Frederik B. Laun

Differentiation of Pancreas Carcinoma From Healthy Pancreatic Tissue Using Multiple b-Values: Comparison of Apparent Diffusion Coefficient and Intravoxel Incoherent Motion Derived Parameters

Objectives:To evaluate in detail the diagnostic performance of diffusion-weighted imaging (DWI) to differentiate pancreas carcinoma from healthy pancreas using the apparent diffusion coefficient (ADC) and parameters derived from the intravoxel incoherent motion (IVIM) theory. Materials and Methods:Twenty-three patients with pancreas carcinoma and 14 volunteers with healthy pancreas were examined at 1.5 Tesla using a single-shot echo-planar imaging DWI pulse sequence. Eleven b-values ranging from 0 to 800 s/mm2 were used. The acquisition was separated into blocks (b0, b25), (b0, b50),...(b0, b800) and each block was acquired in a single expirational breath-hold (TA = 26 seconds) to avoid motion artifacts. The ADC was calculated for all b-values using linear regression yielding ADCtot. By applying the IVIM model, which allows for the estimation of perfusion effects in DWI, the perfusion fraction f and the perfusion free diffusion parameter D were calculated. The diagnostic performance of ADC, f and D as a measure for the differentiation between healthy pancreas and pancreatic carcinoma was evaluated with receiver operating characteristics analysis. Results:In the healthy control group, the ADCtot ranged from 1.53 to 2.01 &mgr;m2/ms with a mean value of 1.71 ± 0.19 &mgr;m2/ms, the perfusion fraction f ranged from 18.5% to 40.4% with a mean value of 25.0 ± 6.2%, and the diffusion coefficient D from 0.94 to 1.28 &mgr;m2/ms with a mean value of 1.13 ± 0.15 &mgr;m2/ms. In patients with pancreas carcinoma, the ADCtot ranged from 0.98 to 1.81 &mgr;m2/ms with a mean value of 1.31 ± 0.24 &mgr;m2/ms, the perfusion fraction f ranged from 0% to 20.4% with a mean value of 8.59 ± 4.6% and the diffusion coefficient D from 0.74 to 1.60 &mgr;m2/ms with a mean value of 1.15 ± 0.22 &mgr;m2/ms. In comparison to healthy pancreatic tissue, a significant reduction of the perfusion fraction f and of ADCtot was found in pancreatic carcinoma (P < 0.00001, 0.0002, respectively). The f value showed more than a 10-fold higher significance level in distinguishing cancerous from normal tissue when compared with the ADCtot value. No significant difference in the diffusion coefficient D was observed between the 2 groups (P > 0.5). In the receiver operating characteristic-analyses, the area under curve for f was 0.991 and significantly larger than ADCtot (P < 0.05). f had the highest sensitivity, specificity, negative predictive value, and positive predictive value with 95.7%, 100%, 93.3%, and 100%, respectively. Conclusions:Using the IVIM-approach, the f value proved to be the best parameter for the differentiation between healthy pancreas and pancreatic cancer. The acquisition of several b-values strongly improved the stability of the parameter estimation thus increasing the sensitivity and specificity to 95.7% and 100% respectively. The proposed method may hold great promise for the non invasive, noncontrast-enhanced imaging of pancreas lesions and may eventually become a screening tool for pancreatic cancer.

Toward an optimal distribution of b values for intravoxel incoherent motion imaging

The intravoxel incoherent motion (IVIM) theory provides a framework for the separation of perfusion and diffusion effects in diffusion-weighted imaging (DWI). To measure the three free IVIM parameters, DWIs with several diffusion weightings b must be acquired. To date, the used b value distributions are chosen heuristically and vary greatly among researchers. In this work, optimal b value distributions for the three parameter fit are determined using Monte-Carlo simulations for the measurement of a low, medium and high IVIM perfusion regime. The first 16 b values of a b value distribution, which was optimized to be appropriate for all three regimes, are {0, 40, 1000, 240, 10, 750, 90, 390, 170, 10, 620, 210, 100, 0, 530 and 970} in units of seconds per square meter. This distribution performed well for all organs and outperformed a distribution frequently used in the literature. In case of limited acquisition time, the b values should be chosen in the given order, but at least 10 b values should be used for current clinical settings. The overall parameter estimation quality depends strongly and nonlinearly on the signal-to-noise ratio (SNR): it is essential that the SNR is considerably higher than a critical SNR. This critical SNR is about 8 for medium and high IVIM perfusion and 50 for the low IVIM perfusion regime. Initial in vivo IVIM measurements were performed in the abdomen and were in keeping with the numerically simulated results.

An in vivo verification of the intravoxel incoherent motion effect in diffusion‐weighted imaging of the abdomen

To investigate the vascular contribution to the measured apparent diffusion coefficient and to validate the Intra Voxel Incoherent Motion theory, the signal as a function of the b‐value was measured in the healthy pancreas with and without suppression of the vascular component and under varying echo times (TE = 50, 70, and 100 msec). The perfusion fraction f and the diffusion coefficient D were extracted from the measured DW‐data using the original Intra Voxel Incoherent Motion‐equation and a modified version of this equation incorporating relaxation effects. First, the perfusion fraction f in the blood suppressed pancreatic tissue decreased significantly (P = 0.03), whereas the diffusion coefficient D did not change with suppression (P = 0.43). Second, the perfusion fraction f increased significantly with increasing echo time (P = 0.0025), whereas the relaxation time compensated perfusion fraction f′ showed no significant dependence on TE (P = 0.31). These results verify a vascular contribution to the diffusion weighted imaging measurement at low b values and support the Intra Voxel Incoherent Motion‐theory. Magn Reson Med, 2010.

Methodological considerations on tract-based spatial statistics (TBSS)

Having gained a tremendous amount of popularity since its introduction in 2006, tract-based spatial statistics (TBSS) can now be considered as the standard approach for voxel-based analysis (VBA) of diffusion tensor imaging (DTI) data. Aiming to improve the sensitivity, objectivity, and interpretability of multi-subject DTI studies, TBSS includes a skeletonization step that alleviates residual image misalignment and obviates the need for data smoothing. Although TBSS represents an elegant and user-friendly framework that tackles numerous concerns existing in conventional VBA methods, it has limitations of its own, some of which have already been detailed in recent literature. In this work, we present general methodological considerations on TBSS and report on pitfalls that have not been described previously. In particular, we have identified specific assumptions of TBSS that may not be satisfied under typical conditions. Moreover, we demonstrate that the existence of such violations can severely affect the reliability of TBSS results. With TBSS being used increasingly, it is of paramount importance to acquaint TBSS users with these concerns, such that a well-informed decision can be made as to whether and how to pursue a TBSS analysis. Finally, in addition to raising awareness by providing our new insights, we provide constructive suggestions that could improve the validity and increase the impact of TBSS drastically.

Sodium MRI using a density‐adapted 3D radial acquisition technique

A density‐adapted three‐dimensional radial projection reconstruction pulse sequence is presented which provides a more efficient k‐space sampling than conventional three‐dimensional projection reconstruction sequences. The gradients of the density‐adapted three‐dimensional radial projection reconstruction pulse sequence are designed such that the averaged sampling density in each spherical shell of k‐space is constant. Due to hardware restrictions, an inner sphere of k‐space is sampled without density adaption. This approach benefits from both the straightforward handling of conventional three‐dimensional projection reconstruction sequence trajectories and an enhanced signal‐to‐noise ratio (SNR) efficiency akin to the commonly used three‐dimensional twisted projection imaging trajectories. Benefits for low SNR applications, when compared to conventional three‐dimensional projection reconstruction sequences, are demonstrated with the example of sodium imaging. In simulations of the point‐spread function, the SNR of small objects is increased by a factor 1.66 for the density‐adapted three‐dimensional radial projection reconstruction pulse sequence sequence. Using analytical and experimental phantoms, it is shown that the density‐adapted three‐dimensional radial projection reconstruction pulse sequence allows higher resolutions and is more robust in the presence of field inhomogeneities. High‐quality in vivo images of the healthy human leg muscle and the healthy human brain are acquired. For equivalent scan times, the SNR is up to a factor of 1.8 higher and anatomic details are better resolved using density‐adapted three‐dimensional radial projection reconstruction pulse sequence. Magn Reson Med, 2009.

Intravoxel incoherent motion MRI for the differentiation between mass forming chronic pancreatitis and pancreatic carcinoma.

Purpose:To determine which of the quantitative parameters obtained from intravoxel incoherent motion diffusion weighted imaging (DWI) is the most significant for the differentiation between pancreatic carcinoma and mass-forming chronic pancreatitis. Materials and Methods:Twenty-nine patients with pancreatic masses were included, 9 proved to have a mass-forming pancreatitis and 20 had a pancreatic carcinoma. The patients were studied using intravoxel incoherent motion DWI with 11 b-values and the apparent diffusion coefficient (ADC), the true diffusion constant (D) and the perfusion fraction (f) were calculated. The diagnostic strength of the parameters was evaluated using receiver operating characteristic analysis. Results:The ADC in chronic pancreatitis was higher than in pancreatic carcinoma with significant differences at b = 50, 75, 100, 150, 200, 300 s/mm2 (ADC50 = 3.17 ± 0.67 vs. 2.55 ± 1.09, ADC75 = 2.46 ± 0.4 vs. 1.93 ± 0.52, ADC100 = 2.28 ± 0.48 vs. 1.73 ± 0.45, ADC150 = 1.97 ± 0.26 vs. 1.63 ± 0.40, ADC200 = 1.98 ± 0.24 vs. 1.53 ± 0.28, and ADC300 = 1.76 ± 0.19 vs. 1.46 ± 0.31 × 10−3 mm2/s). No significant differences were found at b = 25, 400, 600, and 800 s/mm2 (ADC25 = 4.69 ± 0.65 vs. 4.04 ± 1.35, ADC400 = 1.57 ± 0.21 vs. 1.37 ± 0.30, ADC600 = 1.38 ± 0.18 vs. 1.24 ± 0.25, and ADC800 = 1.27 ± 0.10 vs. 1.18 ± 0.19 × 10−3 mm2/s) nor using ADCtot (1.42 ± 0.23 vs. 1.28 ± 0.12 × 10−3 mm2/s). The perfusion fraction f was significantly higher in pancreatitis compared with pancreatic carcinoma (16.3% ± 5.30% vs. 8.2% ± 4.00%, P = 0.0001). There was no significant difference between groups for D (1.07 ± 0.224 × 10−3 mm2/s for chronic pancreatitis and 1.09 ± 0.3 × 10−3 mm2/s for pancreatic carcinoma, P = 0.66). For f, the highest area under the curve (0.894) and combined sensitivity (80%) and specificity (89.9%) were found. Conclusions:There were significant differences in ADC50–300 between chronic pancreatitis and pancreatic carcinoma. Because D is not significantly different between groups, differences in ADC can be attributed mainly to differences in perfusion. The perfusion fraction f proved to be the superior DWI-derived parameter for differentiation of mass-forming pancreatitis and pancreatic carcinoma.

Diffusion-weighted imaging for non-invasive and quantitative monitoring of bone marrow infiltration in patients with monoclonal plasma cell disease: A comparative study with histology

Bone marrow plasma cell infiltration is a crucial parameter of disease activity in monoclonal plasma cell disorders. Until now, the only way to quantify such infiltration was bone marrow biopsy or aspiration. Diffusion‐weighted imaging (DWI) is a magnetic resonance imaging‐technique that may mirror tissue cellularity by measuring random movements of water molecules. To investigate if DWI is capable of assessing bone marrow cellularity in monoclonal plasma cell disease, we investigated 56 patients with multiple myeloma or monoclonal gammopathy of undetermined significance, and 30 healthy controls using DWI of the pelvis and/or the lumbar spine. In 25 of 30 patients who underwent biopsy, bone marrow trephine and DWI could be compared. Of the patients with symptomatic disease 15 could be evaluated after systemic treatment. There was a positive correlation between the DWI‐parameter apparent diffusion coefficient (ADC) and bone marrow cellularity as well as micro‐vessel density (P < 0·001 respectively). ADC was significantly different between patients and controls (P < 0·01) and before and after systemic therapy (P < 0·001). In conclusion, DWI enabled bone marrow infiltration to be monitored in a non‐invasive, quantitative way, suggesting that after further investigations on larger patient groups this might become an useful tool in the clinical work‐up to assess tumour burden.

Opportunities and pitfalls in the quantification of fiber integrity: What can we gain from Q-ball imaging?

The quantification of fiber integrity is central to the clinical application of diffusion imaging. Compared to diffusion tensor imaging (DTI), Q-ball imaging (QBI) allows for the depiction of multiple fiber directions within a voxel. However, this advantage has not yet been shown to translate directly to superior quantification of fiber integrity. Furthermore, recent developments in QBI reconstruction with solid angle consideration have led to sharper and intrinsically normalized orientation distribution functions. The implications of this technique on quantification are also unknown. To investigate this, the generalized fractional anisotropy (GFA) from the original and the more recent QBI reconstruction scheme and the DTI derived fractional anisotropy (FA) were evaluated comparatively using Monte Carlo simulations and real MRI measurements of crossing fiber phantoms. Contrast-to-noise ratio, accuracy, independence of the acquisition setup and the relation of single fiber anisotropies to measured anisotropy in crossings were assessed. In homogeneous single-fiber regions at b-values around 1000 s/mm2, the FA performed best. While the original QBI reconstruction does not show a clear advantage even at higher b-values and in crossing regions, the new reconstruction scheme yields superior properties and is recommended for quantification at higher b-values and especially in regions of heterogeneous fiber configuration.

Evaluation of Diffusion Kurtosis Imaging Versus Standard Diffusion Imaging for Detection and Grading of Peripheral Zone Prostate Cancer.

ObjectivesThe purpose of the study was to evaluate and validate diffusion kurtosis imaging (DKI) for detection grading of peripheral zone prostate cancer (PCa) compared with standard diffusion-weighted imaging (DWI) in a cohort of patients with biopsy-proven PCa. Materials and MethodsIn this retrospective, single-institutional study, 55 patients (age, 67.5 ± 6.9 years; range, 52–84 years) who underwent multiparametric magnetic resonance imaging (MRI) before transperineal magnetic resonance/transrectal ultrasound-guided fusion biopsy were included. Suspicious lesions identified in multiparametric MRI underwent image-guided targeted biopsy procedure using a hybrid magnetic resonance/transrectal ultrasound-guided fusion biopsy system. Multiparametric MRI examinations were performed at 3.0 T using a 16-channel phased array coil. Diffusion kurtosis imaging has been acquired with 9 b values (0, 50, 250, 500, 750, 1000, 1250, 1500, and 2000 s/mm2). In patients with histologically proven PCa, a representative tumor region was determined as region of interest (ROI) on axial T2-weighted images in consensus by 2 board-certified radiologists. For quantitative evaluation, ROIs located in malignant and contralateral tumor-free regions were transferred to diffusion-weighted images. Diffusion kurtosis imaging parameters (Dapp and Kapp) and apparent diffusion coefficient (ADC) values of the ROIs in tumor and contralateral remote areas were calculated. Estimation of the kurtosis-derived parameters was performed using a voxel-by-voxel fit followed by an ROI-based averaging and a second fit to ROI-averaged signal values. A subgroup analysis was performed to determine the influence of aggressiveness of PCa using ADC, Dapp, and Kapp. The receiver operating characteristic (ROC) curves were calculated for DKI parameters and ADC values. ResultsIn the 55 patients, the average prostate-specific antigen level was 12.4 ± 12.6 ng/mL (range, 2.7–75.0 ng/mL), and the median Gleason score was 7 (range, 6–10). Dapp (units, 10−3 mm2/s) was significantly lower in tumor compared with control regions (1.48 ± 0.35 vs 2.00 ± 0.32, P < 0.05), and Kapp was significantly higher (1.01 ± 0.21 vs 0.76 ± 0.14, P < 0.05). Dapp was significantly higher than standard ADC (units, 10−3 mm2/s) both in tumor regions and in controls (1.48 ± 0.35 vs 1.10 ± 0.25 and 2.00 ± 0.32 vs 1.43 ± 0.25, P < 0.05). Neither the ROI-based calculation of the kurtosis parameters nor the application of the noise correction significantly changed the DKI parameter estimation. There was no significant difference for the applied fitting method for DKI-derived parameters considering the differentiation between tumor and control tissue. Subsequent ROC analyses did not reveal a significant difference between DKI and ADC for detection of PCa. Sensitivities derived by Youden J statistics cutoff values ranged from 69% to 91% for DKI parameters; specificities ranged from 71% to 89%. Subgroup analysis for DKI (Dapp, Kapp) and ADC for assessing aggressiveness of PCa found significant difference (P < 0.05) for discrimination between high- and low-grade findings. However, no significant difference could be obtained between standard DWI- and DKI-derived parameters. ConclusionsThe results of this study demonstrated no significant benefit of DKI for detection and grading of PCa as compared with standard ADC in the peripheral zone determined from b values of 0 and 800 s/mm2. For clinical routine application, ADC derived from monoexponential fitting of DWI data remains the standard for characterizing peripheral zone cancer of the prostate.

Prediction of treatment response in head and neck carcinomas using IVIM-DWI: Evaluation of lymph node metastasis.

PURPOSE To obtain diffusion and microperfusion measures in lymph node metastases of head and neck squamous cell carcinomas (HNSCC) using intravoxel incoherent motion (IVIM) imaging. The obtained IVIM parameters were used to characterize lymph nodes in the staging phase and longitudinal follow-up was performed to evaluate the potential predictive value of these parameters considering therapy response. METHODS Fifteen patients with lymph node metastases of histologically confirmed locally advanced HNSCC were examined using diffusion weighted imaging (DWI) before a nonsurgical organ preserving therapy. DWI imaging was performed at 3T using eight different b-values ranging from 0 to 800s/mm(2). Using the IVIM-approach, the perfusion fraction f and the diffusion coefficient D were extracted using a biexponential fit. A follow-up period of 13.5 months was available for all patients. One patient with a macroscopically necrotic lymph node was excluded from analyses. A region of interest (ROI)-analysis was performed in all patients. RESULTS Locoregional failure (LRF) was present in 3 of 15 patients within 13.5 months follow-up. The initial f-value was significantly higher (p=0.01) in patients with LRF (14.5±0.6% vs. 7.7±2.6%) compared to patients with locoregional control (LRC). The initial diffusion coefficient D did not differ significantly (p=0.30) between the two groups (0.97±0.15×10(-3)mm(2)/s vs. 0.88±0.13×10(-3)mm(2)/s). CONCLUSIONS Our results indicate that a high initial perfusion fraction f in lymph nodes may predict poor treatment response in patients with HNSCC due to locoregional failure.