Validation of a Denoising Method Using Deep Learning-Based Reconstruction to Quantify Multiple Sclerosis Lesion Load on Fast FLAIR Imaging

BACKGROUND AND PURPOSE

Accurate quantification of WM lesion load is essential for the care of patients with multiple sclerosis. We tested whether the combination of accelerated 3D-FLAIR and denoising using deep learning-based reconstruction could provide a relevant strategy while shortening the imaging examination.

MATERIALS AND METHODS

Twenty-eight patients with multiple sclerosis were prospectively examined using 4 implementations of 3D-FLAIR with decreasing scan times (4 minutes 54 seconds, 2 minutes 35 seconds, 1 minute 40 seconds, and 1 minute 15 seconds). Each FLAIR sequence was reconstructed without and with denoising using deep learning-based reconstruction, resulting in 8 FLAIR sequences per patient. Image quality was assessed with the Likert scale, apparent SNR, and contrast-to-noise ratio. Manual and automatic lesion segmentations, performed randomly and blindly, were quantitatively evaluated against ground truth using the absolute volume difference, true-positive rate, positive predictive value, Dice similarity coefficient, Hausdorff distance, and F1 score based on the lesion count. The Wilcoxon signed-rank test and 2-way ANOVA were performed.

RESULTS

Both image-quality evaluation and the various metrics showed deterioration when the FLAIR scan time was accelerated. However, denoising using deep learning-based reconstruction significantly improved subjective image quality and quantitative performance metrics, particularly for manual segmentation. Overall, denoising using deep learning-based reconstruction helped to recover contours closer to those from the criterion standard and to capture individual lesions otherwise overlooked. The Dice similarity coefficient was equivalent between the 2-minutes-35-seconds-long FLAIR with denoising using deep learning-based reconstruction and the 4-minutes-54-seconds-long reference FLAIR sequence.

CONCLUSIONS

Denoising using deep learning-based reconstruction helps to recognize multiple sclerosis lesions buried in the noise of accelerated FLAIR acquisitions, a possibly useful strategy to efficiently shorten the scan time in clinical practice.

Influence of inflammatory status in the acute phase of stroke on post-stroke depression

BACKGROUND

Thirty percent of stroke patients will suffer from post-stroke depression (PSD). Recent data suggest that inflammation accounts for a substantial amount of depression. Our primary objective was to assess the association between standard inflammation biomarkers in the acute phase of stroke and PSD at three months. The secondary objective was to elaborate a predictive model of PSD from clinical, biological and radiological data.

METHODS

We performed a retrospective analysis of a single-centre cohort of stroke patients with a three-month follow-up. Serum levels of C-reactive protein (CRP), fibrinogen, leukocyte count and neutrophil to lymphocyte ratio (NLR) were tested at admission and at peak. Mood was assessed at three months using the depression sub-scale of the Hospital Anxiety and Depression Scale (HADS). Association between inflammation biomarkers and HADS was evaluated with multi-linear regression adjusted on clinical and radiological parameters. Logistic predictive models of PSD at three months, with and without inflammation biomarkers, were compared.

RESULTS

Three hundred and forty-eight patients were included, of whom 20.06% developed PSD. Baseline and peak values of all inflammatory markers were associated with the severity of PSD at three months. Area under the curve for the receiver operating characteristic curve of PSD prediction was 0.746 (CI 95% 0.592-0.803) with selected inflammation biomarkers and 0.744 (CI 95% 0.587-0.799) without.

CONCLUSION

Most inflammation biomarkers are weakly associated with PSD, adding negligible value to predictive models. While they suggest the implication of inflammation in PSD pathogenesis, they are useless for the prediction of PSD, underscoring the need for more specific biomarkers.

Acute Ischemic Stroke or Epileptic Seizure? Yield of CT Perfusion in a "Code Stroke" Situation

BACKGROUND AND PURPOSE

The clinical differentiation between acute ischemic stroke and epileptic seizure may be challenging, and making the correct diagnosis could avoid unnecessary reperfusion therapy. We examined the accuracy of CTP in discriminating epileptic seizures from acute ischemic stroke without identified arterial occlusion.

MATERIALS AND METHODS

We retrospectively identified consecutive patients in our emergency department who underwent CTP in the 4.5 hours following the development of an acute focal neurologic deficit who were discharged with a final diagnosis of acute ischemic stroke or epileptic seizure.

RESULTS

Among 95 patients, the final diagnosis was epileptic seizure in 45 and acute ischemic stroke in 50. CTP findings were abnormal in 73% of the patients with epileptic seizure and 40% of those with acute ischemic stroke. Hyperperfusion was observed more frequently in the seizure group (36% versus 2% for acute ischemic stroke) with high specificity (98%) but low sensitivity (35%) for the diagnosis of epileptic seizure. Hypoperfusion was found in 38% of cases in each group and was not confined to a vascular territory in 24% of patients in the seizure group and 2% in the acute ischemic stroke group. The interobserver agreement was good (κ = 0.60) for hypo-, hyper-, and normoperfusion patterns and moderate (κ = 0.41) for the evaluation of vascular systematization.

CONCLUSIONS

CTP patterns helped to differentiate acute ischemic stroke from epileptic seizure in a "code stroke" situation. Our results indicate that a hyperperfusion pattern, especially if not restricted to a vascular territory, may suggest reconsideration of intravenous thrombolysis therapy.

Advanced multiparametric magnetic resonance imaging of multinodular and vacuolating neuronal tumor

BACKGROUND AND PURPOSE

Multinodular and vacuolating neuronal tumor (MVNT) of the cerebrum is a rare brain lesion with suggestive imaging features. The aim of our study was to report the largest series of MVNTs so far and to evaluate the utility of advanced multiparametric magnetic resonance (MR) techniques.

METHODS

This multicenter retrospective study was approved by our institutional research ethics board. From July 2014 to May 2019, two radiologists read in consensus the MR examinations of patients presenting with a lesion suggestive of an MVNT. They analyzed the lesions' MR characteristics on structural images and advanced multiparametric MR imaging.

RESULTS

A total of 64 patients (29 women and 35 men, mean age 44.2 ± 15.1 years) from 25 centers were included. Lesions were all hyperintense on fluid-attenuated inversion recovery and T2-weighted imaging without post-contrast enhancement. The median relative apparent diffusion coefficient on diffusion-weighted imaging was 1.13 [interquartile range (IQR), 0.2]. Perfusion-weighted imaging showed no increase in perfusion, with a relative cerebral blood volume of 1.02 (IQR, 0.05) and a relative cerebral blood flow of 1.01 (IQR, 0.08). MR spectroscopy showed no abnormal peaks. Median follow-up was 2 (IQR, 1.2) years, without any changes in size.

CONCLUSIONS

A comprehensive characterization protocol including advanced multiparametric magnetic resonance imaging sequences showed no imaging patterns suggestive of malignancy in MVNTs. It might be useful to better characterize MVNTs.

Multinodular and Vacuolating Posterior Fossa Lesions of Unknown Significance

Multinodular and vacuolating neuronal tumor of the cerebrum is a rare supratentorial brain tumor described for the first time in 2013. Here, we report 11 cases of infratentorial lesions showing similar striking imaging features consisting of a cluster of low T1-weighted imaging and high T2-FLAIR signal intensity nodules, which we referred to as multinodular and vacuolating posterior fossa lesions of unknown significance. No relationship was found between the location of the lesion and clinical symptoms. A T2-FLAIR hypointense central dot sign was present in images of 9/11 (82%) patients. Cortical involvement was present in 2/11 (18%) of patients. Only 1 nodule of 1 multinodular and vacuolating posterior fossa lesion of unknown significance showed enhancement on postcontrast T1WI. DWI, SWI, MRS, and PWI showed no malignant pattern. Lesions did not change in size or signal during a median follow-up of 3 years, suggesting that multinodular and vacuolating posterior fossa lesions of unknown significance are benign malformative lesions that do not require surgical intervention or removal.

Cervical Spinal Cord DTI Is Improved by Reduced FOV with Specific Balance between the Number of Diffusion Gradient Directions and Averages

BACKGROUND AND PURPOSE

Reduced-FOV DTI is promising for exploring the cervical spinal cord, but the optimal set of parameters needs to be clarified. We hypothesized that the number of excitations should be favored over the number of diffusion gradient directions regarding the strong orientation of the cord in a single rostrocaudal axis.

MATERIALS AND METHODS

Fifteen healthy individuals underwent cervical spinal cord MR imaging at 3T, including an anatomic 3D-Multi-Echo Recombined Gradient Echo, high-resolution full-FOV DTI with a NEX of 3 and 20 diffusion gradient directions and 5 sets of reduced-FOV DTIs differently balanced in terms of NEX/number of diffusion gradient directions: (NEX/number of diffusion gradient directions = 3/20, 5/16, 7/12, 9/9, and 12/6). Each DTI sequence lasted 4 minutes 30 seconds, an acceptable duration, to cover C1-C4 in the axial plane. Fractional anisotropy maps and tractograms were reconstructed. Qualitatively, 2 radiologists rated the DTI sets blinded to the sequence. Quantitatively, we compared distortions, SNR, variance of fractional anisotropy values, and numbers of detected fibers.

RESULTS

Qualitatively, reduced-FOV DTI sequences with a NEX of ≥5 were significantly better rated than the full-FOV DTI and the reduced-FOV DTI with low NEX (N = 3) and a high number of diffusion gradient directions (D = 20). Quantitatively, the best trade-off was reached by the reduced-FOV DTI with a NEX of 9 and 9 diffusion gradient directions, which provided significantly fewer artifacts, higher SNR on trace at b = 750 s/mm² and an increased number of fibers tracked while maintaining similar fractional anisotropy values and dispersion.

CONCLUSIONS

Optimized reduced-FOV DTI improves spinal cord imaging. The best compromise was obtained with a NEX of 9 and 9 diffusion gradient directions, which emphasizes the need for increasing the NEX at the expense of the number of diffusion gradient directions for spinal cord DTI contrary to brain DTI.

In Vivo 7T MR Quantitative Susceptibility Mapping Reveals Opposite Susceptibility Contrast between Cortical and White Matter Lesions in Multiple Sclerosis

BACKGROUND AND PURPOSE

Magnetic susceptibility measured with quantitative susceptibility mapping has been proposed as a biomarker for demyelination and inflammation in patients with MS, but investigations have mostly been on white matter lesions. A detailed characterization of cortical lesions has not been performed. The purpose of this study was to evaluate magnetic susceptibility in both cortical and WM lesions in MS by using quantitative susceptibility mapping.

MATERIALS AND METHODS

Fourteen patients with MS were scanned on a 7T MR imaging scanner with T1-, T2-, and T2*-weighted sequences. The T2*-weighted sequence was used to perform quantitative susceptibility mapping and generate tissue susceptibility maps. The susceptibility contrast of a lesion was quantified as the relative susceptibility between the lesion and its adjacent normal-appearing parenchyma. The susceptibility difference between cortical and WM lesions was assessed by using a t test.

RESULTS

The mean relative susceptibility was significantly negative for cortical lesions (P < 10^-7) but positive for WM lesions (P < 10^-22). A similar pattern was also observed in the cortical (P = .054) and WM portions (P = .043) of mixed lesions.

CONCLUSIONS

The negative susceptibility in cortical lesions suggests that iron loss dominates the susceptibility contrast in cortical lesions. The opposite susceptibility contrast between cortical and WM lesions may reflect both their structural (degree of myelination) and pathologic (degree of inflammation) differences, in which the latter may lead to a faster release of iron in cortical lesions.

MS lesions are better detected with 3D T1 gradient-echo than with 2D T1 spin-echo gadolinium-enhanced imaging at 3T

BACKGROUND AND PURPOSE

In multiple sclerosis, gadolinium enhancement is used to classify lesions as active. Regarding the need for a standardized and accurate method for detection of multiple sclerosis activity, we compared 2D-spin-echo with 3D-gradient-echo T1WI for the detection of gadolinium-enhancing MS lesions.

MATERIALS AND METHODS

Fifty-eight patients with MS were prospectively imaged at 3T by using both 2D-spin-echo and 3D-gradient recalled-echo T1WI in random order after the injection of gadolinium. Blinded and independent evaluation was performed by a junior and a senior reader to count gadolinium-enhancing lesions and to characterize their location, size, pattern of enhancement, and the relative contrast between enhancing lesions and the adjacent white matter. Finally, the SNR and relative contrast of gadolinium-enhancing lesions were computed for both sequences by using simulations.

RESULTS

Significantly more gadolinium-enhancing lesions were reported on 3D-gradient recalled-echo than on 2D-spin-echo (n = 59 versus n = 30 for the junior reader, P = .021; n = 77 versus n = 61 for the senior reader, P = .017). The difference between the 2 readers was significant on 2D-spin-echo (P = .044), for which images were less reproducible (κ = 0.51) than for 3D-gradient recalled-echo (κ = 0.65). Further comparisons showed that there were statistically more small lesions (<5 mm) on 3D-gradient recalled-echo than on 2D-spin-echo (P = .04), while other features were similar. Theoretic results from simulations predicted SNR and lesion contrast for 3D-gradient recalled-echo to be better than for 2D-spin-echo for visualization of small enhancing lesions and were, therefore, consistent with clinical observations.

CONCLUSIONS

At 3T, 3D-gradient recalled-echo provides a higher detection rate of gadolinium-enhancing lesions, especially those with smaller size, with a better reproducibility; this finding suggests using 3D-gradient recalled-echo to detect MS activity, with potential impact in initiation, monitoring, and optimization of therapy.

Atraumatic nonaneurysmal sulcal subarachnoid hemorrhages: a diagnostic workup based on a case series

INTRODUCTION

Atraumatic and nonaneurysmal sulcal subarachnoid hemorrhage (sSAH) is a rare type of cerebrovascular disease with various etiologies previously reported in small case reports. In this study, we propose to analyze clinical presentations, imaging patterns and etiologies in a large case series of such patients in order to propose a diagnostic workup.

METHODS

We retrospectively analyzed clinical and radiological data of consecutive patients with a diagnosis of atraumatic and nonaneurysmal sSAH, admitted to our institution between 2008 and 2011. All patients had both computed tomography (CT) and magnetic resonance imaging (MRI) as a part of their initial evaluation.

RESULTS

30 patients (18 women and 12 men, mean age: 60 years) were identified. The main clinical symptoms at presentation were focal and transient neurological deficit (n = 22) and thunderclap headache (n = 10). Four patients had progressive headache and 4 other had partial or generalized epileptic seizures. MRI abnormalities associated with sSAH were prior hemorrhages, microbleeds, severe leukoencephalopathy and hemosiderosis suggesting cerebral amyloid angiopathy (CAA; n = 9), vasogenic edema in parieto-occipital areas compatible with a posterior reversible encephalopathy syndrome (PRES; n = 3), cortical venous thrombosis (n = 2) and concomitant acute cortical stroke (n = 3). Other underlying causes of sSAH, not diagnosed on MRI, were reversible cerebral vasoconstriction syndrome (RCVS) based on clinical criteria and conventional angiography (n = 4), angiitis diagnosed by skin biopsy (n = 1), vascular malformation diagnosed on CT and digital subtraction angiographies (n = 3), and overanticoagulation (n = 1). Four cases remained unresolved.

CONCLUSION

This study confirmed that sSAH is a rare condition related to a wide spectrum of etiologies. Combination of brain MRI and magnetic resonance angiography and eventually digital subtraction angiography allowed the identification of an underlying etiology for 87% of patients. CAA, RCVS and PRES represented more than 50% of the etiological mechanisms. Among older patients, sSAH was mainly related to CAA while in younger patients, RCVS represented the most frequent etiology.

Perianeurysmal brain inflammation after flow-diversion treatment

BACKGROUND AND PURPOSE

Flow-diverter stents are an alternative treatment for challenging and recurrent aneurysms. Thrombosis of the sac is thought to induce perianeurysmal brain inflammation, but such phenomena have never been studied in flow-diverter devices. We developed imaging data to explain the clinical exacerbation of symptoms after flow-diversion treatment.

MATERIALS AND METHODS

Seventeen patients with unruptured aneurysms were treated by using a flow-diverter device. Clinical symptoms and angiographic and MR imaging features were recorded before and after treatment, during both the acute and chronic phases, to look for inflammatory reaction.

RESULTS

Seven of the 17 patients (41%) showed a delayed clinical aggravation of symptoms posttreatment consisting of a headache (n = 7) with aggravation of pre-existing compressive symptoms (n = 4) and the appearance of compressive symptoms (n = 1). This clinical deterioration was transient; it was observed between 3 and 15 days posttreatment and resolved by day 30. MR imaging revealed signs highly suggestive of perianeurysmal inflammation with vasogenic edema and blood-brain barrier breakdown. The association between MR imaging inflammatory features and clinical aggravation was statistically significant. Large aneurysmal size and its proximity to surrounding brain tissue were predictive of this inflammatory reaction after flow diversion.

CONCLUSIONS

The main finding of the series is that MR imaging-defined perianeurysmal inflammation is observed with a high frequency after treatment of unruptured aneurysms with flow diverters and is, in most cases, associated with a transient clinical deterioration.

Descriptive epidemiology of CNS tumors in France: results from the Gironde Registry for the period 2000-2007

An increase in the incidence of CNS tumors has been observed in many countries in the last decades. The reality of this trend has been much debated, as it has happened during a period when computer-assisted tomography and MRI have dramatically improved the detection of these tumors. The Gironde CNS Tumor Registry provides here the first data on CNS tumor incidence and trends in France for all histological types, including benign and malignant tumors, for the period 2000-2007. Incidence rates were calculated globally and for each histological subtype. For trends, a piecewise log-linear model was used. The overall annual incidence rate was found to be 17.6/100 000. Of this rate, 7.9/100 000 were neuroepithelial tumors and 6.0/100 000 were meningiomas. An overall increase in CNS tumor incidence was observed from 2000 to 2007, with an annual percent change (APC) of +2.33%, which was explained mainly by an increase in the incidence of meningiomas over the 8-year period (APC = +5.4%), and also more recently by an increase in neuroepithelial tumors (APC = +7.45% from 2003). The overall increase was more pronounced in women and in the elderly, with an APC peaking at +24.65% in subjects 85 and over. The increase in the incidence rates we observed may have several explanations: not only improvements in registration, diagnosis, and clinical practice, but also changes in potential risk factors.

[Grading of adults primitive glial neoplasms using arterial spin-labeled perfusion MR imaging]

PURPOSE

We investigated the relationship between tumor blood-flow measurement based on perfusion-imaging by arterial spin-labeling (ASL) and histopathologic findings in adults' primitive glial tumours.

PATIENTS AND METHODS

Thus, 40 primitive brain tumors (8 low-grade and 32 high-grade gliomas according to the Sainte-Anne classification) were imaged using pulsed (n=19) or continuous (n=21) ASL. Relative cerebral blood flow (rCBF=tumoral blood flow/normal cerebral blood flow) between high- and low-grade gliomas were compared.

RESULTS

Using pulsed ASL, differences in mean rCBF were observed in high- and low-grade gliomas although no significant (respectively 1.95 and 1.5). Using continuous ASL, mean rCBF were significantly higher for high-grade than for low-grade gliomas (P<0.05). High-grade gliomas could be discriminated using a CBF threshold of 1.18, with a sensitivity of 88%, specificity of 60%, predictive positive value of 88%, and predictive negative value of 60%.

CONCLUSION

ASL-based perfusion provides a quantitative, non-invasive alternative to dynamic susceptibility contrast perfusion MR methods for evaluating CBF. ASL is a suitable method for gliomas initial staging and could be useful to identify intermediate tumoral evolution.

Final cerebral infarct volume is predictable by MR imaging at 1 week

BACKGROUND AND PURPOSE

Stroke volume, an increasingly used end point in phase II trials, is considered stationary at least 30 days after the ictus. We investigated whether information conveyed by MR imaging measurements of the "final" infarct volume could be assessed as early as the subacute stage (days 3-6), rather than waiting for the chronic stage (days 30-45).

MATERIALS AND METHODS

Ninety-five patients with middle cerebral artery stroke prospectively included in a multicenter study underwent MR imaging during the first 12 hours (MR imaging-1), between days 3 and 6 (MR imaging-2), and between days 30 and 45 (MR imaging-3). We first investigated the relationship between subacute (FLAIR-2) and chronic volumes (FLAIR-3), by using a linear regression model. We then tested the relationship between FLAIR volumes (either FLAIR-2 or FLAIR-3) and functional disability, measured by the mRS at the time of MR imaging-3, by using logistic regression. The performances of the models were assessed by using the AUC in ROC.

RESULTS

A linear association between log FLAIR-2 and log FLAIR-3 volumes was observed. The proportion of FLAIR-3 variation, explained by FLAIR-2, was high (R(2) = 81%), without a covariate that improved this percentage. Both FLAIR-2 and FLAIR-3 were independent predictors of mRS (OR, 0.79 and 0.73; 95% CI, 0.64-0.97 and 0.56-0.96; P = .026 and .023). The performances of the models for the association between either FLAIR volume and mRS did not differ (AUC = 0.897 for FLAIR-2 and 0.888 for FLAIR-3).

CONCLUSIONS

Stroke damage may be assessed by a subacute volume because subacute volume predicts the "true" final volume and provides the same clinical prognosis.

Reliability of the ECASS radiological classification of postthrombolysis brain haemorrhage: a comparison of CT and three MRI sequences

BACKGROUND

Postthrombolysis brain haemorrhagic transformations (HT) are often categorized with the CT-based classification of the European Cooperative Acute Stroke Study (ECASS). However, little is known about the reliability of this classification and its extension to MRI. Our objective was to compare the inter- and intraobserver reliability of this classification on CT and 3 MRI sequences.

METHODS

Forty-three patients with postthrombolysis HT on CT or at least 1 of the 3 MRI sequences: fluid-attenuation inversion recovery (FLAIR), diffusion-weighted imaging (DWI), and T2* gradient recalled echo (T2*GRE) were selected. Twelve control patients without any bleeding were added to avoid a bias based on a pure HT-positive cohort. Each series of images were independently classified with the ECASS method by 6 blinded observers. Inter- and intraobserver reproducibility was categorized from poor to excellent depending on kappa values.

RESULTS

The inter- and intraobserver overall concordance of the classification was good for T2*GRE, DWI and CT (kappa > 0.6) and moderate for FLAIR (kappa < 0.6). The interobserver concordance for parenchymal haematomas was excellent for T2*GRE (kappa > 0.8) and moderate for CT, FLAIR and DWI.

CONCLUSION

The T2*GRE sequence is the most reproducible method to categorize postthrombolysis HT and has an excellent reliability for the severe parenchymal haematoma category, suggesting that this sequence should be used to assess HT in thrombolytic therapy trials.

[Magnetic resonance imaging of central nervous system inflammation]

Magnetic resonance imaging (MRI) is widely used to explore central nervous system inflammatory disorders, especially multiple sclerosis (MS). Advanced MRI methods are bringing more sensitive and specific tools for each step of the inflammatory process. In this review, we discuss the different MRI approaches for inflammatory disorders exploration, especially MS. We give particular emphasize on sensibility and specificity of each MRI approach and we also discuss the current knowledge concerning biological and histopathological substratum that could explain MRI signal with each modality.

Pulsed arterial spin labeling applications in brain tumors: practical review

Few institutions use MRI perfusion without contrast injection called arterial spins labeling (ASL) routinely in clinical setting. After general considerations concerning the different ASL techniques and quantitative issues, we will detail a pulsed sequence that can be used on a clinical 1.5-T MR unit. We will discuss and illustrate the use of ASL in tumoral diseases for diagnosis, gliomas grading, stereotactic biopsy guidance and for follow-up after treatment.

[Diffusion tensor imaging and tractography of the brain and spinal cord]

Diffusion tensor imaging is a magnetic resonance imaging technique that provides details on tissue microstructure and organization well beyond the usual image resolution. With diffusion tensor imaging, diffusion anisotropy can be quantified and subtle white matter changes not normally seen on conventional MRI can be detected. The aim of this article is to review the principles of diffusion tensor imaging and fiber tracking and their applications to the study of the brain, including Alzheimer disease, neuropsychiatric disorders, strokes, multiple sclerosis, brain tumors, and intractable seizures. Emerging applications to spinal cord disorders are also presented.