A Comparative Study Based On IVIM-DW MRI And DSC-MRI In Cerebral Lymphoma

[Objective]The aims of our study were as follows:1) To differentiate primary cerebral lymphoma from high grade astrocytoma and meningioma using IVIM and DSC perfusion imaging.2) To analyze the correlation between perfusion parameters derived from IVIM-MRI and DSC-MRI perfusion imaging.[Materials and Methods]1.ParticipantsWe collected 65 patients with skeptical brain tumor in our hospital during November 2014 to November 2015. All the patients underwent the bi-exponential model IVIM MRI and dynamic susceptibility contrast-enhanced MRI scan of the whole brain.The diagnoses of all patients were confirmed by stereotactic biopsy pathology or surgical pathology.20 cases were high grade astrocytoma (14 neuroblastoma,6 anaplastic astrocytoma),20 cases were meningioma and 11 cases were primary central nervous system lymphoma.2. MR Imaging AcquisitionAll experimental subjects who met the inclusion criteria underwent magnetic resonance scan in our hospital. All patients underwent conventional head MRI (including axial, sagittal andcoronal) multiple b values DWI based on bi-exponential model IVIM, DSC-MRI and the postperfusion enhanced scan (including axial, sagittal and coronal imaging). All the sequences were performed on a Philips Archieve 1.5T (Philips Healthcare, Best, Netherland) magnetic resonance scanner with an 8-channel standard head coil. Conventional MRI imaging sequence included: axial T1WI, transverse T2WI, axial T2-FLAIR and sagittal T2WI.Before the intravenous injection of contrast material, the multiple b values diffusion-weight imaging was performed using single-shot echo planar spin echo (Spin Echo Planner Imaging SS-EPI) sequence and parallel capture spatial sensitivity encoding technology (ASSET). ForIVIM-DWI,13 b values (0,10,20,30,50,80,100,150,200,300,400,600,800s/mm2) were applicated in three orientations, along the x, y, z axes.Dynamic susceptibility contrast-enhanced perfusion scans:SE-EPI sequence scanning parameters are:TR/TE=2000/30ms,FOV=24cmx24cm, matrix=64x24, resolution=3.75mm×3.75mm, slice=4mm, gap=1mm interval, number of slices= 30.A total of 186 phases and 1800 images were acquired, and acquisition time was about 1 minute 45 seconds. When the image appeared after scanning, the bolus injection of Gd-DTPA (gadopentetate dimeglumine and Magnevist; Bayer Healthcare, Berlin, Germany) was performed with a high pressure injector through the cubital vein, with an injection rate of 5ml/s and a dosage of 0.2ml/kg. Subsequently, the same rate was used in the injection of saline. After the DSC imaging, the patients underwent the axial, coronal and sagittal T1-wighted enhancing scan.3. Data processingIntravoxel incoherent motion (IVIM) is comprised of two parts:the true diffusion of water molecules and the pseudo-diffusion and fraction of microcirculation perfusion. IVIM-DWI can characterize the pure diffusion coefficient of the model and perfusion coefficients (Fp) without exogenous contrast agent. Biological tissue in vivo, diffusion-weighted imaging is the measurement of the incoherent movements in voxels, which contains water molecules diffusion and microcirculation perfusion. A range of b values (from low to high values) was applied in fitting the bi-exponential model, using the equation as follows:Si/Sio= (1-Fp).exp (-b×ADCd)+Fp.exp (-b×ADCp) (1),With which it can calculate the coefficient of diffusion and perfusion of the tissue fraction. At low b values, microcirculation perfusion in the ADC value accounts for a larger proportion. As the b values rise, microcirculation perfusion in the ADC value percentage decreases and mainly reflects the real water diffusion within the orga-nization.Diffusion-sequence data were sent to the Philips station using Philips post-processing workstation (Extended workspace Philips Healthcare). A software Pride DWI Tool (version 1.5, Philips Healthcare) was used in data processing. The dispersion data was fitted with a mono-exponential model and bi-e exponential IVIM model for postprocessing. When b> 200 s/mm2, the signal attenuation of D*is negligible, the equation (1) can be simplified as following one:Sb/S0=exp(-b*D)(2)The high b values were put into the equation (2) to obtain the true diffusion coefficient D by linear regression equation. F value and D* values were calculated by nonlinear least squares fitting algorithm data. PRIDE fitting software can calculate R2 by their own data. If the goodness of fitting of data in this group wasregarded as greater than 0.8, then the data can be included in the study.ImageJ software was used to automatically generate the D map, f map and D* map. As the D map showed the highest spatial resolution, we chose the D map to draw a region of interest (ROI). Region of interest was regard as in the enhancing part on T1-weighted image as close to the cancer center as possible. Care was taken to avoid cystic, necrosis and hemorrhage.According to the equation of mono-exponential model (2), two b values of 0 and the 800s/mm2 would generate the ADC map and ADC values were measured at the same level of the lesion.2.2.2 Analysis of DSC-MRI dataAfter the injection of gadolinium contrast agent, the quick repeat measurements can reflect the short T2 weighted signal integrity or T2* loss through a simple linear relationship between reaction of contrast agent concentration in each pixel. Different from ion-enhanced CT, DSC-MRI is not direct, but it can cause the magnetic susceptibility changes. And the changes of proton relaxation rates can be detected as the MR signal changes.The relationship of T2* MRI signal and organization features of tracer concentrations was transferred into the signal intense of a single voxel-time data to trace tissue concentration-time data:Ci(t)=-k.log (S(t)/S(t0))/TE (3)Ci (t) is the concentration of contrast agent in time t organization, s (t) is the organization signal strength at a point in time after intravenous bolus contrast, s (t0) is a baseline signal before the contrast agent injection, while TE echo time, and k is a constant of proportionality that depends on the contrast agent, type of organization, field strength and pulse sequences and other factors.The images would be transmitted to the Philips magnetic resonance perfusion Imaging workstation (Extended workspace Philips Healthcare), CBV map,CBF map and MTT map were generated by neuron perfusion processing.Compared to T1 enhancement image, a radiologist chose the highest CBV color on the color map, and ROI should avoid necrotic or cystic, vascular anomalies. While in the healthy brain the contralateral parenchyma was measured by the same size of ROI, recording rCBV=TBV/CBV values.Two primary radiologists, one of whom was required to repeat measurements, performed all above measurement.4. Statistical AnalysisOne-way ANOVA was used to compare the difference of age between the two groups, and Pearson x2 test was applied to compare the patient gender. All the measurement data were recorded as mean±standard deviation. SPSS20.0 software (Chicago, IL, USA) was used for data analysis, and p<0.05 indicated statistical significance.Medcalc softwarewas used for calculation of ICC (intraclass correlation coefficients) of the three sets of data. ADC value, D value, f value and D* value derived from IVIM were comparedwith one-way ANOVA among the three groups of tumor (high grade astrocytoma, meningioma and lymphoma). Kruskal-Wallis H test was applied to compare the rCBV values among the three groups of tumor (high grade astrocytoma, meningioma and cerebral lymphoma) Pearson or Spearmen correlation analysis were applied between D value and ADC value, IVIM perfusion parameters f and DSC parameters rCBV.[Results]In 65 cases,3 cases of patients were excluded without pathological confirmation. Among the remaining 62 patients,2 patients were diagnosed with recurrent tumors,6 patients were low-grade astrocytoma (pathology DNT), and 3 cases of patients were cavernous hemangioma. The remaining 51 patients, according to the following criteria, included:(a) the patients who were not treated but wasconfirmed by pathology HGAs (n= 20), meningioma (n= 20) or PCNSL (n= 11)7 cases were single and 4 cases were multiple foci among the primary cerebral lymphoma.19 foci were found in the analysis. The locations of the foci were included as follows:7 foci were located in bilateral basal ganglia area,2 were in the frontal lobes,5 were in the temporal lobes,3 were in the parietal lobe and 2were in the corpus callosum.11 cases were near the middle line of cerebrum.15 foci showed iso-intensity in T1 weighted image, the others were hypointensity. The intensity of foci in T2 weighted image was complicomplicated. Two foci were hyperintensity,11 were mild hyperintensity,2 were iso-intensity and 3 were hoperintensity.16 foci showed peritumoral edema including finger-like edema (12/16) and ring-like edema (4/16).15 foci were homogeneous on MRI scan while 2 cases were seen bleeding and the other 2 showed cystic degeneration. The diameters of tumor lesions were measured on T1WI enhancing sequence. The average diameter of maximum foci of the primary cerebral lymphoma was (23±4.21) mm.7 cases were homogeneous enhanced 4 cases were accompanied by spot-like enhancement,2 cases were ring-like enhancement and 2 were heterogeneous enhancement.8 cases showed irregular margins (hand clasping syndrome:1 case; a sharp corner features:2 cases; umbilicus: 5).18 cases were single and 2 cases were multiple foci among the high grade astrocytoma.23 foci were found in the analysis.5 cases were across the lobes. The locations of the maximum foci were included as follows:1 case was located in basal ganglia area,7 were in the frontal lobes,3 were in the temporal lobes,2 were in the parietal lobes.1 was in the insular lobe and lwas in the lateral ventricle. The high grade astrocytoma showed hypointensity in T1 weighted image and hyperintensity in T2 weighted image and T2-FLAIR image.12 cases showed peritumoral edema including finger-like edema (10 cases) and ring-like edema (2 cases). The diameters of tumor lesions were measured on T1WI enhancing sequence. The average diameter of maximum foci of the high grade astrocytoma was (31±5.81) mm. The 20 cases were heterogeneous enhanced:12 cases were garland-like enhancement,2 were ring-like enhancement,3 were heterogeneous enhancement and 2 were enhanced as mass.19 cases were single and 1 case was multiple foci among the meningioma. The locations of the maximum foci were included as follows:3 foci were located in occipital lobe,4 were in the frontal lobes,3 were in the temporal lobes,4 were in the parietal lobe.3 were in the cerebral falx,3 were in the sphenoid bone. The meningioma showed iso-/hypointensity in T1 weighted image. The intensity of foci in T2 weighted image was complicomplicated.5 foci were hyperintensity,3 were mild hyperintensity,6 were iso-intensity and 6 were hoperintensity.The diameters of tumor lesions were measured on T1WI enhancing sequence. The average diameter of maximum foci of the meningioma was (35±2.11) mm.18 cases were homogeneous enhanced and 2 were heterogeneous enhancement with cystic degeneration.Two radiologists measured the three sets of data among three groups. The ADC value, D value, f values and rCBV value showed better consistency between the groups, with an ICC of ADC value=0.882(95% CI:0.792～0.933), D value=0.861 (95%CI:0.667～0.893), f value=0.862(95%CI:0.791～0.963) and rCBV value= 0.892(95%CI:0.857～0.946). D* demonstrated a relatively good consistency, with an ICC=0.657 (95%CI:0.498～0.848)Cerebral lymphoma:ADC value (1.16±0.02)×10-3mm2/s, D value:(0.76±0.02) ×10-3mm2/s, f value (7.13±0.54)% and the D* value (87.80±20.35)×10-3mm2/sHigh grade astrocytoma:ADC value(1.72±0.56)×10-3mm2/s, D value (1.38± 0.06)×10-3mm2/s, f value (8.75±2.88)% and the D* value (90.83±11.83)× 10-3mm2/s.ADC value of meningioma (1.23±0.04)×10-3mm2/s, D value (0.91±0.35)× 10"3mm2/s, f value (14.63±1.21)% and the D* value (121.51 13.60)×10-3mm2/s.The ADC value, D value and f value and D* value derived from cerebral lymphoma were significantly lower than those in high grade astrocytoma and meningioma. The ADC value, D value, f values and D* values of meningioma were higher than the other two types of tumors. The ADC value, d value, f value showed significant differences among the three groups of tumors. D* value did not showed significant difference (Table 3-3).The rCBV valueof cerebral lymphoma:1.18±0.08, the rCBV value of high grade astrocytoma:5.59±0.59, the rCBV value of meningioma:4.56±0.29. Perfusion of cerebral lymphoma was the lowest, as compared with high grade astrocytoma and meningioma and there was significant difference between cerebral lymphoma and high grade astrocytoma (p=0.00).I VIM perfusion parameters of cerebral lymphoma and astrocytoma demonstrated a positive correlation with DSC-MRI perfusion parameters[Conclusion]1. D value and ADC value derived from IVIM-MRI are significantly different among the primary cerebral lymphoma, meningioma and high grade astrocytoma, respectively. The ADC value is the lowest in primary cerebral lymphoma while it is highest in the meningioma. There is no significant difference in D-value or ADC value between primary cerebral lymphoma and the high grade astrocytoma.2. The f value derived from I VIM is different. And the parameters of primary cerebral lymphoma are lower than those of the high grade astrocytoma. No differences are found between the meningioma and primary cerebral lymphoma or meningioma and HGG3. The rCBV values derived from DSC perfusion are different. And the parameters of primary cerebral lymphoma are lower than those of the high grade astrocytoma and meningioma. No differences are found between the meningioma and HGG.4. There is a positive correlation between the f value derived from IVIM and rCBV value derived from the DSC-MRI perfusion imaging in primary cerebral lymphom and high grade astrocytoma.