This article reviews our exploration of structures and functions of the human visual cortex using high resolution (submillimeter) functional magnetic resonance imaging (fMRI). It discusses factors that restrict the spatial resolution of blood oxygenation-level dependent (BOLD) fMRI–the point-spread function of the BOLD signal, limited by both imaging techniques to be used and neurovascular units to be studied, and the signal-to-noise ratio. I offer personal thoughts regarding optimal solutions for dealing with these issues, summarize techniques we have developed over the years for using high resolution fMRI to visualize functional architectures and explore physiological properties in the primary visual cortex of humans, including choices of imaging hardware and pulse sequences, experimental procedures, and stimulation paradigms, and finally offer my personal opinions regarding the future of high resolution fMRI.
This article provides an overview of in vivo magnetic resonance (MR) imaging contrasts obtained for mammalian brain in relation to histological knowledge. Emphasis is paid to the (1) significance of high spatial resolution for the optimization of T1, T2, and magnetization transfer contrast, (2) use of exogenous extra- and intracellular contrast agents for validating endogenous contrast sources, and (3) histological structures and biochemical compounds underlying these contrasts and (4) their relevance to neuroradiology. Comparisons between MR imaging at subnanoliter resolution and histological data indicate that (a) myelin sheaths, (b) nerve cells, and (c) the neuropil are most responsible for observed MR imaging contrasts, while (a) diamagnetic macromolecules, (b) intracellular paramagnetic ions, and (c) extracellular free water, respectively, emerge as the dominant factors. Enhanced relaxation rates due to paramagnetic ions, such as iron and manganese, have been observed for oligodendrocytes, astrocytes, microglia, and blood cells in the brain as well as for nerve cells. Taken together, a plethora of observations suggests that the delineation of specific structures in high-resolution MR imaging of mammalian brain and the absence of corresponding contrasts in MR imaging of the human brain do not necessarily indicate differences between species but may be explained by partial volume effects. Second, paramagnetic ions are required in active cells in vivo which may reduce the magnetization transfer ratio in the brain through accelerated T1 recovery. Third, reductions of the magnetization transfer ratio may be more sensitive to a particular pathological condition, such as astrocytosis, microglial activation, inflammation, and demyelination, than changes in relaxation. This is because the simultaneous occurrence of increased paramagnetic ions (i.e., shorter relaxation times) and increased free water (i.e., longer relaxation times) may cancel T1 or T2 effects, whereas both processes reduce the magnetization transfer ratio.
Purpose: We examined safety issues related to the presence of various metallic dental materials in magnetic resonance (MR) imaging at 7 tesla. Methods: A 7T MR imaging scanner was used to examine 18 kinds of materials, including 8 metals used in dental restorations, 6 osseointegrated dental implants, 2 abutments for dental implants, and 2 magnetic attachment keepers. We assessed translational attraction forces between the static magnetic field and materials via deflection angles read on a tailor-made instrument and compared with those at 3T. Heating effects from radiofrequency during image acquisitions using 6 different sequences were examined by measuring associated temperature changes in agarose-gel phantoms with a fiber-optic thermometer. Results: Deflection angles of the metallic dental materials were significantly larger at 7T than 3T. Among full metal crowns (FMCs), deflection angles were 18.0° for cobalt-chromium (Co-Cr) alloys, 13.5° for nickel-chromium (Ni-Cr) alloys, and 0° for other materials. Deflection angles of the dental implants and abutments were minimal, ranging from 5.0 to 6.5°, whereas the magnetic attachment keepers were strongly attracted to the field, having deflection angles of 90° or more. Increases in temperature of the FMCs were significant but less than 1°C in every sequence. The dental implant of 50-mm length showed significant but mild temperature increases (up to 1.5°C) when compared with other dental implants and abutments, particularly on sequences with high specific absorption rate values. Conclusion: Although most metallic dental materials showed no apparent translational attraction or heating at 7T, substantial attraction forces on the magnetic attachment keepers suggested potential risks to patients and research participants undergoing MR imaging examinations.
Purpose: T1-Cube (GE HealthCare) is a relatively new 3-dimensional (3D) fast spin-echo (FSE)-based magnetic resonance (MR) imaging sequence that uses a variable flip angle to acquire gap-free volume scans. We compared the gadolinium enhancement characteristics of a heterogeneous population of brain tumors imaged by T1-Cube and then 3D fast spoiled gradient recall acquisition in steady state (3D FSPGR) 3-tesla MR imaging to identify the superior modality for specific diagnostic purposes. Methods: We examined 61 lesions from 32 patients using the 2 sequences after administration of gadopentetic acid (Gd-DTPA; 0.1 mmol/kg). Two neuroradiologists independently measured each lesion twice using a region-of-interest (ROI) method. We measured the contrast-to-noise ratio (CNR), the difference in signal intensity (SI) between the tumor and normal white matter relative to the standard deviation (SD) of the SI within the lesion, for both post-contrast 3D FSPGR and post-contrast T1-Cube images of the same tumor and compared modality-specific CNRs for all tumors and in subgroups defined by tumor size, enhancement ratio, and histopathology. Results: The mean CNR was significantly higher on T1-Cube images than 3D FSPGR images for the total tumor population (1.85 ± 0.97 versus 1.12 ± 1.05, P < 0.01) and the histologic types, i.e., metastasis (P < 0.01) and lymphoma (P < 0.05). The difference in CNR was even larger for smaller tumors in the metastatic group (4.95 to 23.5 mm2) (P < 0.01). In contrast, mean CNRs did not differ between modalities for high grade glioma and meningioma. Conclusions: Gadolinium enhancement of brain tumors was generally higher when imaged by T1-Cube than 3D FSPGR, and T1-Cube with Gd enhancement may be superior to 3D FSPGR for detecting smaller metastatic tumors.
Purpose: To shorten acquisition of diffusion kurtosis imaging (DKI) in 1.5-tesla magnetic resonance (MR) imaging, we investigated the effects of the number of b-values, diffusion direction, and number of signal averages (NSA) on the accuracy of DKI metrics. Methods: We obtained 2 image datasets with 30 gradient directions, 6 b-values up to 2500 s/mm2, and 2 signal averages from 5 healthy volunteers and generated DKI metrics, i.e., mean, axial, and radial kurtosis (MK, K∥, and K⊥) maps, from various combinations of the datasets. These maps were estimated by using the intraclass correlation coefficient (ICC) with those from the full datasets. Results: The MK and K⊥ maps generated from the datasets including only the b-value of 2500 s/mm2 showed excellent agreement (ICC, 0.96 to 0.99). Under the same acquisition time and diffusion directions, agreement was better of MK, K∥, and K⊥ maps obtained with 3 b-values (0, 1000, and 2500 s/mm2) and 4 signal averages than maps obtained with any other combination of numbers of b-value and varied NSA. Good agreement (ICC > 0.6) required at least 20 diffusion directions in all the metrics. Conclusion: MK and K⊥ maps with ICC greater than 0.95 can be obtained at 1.5T within 10 min (b-value = 0, 1000, and 2500 s/mm2; 20 diffusion directions; 4 signal averages; slice thickness, 6 mm with no interslice gap; number of slices, 12).
Purpose: We investigated the added value of the hypointensity on hepatocyte-phase (HP) imaging of gadoxetic acid-enhanced MRI (EOB-MRI) in the 2014 version of the Liver Imaging Reporting and Data System (LI-RADS) for distinguishing hepatocellular carcinoma (HCC) from benign hepatic lesions in patients with chronic liver disease. Methods: We retrospectively evaluated targeted lesions (111 HCCs, 28 benign hepatic lesions) of 139 patients (101 men, 38 women; aged 18 to 89 years, mean age, 68 ± 11 years) with chronic liver disease. EOB-MRI and dynamic contrast-enhanced computed tomography (CECT) were performed within 3 months. Two abdominal radiologists independently reviewed 3 imaging datasets: (1) EOB-MRI without an HP image using the LI-RADS system (MR imaging without HP); (2) EOB-MRI with an HP image using a modified version of the LI-RADS system in which hypointensity on the HP image was used as an additional major criterion of malignancy (MR imaging with HP); and (3) dynamic contrast-enhanced computed tomography (CECT) images using the LI-RADS system. We evaluated intra- and inter-reader agreement with kappa statistics along with 95% confidence intervals and compared diagnostic sensitivity and specificity of the 3 imaging datasets with McNemar’s test. Results: The sensitivities of MR imaging were statistically higher with HP (Reader 1, 95% [107/111]; Reader 2, 95% [106/111]) than without HP (Reader 1, 84% [93/111], P = 0.002; Reader 2, 86% [96/111], P = 0.002). Specificity was comparably high between MR imaging with HP (Reader 1, 96% [27/28]; Reader 2, 96% [27/28]) and dynamic CECT (Reader 1, 100% [28/28], P = 0.317; Reader 2, 100% [28/28], P = 0.317) and MR imaging without HP (Reader 1, 96% [27/28], P = 1.00; Reader 2, 100% [28/28], P = 0.317). Conclusion: The use of an HP image from EOB-MRI as an additional major criterion improved the sensitivity of LI-RADS to distinguish HCCs from benign hepatic lesions while retaining high specificity.
Purpose: We performed a quantitative intraindividual comparison of the performance of 0.025- and 0.05-mmol/kg doses for gadoxetic acid-enhanced liver magnetic resonance (MR) imaging. Materials and Methods: Eleven healthy volunteers underwent liver MR imaging twice, once with a 0.025- and once with a 0.05-mmol/kg dose of gadoxetic acid. MR spectroscopy and 3-dimensional gradient-echo T1-weighted images (3D-GRE) were obtained before and 3, 10, and 20 min after injection of the contrast medium to measure T1 and T2 values and signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) performance. During the dynamic phase, highly time-resolved 3D-GRE was used to estimate the relative CNR (CNRrel) of the hepatic artery and portal vein (PV) to the liver. We used paired t-tests to compare the results of different doses. Results: During the hepatobiliary phase, we observed shorter T1 values and higher SNRs of the liver (P < 0.001) and higher liver-to-PV and liver-to-muscle CNRs (P < 0.002) using 0.05 mmol/kg compared to 0.025 mmol/kg. Increasing the dose to 0.05 mmol/kg yielded a greater T1-shortening effect at 10 min delay even compared with 0.025 mmol/kg at 20 min (P < 0.001). During the dynamic phase, the peak CNRrel for the hepatic artery and portal vein were higher using 0.05 mmol/kg (P = 0.007 to 0.035). Conclusion: Use of gadoxetic acid at a dose of 0.05 mmol/kg leads to significantly higher SNR and CNR performance than with 0.025 mmol/kg. Quantitatively, a 10-min delay may be feasible for hepatobiliary-phase imaging when using 0.05 mmol/kg of gadoxetic acid.
Purpose: Our study aim was to clarify the characteristics of hemangiomas with pseudo washout sign (PWS) by comparing their features with those of hemangiomas without PWS on gadolinium-ethoxybenzyl-diethylenetriaminepentaacetic acid (Gd-EOB-DTPA)-enhanced magnetic resonance (MR) imaging. Methods: We evaluated the features of hemangiomas on Gd-EOB-DTPA-enhanced MR imaging of 70 hepatic hemangiomas in 31 patients, investigating the presence of peripheral or central nodular enhancement, diffuse enhancement, and arterioportal shunt during the arterial phase, fill-in enhancement during the portal venous phase, and PWS, which is low signal intensity during the late phase. We visually assessed the intensity of contrast enhancement of the lesion during the arterial, portal venous, late, and hepatobiliary phases using a 4-grade scale and used the Fisher exact and Mann-Whitney U tests to compare hemangiomas with and without PWS. Results: We observed PWS in 33 (47%) of 70 hemangiomas, which were significantly smaller than the hemangiomas without PWS (17.4 mm ± 20.3 versus 30.1 mm ± 28.5; P = 0.005); more frequent diffuse enhancement in hemangiomas with PWS than those without (21.2% versus 2.7%; P = 0.026); and no significant differences in nodular enhancement (P = 0.231), arterioportal shunt (P = 0.403), or fill-in enhancement (P = 0.357) between hemangiomas with and without PWS. Visually determined grades of tumor contrast enhancement were significantly lower in hemangiomas with PWS during the portal venous (P = 0.007) and late (P < 0.001) phases. Conclusions: Small hemangiomas tend to decrease in signal intensity during the portal venous phase and show PWS during the late phase.
Purpose: Q-space imaging (QSI) is a diffusion-weighted imaging (DWI) technique that enables investigation of tissue microstructure. However, for sufficient displacement resolution to measure the microstructure, QSI requires high q-values that are usually difficult to achieve with a clinical scanner. The recently introduced “low q-value method” fits the echo attenuation to only low q-values to extract the root mean square displacement. We investigated the clinical feasibility of the low q-value method for estimating the microstructure of the human corpus callosum using a 3.0-tesla clinical scanner within a clinically feasible scan time. Methods: We performed a simulation to explore the acceptable range of maximum q-values for the low q-value method. We simulated echo attenuations caused by restricted diffusion in the intra-axonal space (IAS) and hindered diffusion in the extra-axonal space (EAS) assuming 100,000 cylinders with various diameters, and we estimated mean axon diameter, IAS volume fraction, and EAS diffusivity by fitting echo attenuations with different maximum q-values. Furthermore, we scanned the corpus callosum of 7 healthy volunteers and estimated the mean axon diameter and IAS volume fraction. Results: Good agreement between estimated and defined values in the simulation study with maximum q-values of 700 and 800 cm−1 suggested that the maximum q-value used in the in vivo experiment, 737 cm−1, was reasonable. In the in vivo experiment, the mean axon diameter was larger in the body of the corpus callosum and smaller in the genu and splenium, and this anterior-to-posterior trend is consistent with previously reported histology, although our mean axon diameter seems larger in size. On the other hand, we found an opposite anterior-to-posterior trend, with high IAS volume fraction in the genu and splenium and a lower fraction in the body, which is similar to the fiber density reported in the histology study. Conclusion: The low q-value method may provide insights into tissue microstructure using a 3T clinical scanner within clinically feasible scan time.
Purpose: We propose and assess 2 novel asymmetric Fourier imaging (AFI) techniques, magnitude-based AFI (MagAFI) and MagAFI combined with projection on to convex sets (POCS) (MagAFI+POCS). MagAFI does not require phase information because it uses only the magnitude image with zero-filling. MagAFI+POCS requires phase information but further reduces image errors. Materials and Methods: We initially compared phase maps obtained using asymmetrically sampled data for the whole of the k-space and symmetrically sampled data for the low frequency part of the k-space. We used one-dimensional simulation data and 3-dimensional gradient echo data for 2 different echo times (TEs) of the brains of volunteers and assessed the differences between the image reconstructed from the full k-space data and AFI images reconstructed from truncated k-space data. We generated AFI images in this study using the zero-filling, Margosian (homodyne), Margosian+POCS (standard POCS), MagAFI, and MagAFI+POCS techniques. Results: We confirmed the assumption of smaller phase errors for the full k-space data than for the symmetric low frequency k-space data. Our proposed MagAFI technique provides images with smaller phase-induced errors than those obtained using conventional methods, including standard POCS methods, which have been regarded as the best methods. MagAFI+POCS improves image quality as well as robustness. Conclusion: Our proposed MagAFI technique achieves a practical balance of image quality and simplicity to perform better than conventional methods using only the 0-filled magnitude image. Combined with POCS, this technique can produce images of even better quality.
Purpose: Hyperintense vessels (HVs) on fluid-attenuated inversion recovery (FLAIR) imaging are associated with the leptomeningeal collateral circulation in cases of arterial occlusive lesions. Nevertheless, the relationship between HVs on FLAIR imaging and arterial circulation time (ACT) on cerebral angiography has not been defined. Methods: We analyzed images of 11 patients with acute occlusion of the distal internal carotid artery or proximal middle cerebral artery and calculated the difference in ACT (DACT) between infarcted and normal hemispheres. ACT was defined as the time interval from the initial opacification of the ipsilateral or contralateral cavernous internal carotid artery to the late arterial phase of the carotid artery territories. We scored HVs on FLAIR imaging using a modified Alberta Stroke Program Early Computerized Tomography Score (ASPECTS) and determined collateral circulation by grading collateral flow. Results: We detected HVs on FLAIR images in 10 patients (median score, 4; range, 0 to 6). Comparison of infarcted and normal hemispheres demonstrated absent or subtle HVs on FLAIR imaging when the DACT was too short (<one second) or too long (>7.98 s) and prominent HVs with moderate DACT (2 to 5 s). The score of HVs on FLAIR was estimated well by DACT using a quadratic regression model (R2 = 0.602) and better than by grading collateral flow (R2 = 0.256). Conclusion: In cases of large arterial occlusion, the hyperintensity of vessels on FLAIR images may be dependent on arterial circulation time via retrograde filling of the leptomeningeal collateral circulation.
Purpose: We attempted to clarify the relationship between the signal intensity (SI) in the hepatobiliary phase of gadoxetic acid-enhanced magnetic resonance (MR) imaging and the efficacy of hepatic arterial infusion chemotherapy (HAIC) in hepatocellular carcinomas (HCCs). Methods: We enrolled 14 patients with HCCs who underwent gadoxetic acid-enhanced MR imaging prior to HAIC using cisplatin and 5-fluorouracil. In the hepatobiliary phase, we calculated the SI of the HCCs and the background liver. In cases with multiple HCCs, we calculated the SI of the largest lesion. Patients were classified into high (n = 7) and low intensity (n = 7) groups based on the median value of the SI ratio (SI of the tumor/SI of the background liver). We analyzed progression-free survival using the Kaplan-Meier method and the log-rank test. In the 5 patients with a history of HCC surgery, we compared the expression of immunohistochemical organic anion-transporting polypeptide (OATP) 8 between the high and low intensity groups by chi-square test. Results: The SI ratios were 0.568 ± 0.093 (mean ± standard deviation) in the high intensity group and 0.251 ± 0.086 in the low intensity group. Compared to the group with low signal intensity, the group with high signal intensity demonstrated significantly lower serum levels of alpha fetoprotein (AFP) (P = 0.0350), significantly higher progression-free survival (P = 0.0108), better differentiation of tumor grade at histologic examination (P = 0.0253), and significantly higher OATP8 expression (P = 0.0253). Conclusion: Patients with HCCs of high SI ratio in the hepatobiliary phase of gadoxetic acid-enhanced MR imaging can respond better to HAIC.
Background and Purpose: We analyzed the ability of a machine learning approach that uses diffusion tensor imaging (DTI) structural connectomes to determine lateralization of epileptogenicity in temporal lobe epilepsy (TLE). Materials and Methods: We analyzed diffusion tensor and 3-dimensional (3D) T1-weighted images of 44 patients with TLE (right, 15, left, 29; mean age, 33.0 ± 11.6 years) and 14 age-matched controls. We constructed a whole brain structural connectome for each subject, calculated graph theoretical network measures, and used a support vector machine (SVM) for classification among 3 groups (right TLE versus controls, left TLE versus controls, and right TLE versus left TLE) following a feature reduction process with sparse linear regression. Results: In left TLE, we found a significant decrease in local efficiency and the clustering coefficient in several brain regions, including the left posterior cingulate gyrus, left cuneus, and both hippocampi. In right TLE, the right hippocampus showed reduced nodal degree, clustering coefficient, and local efficiency. With use of the leave-one-out cross-validation strategy, the SVM classifier achieved accuracy of 75.9 to 89.7% for right TLE versus controls, 74.4 to 86.0% for left TLE versus controls, and 72.7 to 86.4% for left TLE versus right TLE. Conclusion: Machine learning of graph theoretical measures from the DTI structural connectome may give support to lateralization of the TLE focus. The present good discrimination between left and right TLE suggests that, with further refinement, the classifier should improve presurgical diagnostic confidence.
Purpose: We evaluated the feasibility of contrast-enhanced steady-state free precession (ceSSFP) in the assessment of myocardial injury and obstruction of the left ventricular outflow tract (LVOT) in patients with hypertrophic obstructive cardiomyopathy (HOCM) after alcohol septal ablation (ASA). Methods: Twelve patients with HOCM underwent 16 magnetic resonance (MR) examinations following ASA. Precontrast SSFP, ceSSFP and late gadolinium enhancement (LGE) imaging were performed with a 1.5-tesla imager. ceSSFP was performed 3 to 7 min after gadolinium injection. We visually and quantitatively evaluated the signal patterns of the myocardium after ASA on SSFP and LGE MR imaging. We observed the LVOT using ceSSFP in the 3-chamber view. Results: We could visualize ASA-induced myocardial infarction (MI) in all 16 studies by LGE and ceSSFP but in only 6 studies (37.5%) by precontrast SSFP. Contrast was higher between MI and remote myocardium with LGE than ceSSFP (P < 0.01). ASA-induced hypointense regions were well visualized by the 2 sequences after contrast in the 7 patients who underwent MR imaging within 7 weeks of ASA and in a few patients after 80 weeks from ASA. The ceSSFP allowed comparable visualization of the jet flow crossing the LVOT to that derived from echocardiographic data. Conclusion: Contrast-enhanced steady-state free precession allows assessment of myocardial injury as well as of the left ventricular outflow tract after alcohol septal ablation in a single scan without penalty in scan time and cine imaging contrast.
Purpose: We retrospectively evaluated the incidence and related factors of obliteration of the lower bile duct after oral administration of contrast medium (OCM) probably resulting from its regurgitation into the biliary system (OCMRB) as observed on images of MR cholangiopancreatography (MRCP). Methods: We retrospectively analyzed 305 MRCP images in 278 patients obtained between February 2010 and March 2011 using negative OCM with 1.0- and 1.5-tesla clinical units. OCMRB was defined as positive when visualization of the common bile duct was clear on precontrast 2-dimensional (2D) MRCP but obliterated on postcontrast 3-dimensional (3D) MRCP. Two abdominal radiologists reviewed all images in consensus. The incidence of OCMRB was correlated to various clinicoradiological factors. Results: We observed OCMRB on 11 MRCP images in 10 patients (3.6%). Among various clinicoradiological factors, the presence of juxtapapillary diverticula, pneumobilia, and history of intervention to the papilla were suggested as significant factors related to positive OCMRB with multivariate analysis (P < 0.05). Conclusion: OCMRB occurs in about 4% of the patients who undergo MRCP, typically in those with juxtapapillary diverticula, pneumobilia, and history of papillary intervention. Acquisition of MRCP images before OCM may secure visualization of the common bile duct in these patients.
We propose a novel image processing technique that combines images routinely acquired with low and high b values to create a single image that contains clinically useful information without the ambiguity of T2 shine-through. The contrast of resulting images is similar to that of a T2 image, but the signals of pixels with low apparent diffusion coefficient (ADC) values are inverted. The proposed technique takes the threshold ADC value as the one adjustable parameter.