Study On The Correlation Between Microvascular Perfusion Evaluated By VIVM DW-MRI With Low B Value,MVD,and Parameters Of CT Perfusion

Objective(1) With rabbit liver VX2 tumor, to study the influence of different tube voltage, tube current and newer iterative reconstruction algorithm to the radiation dose, image quality and perfusion parameters of hepatic computed tomography perfusion imaging (CTPI), to evaluate the feasibility and accuracy of hepatic CTPI with low scanning parameters, and evaluate the correlation between parameters derived from CTPI with microvessel density (MVD).(2) To explore the influence of different size range and number of b value to parameter derived from intravoxel incoherent motion DW-MRI (IVIM DWI), and to evaluate the correlation between them and CTPI parameters, MVD.MethodsTo build rabbit liver VX2 tumor model by injecting VX2 tumor tissue suspension (presented by ultrasound department of southern hospital of southern medical university) under open surgery, and a total of 25 New-Zealand white rabbit liver VX2 tumor model were successfully established.Three hepatic CTPI were underwent with different scanning parameter (protocolA:tube voltage 80kV, tube current 40 mAs; protocol:tube voltage 80kV, tube current 80 mAs; protocol C:tube voltage 100kV, tube current 80 mAs) for per animal model by using 256-slice spiral CT scanner, and other scanning parameters were unchanged. All CTPI raw data were reconstructed by traditional filtered back projection (FBP) and iterative reconstruction (iDose4 Level3) respectively. With Functional CT software package, to measure the average CT value, image noise and to calculate the image signal noise ratio (SNR) and contrast noise ratio, to measure the hepatic arterial perfusion (HAP), hepatic portal perfusion (HPP), total perfusion (TP), hepatic arterial perfusion index (HPI), blood flow, peak of enhancement index (PEI), time of transit peak (TTP) and blood volume of liver VX2 tumor and normal liver parenchyma on three consecutive maximum image of VX2 tumor. To record the dose-length product (DLP) and to calculate the effective radiation dose of each hepatic CTPI. To compare the difference of radiation dose, objective image quality and perfusion parameters among three CTPI scanning protocols by using repeated measurement data with LMMs and LSD method. To compare the difference of objective image quality and perfusion parameters of hepatic CTPI between FBP and iDose4 Level3 by using pared-sample T test. To evaluate the correlation between CTPI parameters and MVD by using Pearson.correlation analysis.Conventional T1WK T2WI and IVIM DW-MRI were underwent by 1.5 T MRI scanner and special rabbit coil; b values were 0,10,20,30,40,50,75,100,150,300, 500 and 800 s/mm2 (protocol A); or 0,10,20,30,40,50,75,100,150,300,500,800, 1000 and 1200 s/mm2 (protocol B) or 0,20,40,75,100,150,300 及 800 s/mm2 (protocol C). PRIDE DWI Tool and Image J software were used for IVIM and mono-exponential analysis. ROI was manually placed on each liver VX2 lesion at the level of maximum transverse diameter of lesions and were chosen to be as large as possible, consistent with minimal contaminations from surrounding unintended tissues. The ADC value, and D value, f value and D* value were measured by two independent observers. The difference of D value, f value and D* value derived from IVIM DW-MRI and ADC value were analyzed by repeated measurement data with LMMs. The correlation between D value, f value and D* value derived from IVIM DW-MRI and ADC value, and CTPI parameters, MVD were analyzed.All experiment ainimals were given intravenous excess 20% urethane for euthanasia, and immediately followed autopsy for rabbit liver VX2 tumor specimens for HE stain and CD34 immunohistochemical stain, and to count MVD under with 400 times.Results(1) The results of specimens and pathological histologyThe rabbit live VX2 tumors showed single lesion for all animal modess, locating in left lobes in 18 cases and right lobes in 7 cases. The VX2 tumors demonstrated gray fish shape without necrosis, round or oval without a capsule, and with clear boundary between dark red normal liver tissue by naked-eye observing. HE stain showed diffuse tumor cell were arranged with irregular arrangement, but clear obvious boundary between normal liver parenchyma. CD34 stains showed vascular endothelial cell as tan or brown, VX2 tumor cells as blue, and the MVD were about 9.07±2.41 (6-15).(2) The results of CTPICTPI of protocol A were finished in 23 case; protocol B in 24 case, but 2 case were eliminate from experimental group because of severe movement artifacts; protocol C were in 23 cases.1. Radiation doseThe radiation dose of protocol C were highest (DLP:2382.81±1.59 mGy*cm; ED:35.74±0.02 mSv), then protocol B (DLP:1273.86±1.15 mGy*cm; ED:19.11 ±0.18 mSv), protocol A (DLP:632.81±1.07 mGy*cm; ED:9.49± 0.16mSv) were lowest. LMMs showed that there were significantly statistical difference of radiation dose among three protocol (DLP:F=9244245, P=0.000; ED:F=8660468, P=0.0000), and so do between each two (P=0.000).2. Objective image quality2.1 The average CT value of liver tumor VX2 tumorNo matter iDose4 Level3 or FBP reconstruction, LMMs demonstrated there were significantly statistical difference of average CT value of rabbit liver VX2 tumor among three protocol (F=11.184,P=0.000; F=14.651, P=0.000), and the average CT value of protocol C was obviously lower than that of protocol A and B, with significantly statistical difference (P<0.05), the average CT value of protocol B was slightly lower than that of protocol A without significantly statistical difference (P>0.05). Pared-sample T test showed the average CT value of rabbit liver VX2 tumor of all three protocol with FBP reconstruction was obviously higher than that with iDose4 Level3 reconstruction with significantly statistical difference (protocol A:F=8.468,P=0.000; protocol B:F=9.182, P=0.000; protocol C:F=13.264, P=0.000).2.2 The image noise of liver tumor VX2 tumorNo matter iDose4 Level3 or FBP reconstruction, LMMs demonstrated there were significantly statistical difference of image noise of rabbit liver VX2 tumor among three protocol (F=8.131, P=0.001; F=14.845, P=0.000), and the image noise of protocol C was obviously lower than that of protocol A and B, with significantly statistical difference (P<0.05), the image noise of protocol A was slightly higher than that of protocol B, but there was no significantly statistical difference (P>0.05). Pared-sample T test showed the image noise of rabbit liver VX2 tumor of protocol A and B with FBP reconstruction was obviously higher than that with iDose4 Level3 reconstruction with significantly statistical difference (protocol A: F=2.715, P=0.013; protocol B:F=3.813, P=0.001), imoge nosie of protocol C with FBP were slightly higher than that with iDose4 Level3 reconstruction without significantly statistical difference (F=0.626, P=0.538).2.3 The SNR of liver VX2 tumorNo matter iDose4 Level3 or FBP reconstruction, LMMs demonstrated there were no significantly statistical difference of SNR of rabbit liver VX2 tumor among three protocol (F=01, P=0.5; F=13, P=0.272). Pared-sample T test showed the SNR of rabbit liver VX2 tumor of all three protocol with FBP reconstruction was lower than that with iDose4 Level3 reconstruction, but there was significantly statistical difference only for protocol B (F=2.525, P=0.020), but no significantly statistical difference for protocol A and C (F=1.537, P=0.140; F=0.634, P=0.533).2.4 The CNR of liver VX2 tumorNo matter iDose4 Level3 or FBP reconstruction, LMMs demonstrated there were no significantly statistical difference of CNR of rabbit liver VX2 tumor among three protocol ((?)6 P (?) (?);F=(?) P=(?) (?)). Wilcoxon Pared-sample test showed the CNR of rabbit liver VX2 tumor of all three protocol with with iDose4 Level3 reconstruction was slightly higher than that with FBP reconstruction, but there were no significantly statistical difference (protocol A:Z=1.651, P=0.099; protocol B: Z=0.643, P=0.520; protocol C:Z=1.269, P=0.205).2.5 The average CT value of normal liver parenchymaNo matter iDose4 Level3 or FBP reconstruction, LMMs demonstrated there were significantly statistical difference of average CT value of normal liver parenchyma among three protocol(F=(?),P=0.000; F=13, P=0.000), and the average CT value of protocol C was obviously lower than that of protocol A and B, with significantly statistical difference (P<0.05), the average CT value of protocol A was slightly higher than that of protocol B, but there was no significantly statistical difference (P>0.05). Pared-sample T test showed the average CT value of normal liver parenchyma of all three protocol with FBP reconstruction was obviously higher than that with iDose4 Level3 reconstruction with significantly statistical difference (protocol A:F=5.022, P=0.000; protocol B:F=3.833, P=0.001; protocol C:F =3.965, P=0.001).2.6 The image noise of normal liver parenchymaNo matter iDose4 Level3 or FBP reconstruction, LMMs demonstrated there were significantly statistical difference of image noise of normal liver parenchyma among three protocol (F==181, P=0.000; F=(?), P=0.000), and the image noise of protocol C was obviously lower than that of protocol A and B, with significantly statistical difference (P<0.05), the image noise of protocol A was slightly higher than that of protocol B without significantly statistical difference(P>0.05). Pared-sample T test showed the image noise of normal liver parenchyma of all three protocol with FBP reconstruction was obviously higher than that with iDose4 Level3 reconstruction with significantly statistical difference (protocol A:F=2.715, P=0.013; protocol B:F =3.813, P=0.001; F=2.62,5, P=0.016).2.7 The SNR of normal liver parenchymaNo matter iDose4 Level3 or FBP reconstruction, LMMs demonstrated there were significantly statistical difference of SNR of normal liver parenchyma among three protocol (F=1556, P=0.010; F=86, P=0.003), and SNR of protocol C were.slightly higher than that of protocol B without significantly statistical difference (P>0.05) by using FBP, but with significantly statistical difference by using iDose4 Level3 (P=0.033); SNR of protocol C were.obviously higher than that of protocol A with significantly statistical difference (P<0.05); SNR of protocol B were slightly higher than that of protocol A without significantly statistical difference (P>0.05). Pared-sample T test showed the SNR of normal liver parenchyma of all three protocol with FBP reconstruction was obviously lower than that with iDose4 Level3 reconstruction with significantly statistical difference (protocol A:F=4.596, P =0.000; protocol B:F=2.373, P=0.028; protocol C:F=3.678, P=0.001).2.8 The CNR of normal liver parenchymaNo matter iDose4 Level3 or FBP reconstruction, LMMs demonstrated there were no significantly statistical difference of CNR of normal liver parenchyma among three protocol(F=39 P(?) (?) F(?) (?) P(?) (?)). Wilcoxon Pared-sample test showed the CNR of normal liver parenchyma of all three protocol with with iDose4 Level3 reconstruction was slightly higher than that with FBP reconstruction with significantly statistical difference (protocol A:Z=3.806, P=0.000; protocol B: Z=1.894, P=0.048; protocol C:Z=2.416, P=0.016).3. The parameters of CTPI3.1 The CTPI parameters of liver VX2 tumorUsing iDose4 Level3 reconstruction, LMMs demonstrated there were no significantly statistical difference of HAP, HPP, TP, HPI, BF, PEI, TTP and BV of rabbit liver VX2 tumor among three protocol (P>0.05).Using FBP reconstruction, LMMs demonstrated there were no significantly statistical difference of HAP, HPP, TP, HPI, BF, TTP and BV of rabbit liver VX2 tumor among three protocol (P>0.05); there were significantly statistical difference of PEI; PEI of protocol C was obviously lower than that of protocol A and B (P <0.05), but without significantly statistical difference between protocol A and B (P>0.05).Pared-sample T test showed there were no significantly statistical difference of HAP, HPP, TP, HPI, BF, PEI, TTP and BV of rabbit liver VX2 tumor of all three protocol between iDose4 Level3 reconstruction and FBP reconstruction(P>0.05).3.2 The CTPI parameters of normal liver parenchymaUsing iDose4 Level3 reconstruction, LMMs demonstrated there were no significantly statistical difference of HAP, HPP, TP, HPI, BF, TTP and BV of normal liver parenchyma among three protocol (P>0.05), but did not do so of PEI (F=38.795, P=0.000); PEI of protocol C was obviously lower than that of protocol A and B (P <0.05), but did not do so between protocol A and B (P>0.05).Using FBP reconstruction, LMMs demonstrated there were no significantly statistical difference of HAP, HPP, TP, HPI, BF, TTP and BV of normal liver parenchyma among three protocol (P>0.05), but did not do so of PEI (F=41.344, P =0.000); PEI of protocol C was obviously lower than that of protocol A and B (P <0.05), but did not do so between protocol A and B (P>0.05).Pared-sample T test showed there were no significantly statistical difference of HAP, HPP, TP, HPI, BF, PE, TTP and BV of normal liver parenchyma of all three protocol between iDose4 Level3 reconstruction and FBP reconstruction(P>0.05).4. The correlation between CTPI parameters and MVD of liver VX2 tumorNo matter which reconstruction algorithm, no matter which CTPI protocol, Pearson correlation analysis demonstated that there were best correlation in general between MVD and HAP, TP, BF of rabbit liver VX2 tumor, especially HAP, with positive linear relationship; there were no correlation or poor correlation in general between MVD and HPI, PEI, TTP, BV of rabbit liver VX2 tumor.(3) The results of IVIM DW-MRIIn our study, IVIM DW-MRI examinations were performed in 16 rabbit liver VX2 tumor model.1. Conventional MRI examinationConventional MRI demonstrated single lesion in all 16 liver VX2 tumor model, of which located in left lobe in 10 cases, in right lobe in 6 cases. The liver VX2 tumor showed low signal with clear boundary on TiWI, and slightly high signal with clear boundary on T2WI and T2WI fat suppression sequence. To measure on the T2WI, the average maximum diameter of 16 cases of rabbit liver VX2 tumor was about 17.02± 7.55 mm.2. The signal change and signal attenuation curve of IVIM DW-MRIIVIM DW-MRI demonstrated that the diffusion of all liver VX2 tumor were restricted, with high signal, and no matter the number of b value, the signal gradually increased with the increase of b value, but so do the image noise, and SNR decreased. The signal attenuation curves of rabbit liver VX2 tumor among three b value protocol were similar, with nonlinear attenuation process, especially when b value was< 200 s/mm2, but with basic linear attenuation process when b value was> 200 s/mm2.3. The total acquisition timeThe total acquisition time of protocol A was 263.0s; protocol B 630.2 s; protocol C 158.1s.4. Quantitative parameters derived from IVIM DW-MRI and ADC valueLMMs showed there was no significantly statistical difference of D value, f value, D* value and ADC value among three different b value protocol (P>0.05). The absolute value of D value, f value, D* value and ADC value of protocol B were relatively smaller than those of protocol A and C, the latter were relatively more close.5. The correlation between quantitative parameters derived from IVIM DW-MRI and CTPI parametersNo matter iDose4 Level3or FBP reconstruction, the Pearson correlation analysis between quantitative parameters (D value, f value, D* value and ADC value) derived from IVIM DW-MRI with protocol A and CTPI parameters (HAP, HPI, BF, BV) with three protocol were performed respectively. In general, there was best correlation between D value and HAP, with linear relationship, not good enough correlation between D value and HPI, BF, BV respectively. There was best correlation between f value and BV, with linear relationship, not good enough correlation between f value and BF, and no correlation between f value and HAP, HPI. There were no correlation between D* and HAP, HPI, BF, BV respectively. There was no correlation between ADC value and HPI, and not good enough correlation between ADC value and HAP, BF, BV respectively.6. The correlation between quantitative parameters derived from IVIM DW-MRI and MVDPearson correlation analysis showed there was median correlation between D value derived from IVIM DW-MRI of protocol A and MVD, with linear relationship (r=0.509, P=0.044). But there was no correlation or poor correlation between f value, D*value, ADC value and MVD respectively (r=0.281, P=0.291; r=0.099, P=0.715; r=0.359, P=0.172).Conclusion(1) It is feasible and accurate to perform CTPI with low radiation dose1. There is no significant influence of tube current to tissue CT value, while there is significant influence of tube voltage to tissue CT value, the tissue CT value decrease with tube voltage increasing, and increase with tube voltage decreasing;2. The CTPI parameter with low tube voltage, low tube current (low radiation dose) is same to that with high tube voltage, high tube current (high radiation dose);3. Iterative reconstruction (iDose4) can be able to decrease image noise, to increase image signal noise ratio and contrast noise ratio and to improve image quality for progressively deceasing scanning parameters and radiation dose, while there is no obvious influence to CTPI parameters.4. CTPI parameters can be used in evaluating the angiogenesis in tumor4.1 There is high correlation between HAP, TP,BF and MVD, but no correlation between HPI, BV and MVD;4.2 The combination of HAP, HPI and BF could be used as sensitive index for MVD and arterialization of angiogenesis in tumor;4.3 The relationship between BV and pathological features of tumor angiogenesis is very complicated, which remain to be further research.(2) To a certain extent, IVIM DW-MRI can be used to reflect microcirculation perfusion, and to separate the microcirculation perfusion from pure water molecule diffusion1. There is high correlation between f value and BV derived from CTPI, so f value could be used to reflect tissue BV;2. There is no correlation between D* value and CTPI parameters, MVD, so it could not considered that D* could reflect microcirculaiton, and its specific function and mechanism still needs to be further study;3. The ADC value is higher than D value, and there is median correlation between both of them and MVD, but it still needs to further study the mechanism and influence factors;4. The influence of the size range and number of b value to quantitative parameters derived from IVIM DW-MRI and ADC value is not obvious;5. The higher the b value and the greater number the b value, the longer the acquistion time; To choose 8 b value at most and about 500 s/mm2 at maximum is recommended.