Simulation of acquisition shifts in T2 weighted fluid-attenuated inversion recovery magnetic resonance images to stress test artificial intelligence segmentation networks.

Purpose: To provide a simulation framework for routine neuroimaging test data, which allows for "stress testing" of deep segmentation networks against acquisition shifts that commonly occur in clinical practice for T2 weighted (T2w) fluid-attenuated inversion recovery magnetic resonance imaging protocols. Approach: The approach simulates "acquisition shift derivatives" of MR images based on MR signal equations. Experiments comprise the validation of the simulated images by real MR scans and example stress tests on state-of-the-art multiple sclerosis lesion segmentation networks to explore a generic model function to describe the F1 score in dependence of the contrast-affecting sequence parameters echo time (TE) and inversion time (TI). Results: The differences between real and simulated images range up to 19% in gray and white matter for extreme parameter settings. For the segmentation networks under test, the F1 score dependency on TE and TI can be well described by quadratic model functions (R2>0.9). The coefficients of the model functions indicate that changes of TE have more influence on the model performance than TI. Conclusions: We show that these deviations are in the range of values as may be caused by erroneous or individual differences in relaxation times as described by literature. The coefficients of the F1 model function allow for a quantitative comparison of the influences of TE and TI. Limitations arise mainly from tissues with a low baseline signal (like cerebrospinal fluid) and when the protocol contains contrast-affecting measures that cannot be modeled due to missing information in the DICOM header.

Vessel size imaging (VSI) by robust magnetic resonance (MR) relaxometry: MR-VSI of solid tumors in correlation with immunohistology and intravital microscopy.

The aim of this study was to evaluate a robust magnetic resonance (MR) vessel size imaging (VSI) method for the noninvasive assessment of mean vessel size in solid tumors in a clinical dose range of ultrasmall superparamagnetic particles of iron oxide (USPIO). Therefore, USPIO-enhanced MR-VSI was performed on DU-4475, MDA-MB-435, and EOMA tumor-bearing mice xenografts with known differences in angiogenic activity and vessel morphology. MR results were compared to vessel sizes determined by immunohistochemistry (anti-CD31) and by intravital microscopy (IVM). MR-VSI revealed significantly different mean vessel sizes between the xenograft models at both USPIO doses (DU-4475: 20.6 ± 4.9 µm; MDA-MB-435: 37.4 ± 8.8 µm; and EOMA: 60.3 ±9.6 µm at 80 µmol/kg; p < .05). Immunohistochemistry revealed lower values for all tumor entities, whereas the size distribution was in line with MR-measurements. IVM corroborated the MR results for DU-4475 and MDA-MB435, but showed similar vessel sizes for MDA-MB-435 and EOMA. Our MR-VSI method allowed a noninvasive estimation of the mean vessel size in mice xenograft solid tumors with variable vascularity using a clinically relevant USPIO dose range.

Dynamic and simultaneous MR measurement of R1 and R2* changes during respiratory challenges for the assessment of blood and tissue oxygenation.

This work presents a novel method for the rapid and simultaneous measurement of R1 and R2* relaxation rates. It is based on a dynamic short repetition time steady-state spoiled multigradient-echo sequence and baseline R1 and B1 measurements. The accuracy of the approach was evaluated in simulations and a phantom experiment. The sensitivity and specificity of the method were demonstrated in one volunteer and in four patients with intracranial tumors during carbogen inhalation. We utilized (ΔR2*, ΔR1) scatter plots to analyze the multiparametric response amplitude of each voxel within an area of interest. In normal tissue R2* decreased and R1 increased moderately in response to the elevated blood and tissue oxygenation. A strong negative ΔR2* and ΔR1 response was observed in veins and some tumor areas. Moderate positive ΔR2* and ΔR1 response amplitudes were found in fluid-rich tissue as in cerebrospinal fluid, peritumoral edema, and necrotic areas. The multiparametric approach was shown to increase the specificity and sensitivity of oxygen-enhanced MRI compared to measuring ΔR2* or ΔR1 alone. It is thus expected to provide an optimal tool for the identification of tissue areas with low oxygenation, e.g., in tumors with compromised oxygen supply.

Current limitations of molecular magnetic resonance imaging for tumors as evaluated with high-relaxivity CD105-specific iron oxide nanoparticles.

OBJECTIVE: Tumor imaging via molecular magnetic resonance imaging (MRI) that uses specific superparamagnetic iron oxide particles (SPIOs) has been addressed in the literature several times in the last 20 years. To our knowledge, none of the reported approaches is currently used for routine clinical diagnostic evaluation, nor are any in clinical development. This raises questions as to whether SPIO-enhanced molecular MRI is sensitive and specific enough for use in clinical practice. The aim of our preclinical study was to investigate the minimum requirements for obtaining sensitive molecular MRI for use in tumor evaluations under optimal conditions. The well-vascularized F9 teratocarcinoma tumor model, which exhibits high levels of the highly accessible target CD105 (endoglin), was used to compare the accumulation and visualization of target-specific SPIOs by MRI. MATERIAL AND METHODS: Superparamagnetic iron oxide particles were optimized in the following ways: (a) proton relaxivity was increased for higher imaging sensitivity, (b) a coating material was used for optimal loading density of the αCD105 antibody, and (c) binding activity to the target CD105 was increased. Binding activity and specificity were confirmed in vitro using enzyme-linked immunosorbent assay and in vivo using pharmacokinetic and biodistribution studies of 11 F9 teratoma-bearing mice together with micro-autoradiography. CD105 target expression was determined using immunohistochemistry and quantitative enzyme-linked immunosorbent assay. The transverse relaxation rate R2* was quantified by 3.0-T MRI in the tumors, kidneys, and muscles before and up to 60 minutes after injection in 11 mice. The use of [Fe]-labeled SPIOs for all in vivo experiments allowed for the direct correlation of the imaging results with SPIO accumulation. RESULTS: High-relaxivity αCD105-polyacrylic acid-SPIOs (r2 up to 440 L mmol Fe s) with strong binding activity accumulated specifically in tumors (1.4% injected dose/g) and kidneys (4.1% injected dose/g) in a manner dependent on the target concentration. The accumulation occurred within the first 3 minutes after injection. Visualization of specific SPIOs was accomplished with MRI. In contrast to the successful use of MRI in all examined kidneys (mean ± SEM ΔR2*, 61 ± 11 s), only 6 of 11 tumors (mean ± SEM ΔR2*, 15 ± 7 s) showed a clear signal when compared with the control even though optimal conditions were used. CONCLUSION: The accumulation of CD105-specific SPIOs in F9 mouse teratomas was robust. However, visualization of the specifically accumulated SPIOs by MRI was not reliable because of its limited signal detection sensitivity. We postulate that it will be challenging to improve the imaging properties of targeted SPIOs further. Therefore, molecular MRI by targeted SPIOs is currently not suitable for clinical tumor imaging using routinely applicable sequences and field strength.

Free-breathing quantitative dynamic contrast-enhanced magnetic resonance imaging in a rat liver tumor model using dynamic radial T(1) mapping.

OBJECTIVES: : The high sensitivity to motion artifacts is a major limiting factor for applying the dynamic 3D T1-weighted gradient-echo (3D T1w GRE) technique for dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) experiments in small rodents. Dynamic quantification of the relaxation rate R1 (1/T1) presents an alternative approach to reduce these motion artifacts. In this work, an optimized 2D single-shot Look-Locker based T1 mapping technique, named GOLD, applying radial sampling in the golden-angle view order and contrast-enhancing k-space filter was evaluated for its use in free-breathing quantitative DCE-MRI of rat liver on a clinical 1.5 T MRI system. MATERIALS AND METHODS: : In vitro measurements and initial in vivo experiments in healthy rats were performed to evaluate the accuracy and resilience of the GOLD technique to motion artifacts. Unifocal hepatocellular carcinoma (HCC) was established in 20 male Buffalo rats. Twelve days after tumor cell implantation, animals were screened for intrahepatic tumor nodules by high-resolution T2-weighted MRI. Quantitative DCE-MRI experiments applying bolus injected gadopentetate dimeglumine were performed in 11 HCC-bearing rats using the GOLD technique. For comparison, a standard 3D T1w GRE sequence was applied in 6 additional rats. RESULTS: : Phantom experiments showed good agreement for T1 values measured by the GOLD method and an inversion recovery spectroscopy measurement. The in vivo experiments in healthy rats confirmed the robustness of the GOLD method in T1 value determination and its resilience to motion artifacts. Gadopentetate dimeglumine concentration (CGd) time curves determined from free-breathing GOLD-based DCE-MRI experiments of HCC-bearing rats allowed reliable and robust pharmacokinetic modeling (K, ve, lag time Td, and slow washout rate rwo) of tumor, liver, and spinal muscle. In comparison to the dynamic 3D T1w GRE, the GOLD method showed less variation and jitter in the CGd time curves and significantly increased accuracy (in terms of the goodness of fit) in the pharmacokinetic modeling. Significant differences were detected for K and ve with the 3D T1w GRE method apparently underestimating those parameters. CONCLUSIONS: : The GOLD technique allowed dynamic sampling of 2D axial T1 maps of the rat abdomen with 6-second temporal resolution enabling simultaneous and robust pharmacokinetic modeling of HCC, normal liver, and spinal muscle.

Monitoring of bevacizumab-induced antiangiogenic treatment effects by "steady state" ultrasmall superparamagnetic iron oxide particles magnetic resonance imaging using robust multiecho ΔR2* relaxometry.

RATIONALE AND OBJECTIVES: The purpose of this study was to evaluate whether "steady state" magnetic resonance imaging (MRI) using a robust multiecho ΔR2* MR relaxometry technique is suitable for the early assessment of a clinically approved antiangiogenic treatment regimen using bevacizumab (Avastin). METHODS: A673 rhabdomyosarcoma-bearing mice were treated with bevacizumab (n = 6) or saline as control, respectively (n = 6). MRI using a multigradient echo sequence was performed before and after 2 doses of 100 µg bevacizumab at baseline and day 7. Ultrasmall superparamagnetic iron oxide particles (SH U 555 C) induced changes of the transverse relaxation rate R2* (ΔR2*) were measured in regions of interest. From these results, the vascular volume fraction was estimated, providing a surrogate marker for the microvessel density (MVD). The actual MVD was determined by immunohistochemistry and correlated with the MRI results. RESULTS: Bevacizumab treatment resulted in a significant reduction of the ΔR2* values compared with the control group (bevacizumab: 10.47 ± 0.78 seconds(-1) vs. control: 17.91 ± 2.63 seconds(-1); P = 0.01), reflecting the significant decrease of the vascular volume fraction by 33% (bevacizumab: 2.21% ± 0.15% vs. control: 3.31% ± 0.22%; P = 0.001). Immunohistochemistry confirmed the MR results showing an approximately 25% reduction of the MVD after treatment (bevacizumab: 7.11 ± 0.3 vs. control: 9.45 ± 0.38; P = 0.001). CONCLUSION: Multiecho ΔR2* MR relaxometry allows an early and quantitative assessment of tumor vascularization changes in response to an antiangiogenic treatment with a clinically approved vascular endothelial growth factor inhibitor. With the availability of long circulating ultrasmall superparamagnetic iron oxide particles s for clinical use, this imaging technique could be instantly translated to antiangiogenic treatment monitoring in clinical studies.

Accelerated T2 mapping for characterization of prostate cancer.

Prostate T(2) mapping was performed in 34 consecutive patients using an accelerated multiecho spin-echo sequence with 4-fold k-space undersampling leading to a net acceleration factor of 3.3 on a 3T scanner. The mean T(2) values from the accelerated and conventional, unaccelerated sequences demonstrated a very high correlation (r = 0.99). Different prostate segments demonstrated similarly good interscan reproducibility (p = not significant) with slightly larger difference at base: 2.0% ± 1.6% for left base and 2.1% ± 1.1% for right base. In patients with subsequent targeted biopsy, T(2) values of histologically proven malignant tumor areas were significantly lower than the suspicious looking but nonmalignant lesions (p < 0.05) and normal areas (p < 0.001): 100 ± 10 ms for malignant tumors, 114 ± 23 ms for suspicious lesions and 149 ± 32 ms for normal tissues. The proposed method can provide an effective approach for accelerated T(2) quantification for prostate patients.

Concurrent MR blood volume and vessel size estimation in tumors by robust and simultaneous ΔR2 and ΔR2* quantification.

This work presents a novel method for concurrent estimation of the fractional blood volume and the mean vessel size of tumors based on a multi-gradient-echo-multi-spin-echo sequence and the injection of a super-paramagnetic blood-pool agent. The approach further comprises a post-processing technique for simultaneous estimation of changes in the transverse relaxation rates R(2) and R(2)*, which is robust against global B(0) and B(1) field inhomogeneities and slice imperfections. The accuracy of the simultaneous ΔR(2) and ΔR(2)* quantification approach is evaluated in a phantom. The simultaneous blood volume and vessel size estimates, obtained with MR, compare well to the immunohistological findings in a preclinical experiment (HT1080 cells, implanted in nude mice). Clinical translation is achieved in a patient with a pleomorphic sarcoma in the left pubic bone. The latter demonstrates the robustness of the technique against changes in the contrast agent concentration in blood during washout.

Changes in the MR relaxation rate R(2)* induced by respiratory challenges at 3.0 T: a comparison of two quantification methods.

The consistent determination of changes in the transverse relaxation rate R(2)* (ΔR(2)*) is essential for the mapping of the effect of hyperoxic and hypercapnic respiratory challenges, which enables the noninvasive assessment of blood oxygenation changes and vasoreactivity by MRI. The purpose of this study was to compare the performance of two different methods of ΔR(2)* quantification from dynamic multigradient-echo data: (A) subtraction of R(2)* values calculated from monoexponential decay functions; and (B) computation of ΔR(2)* echo-wise from signal intensity ratios. A group of healthy volunteers (n = 12) was investigated at 3.0 T, and the brain tissue response to carbogen and CO(2)-air inhalation was registered using a dynamic multigradient-echo sequence with high temporal and spatial resolution. Results of the ΔR(2)* quantification obtained by the two methods were compared with respect to the quality of the voxel-wise ΔR(2)* response, the number of responding voxels and the behaviour of the 'global' response of all voxels with significant R(2)* changes. For the two ΔR(2)* quantification methods, we found no differences in the temporal variation of the voxel-wise ΔR(2)* responses or in the detection sensitivity. The maximum change in the 'global' response was slightly smaller when ΔR(2)* was derived from signal intensity ratios. In conclusion, this first methodological comparison shows that both ΔR(2)* quantifications, from monoexponential approximation as well as from signal intensity ratios, are applicable for the monitoring of R(2)* changes during respiratory challenges.

Quantification of the magnetic resonance signal response to dynamic (C)O(2)-enhanced imaging in the brain at 3 T: R*(2) BOLD vs. balanced SSFP.

PURPOSE: To compare two magnetic resonance (MR) contrast mechanisms, R*(2) BOLD and balanced SSFP, for the dynamic monitoring of the cerebral response to (C)O(2) respiratory challenges. MATERIALS AND METHODS: Carbogen and CO(2)-enriched air were delivered to 9 healthy volunteers and 1 glioblastoma patient. The cerebral response was recorded by two-dimensional (2D) dynamic multi-gradient-echo and passband-balanced steady-state free precession (bSSFP) sequences, and local changes of R*(2) and signal intensity were investigated. Detection sensitivity was analyzed by statistical tests. An exponential signal model was fitted to the global response function delivered by each sequence, enabling quantitative comparison of the amplitude and temporal behavior. RESULTS: The bSSFP signal changes during carbogen and CO(2)/air inhalation were lower compared with R*(2) BOLD (ca. 5% as opposed to 8-13%). The blood-oxygen-level-dependent (BOLD) response amplitude enabled differentiation between carbogen and CO(2)/air by a factor of 1.4-1.6, in contrast to bSSFP, where differentiation was not possible. Furthermore, motion robustness and detection sensitivity were higher for R*(2) BOLD. CONCLUSION: Both contrast mechanisms are well suited to dynamic (C)O(2)-enhanced MR imaging, although the R*(2) BOLD mechanism was demonstrated to be superior in several respects for the chosen application. This study suggests that the R*(2) BOLD and bSSFP-response characteristics are related to different physiologic mechanisms.

Intracranial tumor response to respiratory challenges at 3.0 T: impact of different methods to quantify changes in the MR relaxation rate R2*.

PURPOSE: To compare two DeltaR2* quantification methods for analyzing the response of intracranial tumors to different breathing gases. The determination of changes in the magnetic resonance imaging (MRI) relaxation rate R2* (DeltaR2*), induced by hyperoxic and hypercapnic respiratory challenges, enables the noninvasive assessment of blood oxygenation changes and vasoreactivity. MATERIALS AND METHODS: Sixteen patients with various intracranial tumors were examined at 3.0 T. The response to respiratory challenges was registered using a dynamic multigradient-echo sequence with high temporal and spatial resolution. At each dynamic step, DeltaR2* was derived in two different ways: 1) by subtraction of R2* values obtained from monoexponential decay functions, 2) by computing DeltaR2* echo-wise from signal intensity ratios. The sensitivity for detection of responding voxels and the behavior of the "global" response were investigated. RESULTS: Significantly more responding voxels (about 4%) were found for method (1). The "global" response was independent from the chosen quantification method but showed slightly larger changes (about 6%) when DeltaR2* was derived from method (1). CONCLUSION: Similar results were observed for the two methods, with a slightly higher detection sensitivity of responding voxels when DeltaR2* was obtained from monoexponential approximation.

Tumor blood volume determination by using susceptibility-corrected DeltaR2* multiecho MR.

PURPOSE: To evaluate a susceptibility-corrected multiecho magnetic resonance (MR) relaxometry technique for an accurate and robust determination of DeltaR2* as a noninvasive surrogate parameter of the perfused tumor blood volume. MATERIALS AND METHODS: All experiments were approved by the institutional animal care committee. In a glass tube phantom with different superparamagnetic iron oxide (SPIO) particle concentrations and at tumor mice xenografts with DU-4475, HT-1080, and MDA-MB-435 tumors (n = 15 total, n = 5 per model) with different degrees of neovascularization after injection of different ultrasmall SPIO (USPIO) doses changes of the transverse relaxation rate (DeltaR2*) were determined by using a fixed echo time (TE) of 22 msec and a susceptibility-corrected multigradient-echo technique. The mean DeltaR2* value and the vascular volume fraction (VVF) of each tumor was determined and compared with independent in vivo fluorescent tumor perfusion measurements and histologic analysis helped determine microvessel density (MVD). Statistical differences were tested by using analysis of variance and linear correlations. RESULTS: For the phantom study, DeltaR2* maps calculated with a fixed TE of 22 msec showed a higher standard deviation of the noise index compared with the susceptibility-corrected multiecho technique. For the xenograft model, mean tumor DeltaR2* values (+/- standard error of the mean) showed significant differences between the various tumors (eg, DU-4475: 12.3 sec(-1) +/- 2.67, HT-1080: 36.47 sec(-1) +/- 5.84, and MDA-MB-435: 64.01 sec(-1) +/- 8.87 at 80 mumol of iron per kilogram; P < .05). DeltaR2* values increased dose dependently and in a linear fashion, resulting in reproducibly stable VVF measurements. Fluorescent tumor perfusion measurements and MVD counts corroborated the MR results. CONCLUSION: Susceptibility-corrected multiecho MR relaxometry allows a highly accurate and robust determination of DeltaR2* and VVF with an excellent dynamic range for tumor characterization at clinically relevant doses of USPIO.