive groups of T1rho images with different spin lock time (TSL=0 10,20,40,60 ms) and T1rho-mapping based on single exponential decay model fitting could distinguish prostate and surrounding structures satisfactorily,as well as the peripheral zone and central gland. But behaved not well to distinguish the central zone and transitional zone.The boundary of some hyperplastic nodules with larger volume could be identified. With the increasing of spin lock time, the signal to noise ratio (SNR) of T1rho image is reduced, and the SNR of TSL=0ms is the highest.3.2.2 Parameter of T1rho in BPHIn BPH cases, the thickness of the peripheral zone is related to the degree of central gland hyperplasia. When the peripheral zone is compressed, the T1rho image will not be able to identify the peripheral zone clearly. Peripheral zone tissue on T2WI showed homogeneous high signal intensity, but showed homogeneous or inhomogeneous low signal intensity in each spin lock T1rho relaxation diagram and in T1rho-mapping.The T1rho value of ROIs was 0ms or vary greatly with different ROI.The T1rho value of stromal hyperplastic nodules was 68.65+9.51 ms and showed a low signal intensity in T2WI.3.2.3 Parameter of T1rho in prostate cancerTumor in the peripheral zone was not easy to find out in T1rho images when peripheral zone or tumor size is small. When the peripheral zone was thicker with obviously long T2 signal, tumor performed slightly higher signal in a low signal background. It was difficult to identify prostate cancer foci from hyperplastic nodule in transitional zone. T1rho value of tumor was 65.33+7.68 ms.3.2.4 The Comparison of T1rho value in BPH and PCaTwo independent samples t test showed parameter T1rho had no statistically significant difference between PCa and hyperplastic nodule in transitional zone (P=0.142)Conclusions:1. Kapp value is higher in prostate cancer than in benign prostatic hyperplasia, but Dapp is just the opposite. DKI sequence can be used to identify prostate cancer and benign prostatic hyperplasia.2. Kapp was positively correlated with Gleason score, and Dapp were negatively correlated with Gleason score. DKI parameters can help to predict PCa differentiation.3. DKI have potential to differentiate low and intermediate-high grade prostate cancer, but failed to differentiate intermediate and hign grade prostate cancer, as well as prostate cancer with Gleasen score of 3+4 and 4+3.4. T1rho imaging failed to differentiate prostate cancer and stromal hyperplasia nodules. Peripheral zone in BPH with long T2 signal but showed very low signal intensity in T1rho imaging (T1rho= Oms), which suggested T1rho imaging in our study might not be suitable for the diagnosis of prostate cancer.