Fluid-attenuated Inversion Recovery Research Articles

Rapid imaging of intravenous gadolinium-based contrast agent (GBCA) entering ventricular cerebrospinal fluid (CSF) through the choroid plexus in healthy human subjects

BackgroundPathways for intravenously administered gadolinium-based-contrast-agents (GBCAs) entering cerebrospinal-fluid (CSF) circulation in the human brain are not well-understood. The blood-CSF-barrier (BCSFB) in choroid-plexus (CP) has long been hypothesized to be a main entry-point for intravenous-GBCAs into CSF. Most existing studies on this topic were performed in animals and human patients with various diseases. Results in healthy human subjects are limited. Besides, most studies were performed using MRI methods with limited temporal resolution and significant partial-volume effects from blood and CSF.MethodsThis study employs the recently developed dynamic-susceptibility-contrast-in-the-CSF (cDSC) MRI approach to measure GBCA-distribution in the CSF immediately and 4 h after intravenous-GBCA administration in healthy subjects. With a temporal resolution of 10 s, cDSC MRI can track GBCA-induced CSF signal changes during the bolus phase, which has not been investigated previously. It employs a long echo-time (TE = 1347 ms) to suppress tissue and blood signals so that pure CSF signal is detected with minimal partial-volume effects. GBCA concentration in the CSF can be estimated from cDSC MRI. In this study, cDSC and FLAIR MRI were performed immediately and 4 h after intravenous GBCA administration in 25 healthy volunteers (age 48.9 ± 19.5 years; 14 females). Paired t-tests were used to compare pre-GBCA and post-GBCA signal changes, and their correlations with age were evaluated using Pearson-correlation-coefficients.ResultsAt ~ 20 s post-GBCA, GBCA-induced cDSC signal changes were detected in the CSF around CP (ΔS/S = − 2.40 ± 0.30%; P < .001) but not in the rest of lateral ventricle (LV). At 4 h, significant GBCA-induced cDSC signal changes were observed in the entire LV (ΔS/S = − 7.58 ± 3.90%; P = .002). FLAIR MRI showed a similar trend. GBCA-induced CSF signal changes did not correlate with age.ConclusionsThese results provided direct imaging evidence that GBCAs can pass the BCSFB in the CP and enter ventricular CSF immediately after intravenous administration in healthy human brains. Besides, our results in healthy subjects established a basis for clinical studies in brain diseases exploiting GBCA-enhanced MRI to detect BCSFB dysfunction.

Quantifying the reproducibility and longitudinal repeatability of radiomics features in magnetic resonance Image-Guide accelerator Imaging: A phantom study

ObjectiveThis study aimed to quantitatively evaluate the inter-platform reproducibility and longitudinal acquisition repeatability of MRI radiomics features in Fluid-Attenuated Inversion Recovery (FLAIR), T2-weighted (T2W), and T1-weighted (T1W) sequences on MR-Linac systems using an American College of Radiology (ACR) phantom. Materials and MethodsThis study used two MR-Linac systems (A and B) in different cancer centers. The ACR phantom was scanned on system A daily for 30 consecutive days to evaluate longitudinal repeatability. Additionally, retest data were collected after repositioning the phantom. Inter-platform reproducibility was assessed by conducting scans under identical conditions using system B. Regions of interest were delineated on the T1W sequence from system A and mapped to other sequences via rigid registration. Intra-observer and inter-observer comparisons were conducted. Repeatability and reproducibility were assessed using the intraclass correlation coefficient (ICC) and coefficient of variation (CV). Robust radiomics features were identified based on ICC>0.9 and CV<10 %. ResultsAnalysis showed that a higher proportion of radiomics features derived from longitudinal FLAIR sequence (51.65 %) met robustness criteria compared to T2W (48.35 %) and T1W (43.96 %). Additionally, more inter-platform features from the FLAIR sequence (62.64 %) were robust compared to T2W (42.86 %) and T1W (39.56 %). Test-retest and intra-observer repeatability were excellent across all sequences, with a median ICC of 0.99 and CV<5%. However, inter-observer reproducibility was inferior, especially for the T1W sequence. ConclusionsDifferent sequences show variations in repeatability and reproducibility. The FLAIR sequence demonstrated advantages in both longitudinal repeatability and inter-platform reproducibility. Caution is warranted when interpreting data, particularly in longitudinal or multiplatform radiomics studies.

Hypertension and cerebral blood flow in the development of Alzheimer's disease.

We investigated the interactive associations between amyloid and hypertension on the entorhinal cortex (EC) tau and atrophy and the role of cerebral blood flow (CBF) as a shared mechanism by which amyloid and hypertension contribute to EC tau and regional white matter hyperintensities (WMHs). We analyzed data from older adults without dementia participating in the Add-Tau study (NCT02958670, n=138) or Alzheimer's Disease Neuroimaging Initiative (ADNI) (n=523) who had available amyloid-positron emission tomography (PET), tau-PET, fluid-attenuated inversion recovery (FLAIR), and T1-weighted magnetic resonance imaging (MRI). A subsample in both cohorts had available arterial spin labeling (ASL) MRI (Add-Tau: n=78; ADNI: n=89). The detrimental effects of hypertension on AD pathology and EC thickness were more pronounced in the Add-Tau cohort. Increased amyloid burden was associated with decreased occipital gray matter CBF in the ADNI cohort. In both cohorts, lower regional gray matter CBF was associated with higher EC tau and posterior WMH burden. Reduced cerebral perfusion may be one common mechanism through which hypertension and amyloid are related to increased EC tau and WMH volume. Hypertension is associated with increased entorhinal cortex (EC) tau, particularly in the presence of amyloid. Decreased cortical cerebral blood flow (CBF) is associated with higher regional white matter hyperintensity volume. Increasing amyloid burden is associated with decreasing CBF in the occipital lobe. MTL CBF and amyloid are synergistically associated with EC tau.

Utilizing Radiomics of Peri-Lesional Edema in T2-FLAIR Subtraction Digital Images to Distinguish High-Grade Glial Tumors From Brain Metastasis.

Differentiating high-grade glioma (HGG) and isolated brain metastasis (BM) is important for determining appropriate treatment. Radiomics, utilizing quantitative imaging features, offers the potential for improved diagnostic accuracy in this context. To differentiate high-grade (grade 4) glioma and BM using machine learning models from radiomics data obtained from T2-FLAIR digital subtraction images and the peritumoral edema area. Retrospective. The study included 1287 patients. Of these, 602 were male and 685 were female. Of the 788 HGG patients included in the study, 702 had solitary masses. Of the 499 BM patients included in the study, 112 had solitary masses. Initially, the model was developed and tested on solitary masses. Subsequently, the model was developed and tested separately for all patients (solitary and multiple masses). Axial T2-weighted fast spin-echo sequence (T2WI) and T2-weighted fluid-attenuated inversion recovery sequence (T2-FLAIR), using 1.5-T and 3.0-T scanners. Radiomic features were extracted from digitally subtracted T2-FLAIR images in the area of peritumoral edema. The maximum relevance-minimum redundancy (mRMR) method was then used for dimensionality reduction. The naive Bayes algorithm was used in model development. The interpretability of the model was explored using SHapley Additive exPlanations (SHAP). Chi-square test, one-way analysis of variance, and Kruskal-Wallis test were performed. The P values <0.05 were considered statistically significant. The performance metrics include area under curve (AUC), sensitivity (SENS), and specificity (SPEC). The mean age of HGG patients was 61.4 ± 13.2 years and 61.7 ± 12.2 years for BM patients. In the external validation cohort, the model achieved AUC: 0.991, SENS: 0.983, and SPEC: 0.922. The external cohort results for patients with solitary lesions were AUC: 0.987, SENS: 0.950, and SPEC: 0.922. The artificial intelligence model, developed with radiomics data from the peritumoral edema area in T2-FLAIR digital subtraction images, might be able to differentiate isolated BM from HGG. 3 TECHNICAL EFFICACY: Stage 2.

Prediction of Glioma enhancement pattern using a MRI radiomics-based model.

Contrast-MRI scans carry risks associated with the chemical contrast agents. Accurate prediction of enhancement pattern of gliomas has potential in avoiding contrast agent administration to patients. This study aimed to develop a machine learning radiomics model that can accurately predict enhancement pattern of gliomas based on T2 fluid attenuated inversion recovery images. A total of 385 cases of pathologically-proven glioma were retrospectively collected with preoperative magnetic resonance T2 fluid attenuated inversion recovery images, which were divided into enhancing and non-enhancing groups. Predictive radiomics models based on machine learning with 6 different classifiers were established in the training cohort (n = 201), and tested both in the internal validation cohort (n = 85) and the external validation cohort (n = 99). Receiver-operator characteristic curve was used to assess the predictive performance of these radiomics models. This study demonstrated that the radiomics model comprising of 15 features using the Gaussian process as a classifier had the highest predictive performance in both the training cohort and the internal validation cohort, with the area under the curve being 0.88 and 0.80, respectively. This model showed an area under the curve, sensitivity, specificity, positive predictive value and negative predictive value of 0.81, 0.98, 0.61, 0.82, 0.76 and 0.96, respectively, in the external validation cohort. This study suggests that the T2-FLAIR-based machine learning radiomics model can accurately predict enhancement pattern of glioma.

CSF Susceptibility Variation in Patient With Intracranial Hemorrhage: Implications for Quantitative Susceptibility Mapping Reference Selection.

Quantitative susceptibility mapping (QSM) is an emerging MRI technique with multiple clinical applications. As tissue susceptibility cannot be directly measured using MRI, QSM imaging techniques must indirectly compute susceptibility values, requiring regularization methods. CSF is a popular choice for regularization due to its near water susceptibility in healthy controls. However, the impact of pus, elevated protein, or blood dissolved in CSF on QSM regularization is not well defined. This study aimed to investigate the effects of intracranial hemorrhage (ICH) on selecting CSF as reference for QSM imaging. A total of 87 subjects, 53 with ICH (5 intraventricular, 19 subarachnoid, 27 both, and 2 intraparenchymal only) and 37 without hemorrhage (27 with MS, 10 without MS), were included in this study. Imaging was performed using 3D multiecho gradient echo, FLAIR, and multiecho complex total field inversion (mcTFI) at 3 T. McTFI with and without CSF zero-referencing regularization was generated from the 3DMEGRE data and reviewed with FLAIR images. Regions of hemorrhagic (H+) and nonhemorrhagic (H-) CSF were manually selected in reference to head CT and FLAIR images by a PGY III diagnostic radiology resident and Certificate of Added Qualification-certified neuroradiologist with 10 years' experience. Paired Student t test and one-way ANOVA were used with post hoc multicomparisons. A P value <0.05 was considered statistically significant. Areas of H- CSF were noted to have higher regularized QSM values in subjects with ICH relative to subjects without. Unregularized H- QSM values were also noted to have a systematically higher value in ICH subjects relative to subjects without blood. Subjects with MS and without ICH did not show significant difference in H- CSF regularized or unregularized QSM values. QSM values of areas suggested to not have hemorrhage on other imaging showed significantly higher QSM values in ICH subjects relative to subjects without ICH. Additionally, areas of hemorrhage did not show significant QSM value difference between regularized and unregularized QSM images. These findings suggest that, in subjects with any area of ICH, QSM values for no-hemorrhagic areas may be significantly altered using CSF regularization relative to subjects without ICH, with implications for intra- and intersubject QSM value analysis.

Comparing the imaging characteristics of middle-aged patients with multiple sclerosis and CADASIL: A retrospective cross-sectional study

BackgroundFew studies have quantitatively analyzed the imaging disparities between multiple sclerosis (MS) and cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL). We aimed to compare the imaging characteristics of MS and CADASIL in middle-aged patients. Materials and MethodsThis retrospective study used a single-center database and included patients aged 40–60 years with MS and CADASIL who underwent the designated imaging protocol including 3D T1-weighted imaging and fluid attenuated inversion recovery (FLAIR), diffusion tensor imaging and susceptibility-weighted imaging between January 2018 and March 2023. Patients with MRI-detected macrobleeds were excluded. ResultsA total of 27 patients with MS (mean age, 46.7 years ± 4.4, 8 men) and 30 patients with CADASIL (mean age, 51.6 years ± 5.8, 14 men) were included. No significant differences were observed in the Fazekas grades of white matter lesions (WMLs). Patients with CADASIL exhibited greater external capsule involvement (56.7% vs.18.5 %; p = 0.006), whereas the MS group had more lesions in the corpus callosum (81.5% vs. 53.3 %, p = 0.02) and brainstem (74.1% vs. 46.7 %, p = 0.04). The CADASIL group exhibited a higher incidence of microbleeds (12.07 vs. 0.11, p = 0.001). The WMLs in the MS group exhibited a lower T1 lesion/cerebrospinal fluid signal index (2.206 vs. 2.882, p < 0.001). A value of ≤2.57 demonstrated a sensitivity of 92.6 % and a specificity of 90.0 % in differentiating MS. Patients with MS had a thinner corpus callosum (7.18 mm vs 7.86 mm, p = 0.04), while patients with CADASIL showed significantly higher mean diffusivity (0.8776 × 10–3 vs. 0.7637 × 10–3 mm2/s, p = 0.03) and lower fractional anisotropy (0.7581 vs. 0.8389, p = 0.04) in the splenium of the corpus callosum. ConclusionMiddle-aged patients with MS and CADASIL showed comparable Fazekas grades for WMLs. However, lesion distribution, T1 signal characteristics, and splenic diffusivity changes can help differentiate between MS and CADASIL.

Cerebrovascular disease is associated with Alzheimer's plasma biomarker concentrations in adults with Down syndrome.

By age 40 years, over 90% of adults with Down syndrome have Alzheimer's disease pathology and most progress to dementia. Despite having few systemic vascular risk factors, individuals with Down syndrome have elevated cerebrovascular disease markers that track with the clinical progression of Alzheimer's disease, suggesting a role of cerebrovascular disease that is hypothesized to be mediated by inflammatory factors. This study examined the pathways through which small vessel cerebrovascular disease contributes to Alzheimer's disease-related pathophysiology and neurodegeneration in adults with Down syndrome. One hundred eighty-five participants from the Alzheimer's Biomarkers Consortium-Down Syndrome [mean (SD) age = 45.2 (9.3) years] with available MRI and plasma biomarker data were included in this study. White matter hyperintensity (WMH) volumes were derived from T2-weighted fluid-attenuated inversion recovery MRI scans, and plasma biomarker concentrations of amyloid beta 42/40, phosphorylated tau 217, astrocytosis (glial fibrillary acidic protein) and neurodegeneration (neurofilament light chain) were measured with ultrasensitive immunoassays. We examined the bivariate relationships of WMH, amyloid beta 42/40, phosphorylated tau 217 and glial fibrillary acidic protein with age-residualized neurofilament light chain across Alzheimer's disease diagnostic groups. A series of mediation and path analyses examined statistical pathways linking WMH and Alzheimer's disease pathophysiology to promote neurodegeneration in the total sample and groups stratified by clinical diagnosis. There was a direct and indirect bidirectional effect through the glial fibrillary acidic protein of WMH on phosphorylated tau 217 concentration, which was associated with neurofilament light chain concentration in the entire sample. Amongst cognitively stable participants, WMH was directly and indirectly, through glial fibrillary acidic protein, associated with phosphorylated tau 217 concentration, and in those with mild cognitive impairment, there was a direct effect of WMH on phosphorylated tau 217 and neurofilament light chain concentrations. There were no associations of WMH with biomarker concentrations among those diagnosed with dementia. The findings from this cross-sectional study suggest that among individuals with Down syndrome, cerebrovascular disease promotes neurodegeneration by increasing astrocytosis and tau pathophysiology in the presymptomatic phases of Alzheimer's disease, but future studies will need to confirm these associations with longitudinal data. This work joins an emerging literature that implicates cerebrovascular disease and its interface with neuroinflammation as a core pathological feature of Alzheimer's disease in adults with Down syndrome.

Watershed regions are more susceptible to tissue microstructural injury in multiple sclerosis.

Histopathologic studies report higher concentrations of multiple sclerosis white matter lesions in watershed areas of the brain, suggesting that areas with relatively lower oxygen levels may be more vulnerable to disease. However, it is unknown at what point in the disease course lesion predilection for watershed territories begins. Accordingly, we studied a cohort of people with newly diagnosed disease and asked whether (1) white matter lesions disproportionally localize to watershed-regions and (2) the degree of microstructural injury in watershed-lesions is more severe. Fifty-four participants, i.e. 38 newly diagnosed people with multiple sclerosis, clinically isolated syndrome or radiologically isolated syndrome, and 16 age- and sex-matched healthy controls underwent brain magnetic resonance imaging. T1-weighted and T2-weighted fluid-attenuated inversion recovery sequences, selective inversion recovery quantitative magnetisation transfer images, and the multi-compartment diffusion imaging with the spherical mean technique were acquired. We computed the macromolecular-to-free pool size ratio, and the apparent axonal volume fraction maps to indirectly estimate myelin and axonal integrity, respectively. We produced a flow territory atlas in each subject's native T2-weighted fluid-attenuated inversion recovery images using a T1-weighted magnetic resonance imaging template in the Montreal Neurological Institute 152 space. Lesion location relative to the watershed, non-watershed and mixed brain vascular territories was annotated. The same process was performed on the T2-weighted fluid-attenuated inversion recovery images of the healthy controls using 294 regions of interest. Generalized linear mixed models for continuous outcomes were used to assess differences in size, pool size ratio and axonal volume fraction between lesions/regions of interests (in healthy controls) situated in different vascular territories. In patients, we assessed 758 T2-lesions and 356 chronic black holes (cBHs). The watershed-territories had higher relative and absolute concentrations of T2-lesions (P≤0.041) and cBHs (P≤0.036) compared to either non-watershed- or mixed-zones. T2-lesions in watershed-areas also had lower pool size ratio relative to T2-lesions in either non-watershed- or mixed-zones (P = 0.039). These results retained significance in the sub-cohort of people without vascular comorbidities and when accounting for periventricular lesions. In healthy controls, axonal volume fraction was higher only in mixed-areas regions of interest compared to non-watershed-ones (P = 0.008). No differences in pool size ratio were seen. We provide in vivo evidence that there is an association between arterial vascularisation of the brain and multiple sclerosis-induced tissue injury as early as the time of disease diagnosis. Our findings underline the importance of oxygen delivery and healthy arterial vascularisation to prevent lesion formation and foster a better outcome in multiple sclerosis.

Prediction of early recurrence of adult-type diffuse gliomas following radiotherapy using multi-modal magnetic resonance images.

Adult-type diffuse gliomas are among the central nervous system's most aggressive malignant primary neoplasms. Despite advancements in systemic therapies and technological improvements in radiation oncology treatment delivery, the survival outcome for these patients remains poor. Fast and accurate assessment of tumor response to oncologic treatments is crucial, as it can enable the early detection of recurrent or refractory gliomas, thereby allowing timely intervention with life-prolonging salvage therapies. Radiomics is a developing field with great potential to improve medical image interpretation. This study aims to apply a radiomics-based predictive model for classifying response to radiotherapy within the first 3 months post-treatment. Ninety-five patients were selected from the Burdenko Glioblastoma Progression Dataset. Tumor regions were delineated in the axial plane on contrast-enhanced T1(CE T1W) and T2 fluid-attenuated inversion recovery (T2_FLAIR) magnetic resonance imaging (MRI). Hand-crafted radiomic (HCR) features, including first- and second-order features, were extracted using PyRadiomics (3.7.6) in Python (3.10). Then, recursive feature elimination with a random forest (RF) classifier was applied for feature dimensionality reduction. RF and support vector machine (SVM) classifiers were built to predict treatment outcomes using the selected features. Leave-one-out cross-validation was employed to tune hyperparameters and evaluate the models. For each segmented target, 186 HCR features were extracted from the MRI sequence. Using the top-ranked radiomic features from a combination of CE T1W and T2_FLAIR, an optimized classifier achieved the highest averaged area under the curve (AUC) of 0.829±0.075 using the RF classifier. The HCR features of CE T1W produced the worst outcomes among all models (0.603±0.024 and 0.615±0.075 for RF and SVM classifiers, respectively). We developed and evaluated a radiomics-based predictive model for early tumor response to radiotherapy, demonstrating excellent performance supported by high AUC values. This model, harnessing radiomic features from multi-modal MRI, showed superior predictive performance compared to single-modal MRI approaches. These results underscore the potential of radiomics in clinical decision support for this disease process.

Assessment of multi-modal magnetic resonance imaging for glioma based on a deep learning reconstruction approach with the denoising method.

Deep learning reconstruction (DLR) with denoising has been reported as potentially improving the image quality of magnetic resonance imaging (MRI). Multi-modal MRI is a critical non-invasive method for tumor detection, surgery planning, and prognosis assessment; however, the DLR on multi-modal glioma imaging has not been assessed. To assess multi-modal MRI for glioma based on the DLR method. We assessed multi-modal images of 107 glioma patients (49 preoperative and 58 postoperative). All the images were reconstructed with both DLR and conventional reconstruction methods, encompassing T1-weighted (T1W), contrast-enhanced T1W (CE-T1), T2-weighted (T2W), and T2 fluid-attenuated inversion recovery (T2-FLAIR). The image quality was evaluated using signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and edge sharpness. Visual assessment and diagnostic assessment were performed blindly by neuroradiologists. In contrast with conventionally reconstructed images, (residual) tumor SNR for all modalities and tumor to white/gray matter CNR from DLR images were higher in T1W, T2W, and T2-FLAIR sequences. The visual assessment of DLR images demonstrated the superior visualization of tumor in T2W, edema in T2-FLAIR, enhanced tumor and necrosis part in CE-T1, and fewer artifacts in all modalities. Improved diagnostic efficiency and confidence were observed for preoperative cases with DLR images. DLR of multi-modal MRI reconstruction prototype for glioma has demonstrated significant improvements in image quality. Moreover, it increased diagnostic efficiency and confidence of glioma.

Improved Dementia Prediction in Cerebral Small Vessel Disease Using Deep Learning-Derived Diffusion Scalar Maps From T1.

Cerebral small vessel disease is the most common pathology underlying vascular dementia. In small vessel disease, diffusion tensor imaging is more sensitive to white matter damage and better predicts dementia risk than conventional magnetic resonance imaging sequences, such as T1 and fluid attenuation inversion recovery, but diffusion tensor imaging takes longer to acquire and is not routinely available in clinical practice. As diffusion tensor imaging-derived scalar maps-fractional anisotropy (FA) and mean diffusivity (MD)-are frequently used in clinical settings, one solution is to synthesize FA/MD from T1 images. We developed a deep learning model to synthesize FA/MD from T1. The training data set consisted of 4998 participants with the highest white matter hyperintensity volumes in the UK Biobank. Four external validations data sets with small vessel disease were included: SCANS (St George's Cognition and Neuroimaging in Stroke; n=120), RUN DMC (Radboud University Nijmegen Diffusion Tensor and Magnetic Resonance Imaging Cohort; n=502), PRESERVE (Blood Pressure in Established Cerebral Small Vessel Disease; n=105), and NETWORKS (n=26), along with 1000 normal controls from the UK Biobank. The synthetic maps resembled ground-truth maps (structural similarity index >0.89 for MD maps and >0.80 for FA maps across all external validation data sets except for SCANS). The prediction accuracy of dementia using whole-brain median MD from the synthetic maps is comparable to the ground truth (SCANS ground-truth c-index, 0.822 and synthetic, 0.821; RUN DMC ground truth, 0.816 and synthetic, 0.812) and better than white matter hyperintensity volume (SCANS, 0.534; RUN DMC, 0.710). We have developed a fast and generalizable method to synthesize FA/MD maps from T1 to improve the prediction accuracy of dementia in small vessel disease when diffusion tensor imaging data have not been acquired.

NnU-Net-based Segmentation of Tumor Subcompartments in Pediatric Medulloblastoma Using Multiparametric MRI: A Multi-institutional Study.

Purpose To evaluate nnU-Net-based segmentation models for automated delineation of medulloblastoma tumors on multi-institutional MRI scans. Materials and Methods This retrospective study included 78 pediatric patients (52 male, 26 female), with ages ranging from 2 to 18 years, with medulloblastomas, from three different sites (28 from hospital A, 18 from hospital B, and 32 from hospital C), who had data available from three clinical MRI protocols (gadolinium-enhanced T1-weighted, T2-weighted, and fluid-attenuated inversion recovery). The scans were retrospectively collected from the year 2000 until May 2019. Reference standard annotations of the tumor habitat, including enhancing tumor, edema, and cystic core plus nonenhancing tumor subcompartments, were performed by two experienced neuroradiologists. Preprocessing included registration to age-appropriate atlases, skull stripping, bias correction, and intensity matching. The two models were trained as follows: (a) the transfer learning nnU-Net model was pretrained on an adult glioma cohort (n = 484) and fine-tuned on medulloblastoma studies using Models Genesis and (b) the direct deep learning nnU-Net model was trained directly on the medulloblastoma datasets, across fivefold cross-validation. Model robustness was evaluated on the three datasets when using different combinations of training and test sets, with data from two sites at a time used for training and data from the third site used for testing. Results Analysis on the three test sites yielded Dice scores of 0.81, 0.86, and 0.86 and 0.80, 0.86, and 0.85 for tumor habitat; 0.68, 0.84, and 0.77 and 0.67, 0.83, and 0.76 for enhancing tumor; 0.56, 0.71, and 0.69 and 0.56, 0.71, and 0.70 for edema; and 0.32, 0.48, and 0.43 and 0.29, 0.44, and 0.41 for cystic core plus nonenhancing tumor for the transfer learning and direct nnU-Net models, respectively. The models were largely robust to site-specific variations. Conclusion nnU-Net segmentation models hold promise for accurate, robust automated delineation of medulloblastoma tumor subcompartments, potentially leading to more effective radiation therapy planning in pediatric medulloblastoma. Keywords: Pediatrics, MR Imaging, Segmentation, Transfer Learning, Medulloblastoma, nnU-Net, MRI Supplemental material is available for this article. © RSNA, 2024 See also the commentary by Rudie and Correia de Verdier in this issue.

Differentiation between multifocal CNS lymphoma and glioblastoma based on MRI criteria

PurposeDifferentiating between glioblastoma (GB) with multiple foci (mGB) and multifocal central nervous system lymphoma (mCNSL) can be challenging because these cancers share several features at first appearance on magnetic resonance imaging (MRI). The aim of this study was to explore morphological differences in MRI findings for mGB versus mCNSL and to develop an interpretation algorithm with high diagnostic accuracy.MethodsIn this retrospective study, MRI characteristics were compared between 50 patients with mGB and 50 patients with mCNSL treated between 2015 and 2020. The following parameters were evaluated: size, morphology, lesion location and distribution, connections between the lesions on the fluid-attenuated inversion recovery sequence, patterns of contrast enhancement, and apparent diffusion coefficient (ADC) values within the tumor and the surrounding edema, as well as MR perfusion and susceptibility weighted imaging (SWI) whenever available.ResultsA total of 187 mCNSL lesions and 181 mGB lesions were analyzed. The mCNSL lesions demonstrated frequently a solid morphology compared to mGB lesions, which showed more often a cystic, mixed cystic/solid morphology and a cortical infiltration. The mean measured diameter was significantly smaller for mCNSL than mGB lesions (p < 0.001). Tumor ADC ratios were significantly smaller in mCNSL than in mGB (0.89 ± 0.36 vs. 1.05 ± 0.35, p < 0.001). The ADC ratio of perilesional edema was significantly higher (p < 0.001) in mCNSL than in mGB. In SWI / T2*-weighted imaging, tumor-associated susceptibility artifacts were more often found in mCNSL than in mGB (p < 0.001).ConclusionThe lesion size, ADC ratios of the lesions and the adjacent tissue as well as the vascularization of the lesions in the MR-perfusion were found to be significant distinctive patterns of mCNSL and mGB allowing a radiological differentiation of these two entities on initial MRI. A diagnostic algorithm based on these parameters merits a prospective validation.

Deep learning for efficient reconstruction of highly accelerated 3D FLAIR MRI in neurological deficits.

To compare compressed sensing (CS) and the Cascades of Independently Recurrent Inference Machines (CIRIM) with respect to image quality and reconstruction times when 12-fold accelerated scans of patients with neurological deficits are reconstructed. Twelve-fold accelerated 3D T2-FLAIR images were obtained from a cohort of 62 patients with neurological deficits on 3T MRI. Images were reconstructed offline via CS and the CIRIM. Image quality was assessed in a blinded and randomized manner by two experienced interventional neuroradiologists and one experienced pediatric neuroradiologist on imaging artifacts, perceived spatial resolution (sharpness), anatomic conspicuity, diagnostic confidence, and contrast. The methods were also compared in terms of self-referenced quality metrics, image resolution, patient groups and reconstruction time. In ten scans, the contrast ratio (CR) was determined between lesions and white matter. The effect of acceleration factor was assessed in a publicly available fully sampled dataset, since ground truth data are not available in prospectively accelerated clinical scans. Specifically, 451 FLAIR scans, including scans with white matter lesions, were adopted from the FastMRI database to evaluate structural similarity (SSIM) and the CR of lesions and white matter on ranging acceleration factors from four-fold up to 12-fold. Interventional neuroradiologists significantly preferred the CIRIM for imaging artifacts, anatomic conspicuity, and contrast. One rater significantly preferred the CIRIM in terms of sharpness and diagnostic confidence. The pediatric neuroradiologist preferred CS for imaging artifacts and sharpness. Compared to CS, the CIRIM reconstructions significantly improved in terms of imaging artifacts and anatomic conspicuity (p < 0.01) for higher resolution scans while yielding a 28% higher SNR (p = 0.001) and a 5.8% lower CR (p = 0.04). There were no differences between patient groups. Additionally, CIRIM was five times faster than CS was. An increasing acceleration factor did not lead to changes in CR (p = 0.92), but led to lower SSIM (p = 0.002). Patients with neurological deficits can undergo MRI at a range of moderate to high acceleration. DL reconstruction outperforms CS in terms of image resolution, efficient denoising with a modest reduction in contrast and reduced reconstruction times.

Improving the lesion appearance on FLAIR images synthetized from quantitative MRI: a fast, hybrid approach.

The image quality of synthetized FLAIR (fluid attenuated inversion recovery) images is generally inferior to its conventional counterpart, especially regarding the lesion contrast mismatch. This work aimed to improve the lesion appearance through a hybrid methodology. We combined a full brain 5-min MR-STAT acquisition followed by FLAIR synthetization step with an ultra-under sampled conventional FLAIR sequence and performed the retrospective and prospective analysis of the proposed method on the patient datasets and a healthy volunteer. All performance metrics of the proposed hybrid FLAIR images on patient datasets were significantly higher than those of the physics-based FLAIR images (p < 0.005), and comparable to those of conventional FLAIR images. The small difference between prospective and retrospective analysis on a healthy volunteer demonstrated the validity of the retrospective analysis of the hybrid method as presented for the patient datasets. The proposed hybrid FLAIR achieved an improved lesion appearance in the clinical cases with neurological diseases compared to the physics-based FLAIR images, Future prospective work on patient data will address the validation of the method from a diagnostic perspective by radiological inspection of the new images over a larger patient cohort.

Ultrafast Brain MRI at 3 T for MS: Evaluation of a 51-Second Deep Learning-Enhanced T2-EPI-FLAIR Sequence.

In neuroimaging, there is no equivalent alternative to magnetic resonance imaging (MRI). However, image acquisitions are generally time-consuming, which may limit utilization in some cases, e.g., in patients who cannot remain motionless for long or suffer from claustrophobia, or in the event of extensive waiting times. For multiple sclerosis (MS) patients, MRI plays a major role in drug therapy decision-making. The purpose of this study was to evaluate whether an ultrafast, T2-weighted (T2w), deep learning-enhanced (DL), echo-planar-imaging-based (EPI) fluid-attenuated inversion recovery (FLAIR) sequence (FLAIRUF) that has targeted neurological emergencies so far might even be an option to detect MS lesions of the brain compared to conventional FLAIR sequences. Therefore, 17 MS patients were enrolled prospectively in this exploratory study. Standard MRI protocols and ultrafast acquisitions were conducted at 3 tesla (T), including three-dimensional (3D)-FLAIR, turbo/fast spin-echo (TSE)-FLAIR, and FLAIRUF. Inflammatory lesions were grouped by size and location. Lesion conspicuity and image quality were rated on an ordinal five-point Likert scale, and lesion detection rates were calculated. Statistical analyses were performed to compare results. Altogether, 568 different lesions were found. Data indicated no significant differences in lesion detection (sensitivity and positive predictive value [PPV]) between FLAIRUF and axially reconstructed 3D-FLAIR (lesion size ≥3 mm × ≥2 mm) and no differences in sensitivity between FLAIRUF and TSE-FLAIR (lesion size ≥3 mm total). Lesion conspicuity in FLAIRUF was similar in all brain regions except for superior conspicuity in the occipital lobe and inferior conspicuity in the central brain regions. Further findings include location-dependent limitations of signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) as well as artifacts such as spatial distortions in FLAIRUF. In conclusion, FLAIRUF could potentially be an expedient alternative to conventional methods for brain imaging in MS patients since the acquisition can be performed in a fraction of time while maintaining good image quality.

Deep learning-based segmentation in MRI-(immuno)histological examination of myelin and axonal damage in normal-appearing white matter and white matter hyperintensities.

The major vascular cause of dementia is cerebral small vessel disease (SVD). Its diagnosis relies on imaging hallmarks, such as white matter hyperintensities (WMH). WMH present a heterogenous pathology, including myelin and axonal loss. Yet, these might be only the "tip of the iceberg." Imaging modalities imply that microstructural alterations underlie still normal-appearing white matter (NAWM), preceding the conversion to WMH. Unfortunately, direct pathological characterization of these microstructural alterations affecting myelinated axonal fibers in WMH, and especially NAWM, is still missing. Given that there are no treatments to significantly reduce WMH progression, it is important to extend our knowledge on pathological processes that might already be occurring within NAWM. Staining of myelin with Luxol Fast Blue, while valuable, fails to assess subtle alterations in white matter microstructure. Therefore, we aimed to quantify myelin surrounding axonal fibers and axonal- and microstructural damage in detail by combining (immuno)histochemistry with polarized light imaging (PLI). To study the extent (of early) microstructural damage from periventricular NAWM to the center of WMH, we refined current analysis techniques by using deep learning to define smaller segments of white matter, capturing increasing fluid-attenuated inversion recovery signal. Integration of (immuno)histochemistry and PLI with post-mortem imaging of the brains of individuals with hypertension and normotensive controls enables voxel-wise assessment of the pathology throughout periventricular WMH and NAWM. Myelin loss, axonal integrity, and white matter microstructural damage are not limited to WMH but already occur within NAWM. Notably, we found that axonal damage is higher in individuals with hypertension, particularly in NAWM. These findings highlight the added value of advanced segmentation techniques to visualize subtle changes occurring already in NAWM preceding WMH. By using quantitative MRI and advanced diffusion MRI, future studies may elucidate these very early mechanisms leading to neurodegeneration, which ultimately contribute to the conversion of NAWM to WMH.

Examining the relative contribution of slow-burning inflammation and chronic demyelination to axonal damage in chronic multiple sclerosis lesions

Background and ObjectivesSlow-burning inflammation at the edge, and chronic demyelination at the core, of established multiple sclerosis (MS) lesions are potential mediators of disease progression. However, their relative contribution to progressive axonal damage has not been explored. Therefore, in this study, we investigated the comparative contribution of slow-burning inflammation and chronic demyelination to axonal attrition within MS lesions by measuring progressive tissue rarefaction. In addition, we use the visual system as a model to investigate the effect of chronic demyelination on the acceleration of axonal death in a sub-group of patients with unilateral optic neuritis. MethodsPre- and post-gadolinium 3D-T1, 3D FLAIR, diffusion tensor images, Optical Coherence tomography and multifocal visual evoked potentials were acquired from 52 relapsing-remitting MS patients who completed at least 5 years follow-up.Lesion expansion was measured using custom software, and the rate of tissue rarefication inside lesion core was assessed by measuring increase of normalized mean diffusivity (nMD). Axonal loss was also examined in eyes with severe optic nerve demyelination. ResultsAmong the 361 lesions analyzed, 104 were expanding (a minimum of 4 % expansion per year) and 257 were stable. Expanding lesions showed a significantly higher rate of progressive tissue rarefication inside lesion (1.12 % per year) core compared to stable lesions (0.21 % per year, p = 0.01). The magnitude of nMD change was significantly correlated with the rate of lesion expansion (r = 0.4, p < 0.001). Analysis of retinal ganglion cells in eyes with severe optic nerve demyelination (Inter-eye latency delay of >10 ms) revealed a similar rate of axonal loss (0.19 %) to the degree of tissue rarefaction observed in stable lesions (0.21 %). DiscussionThe results of the study suggest that the slow-burning inflammation at the lesion's edge (as measured by lesion expansion), is likely to have a greater impact on tissue damage (as measured by nMD change), when compared to stable chronically demyelinated lesions. The similar modest degree of tissue damage was also observed in chronically demyelinated fibers of the optic nerve.

Contrast-Enhancing Lesion Segmentation in Multiple Sclerosis: A Deep Learning Approach Validated in a Multicentric Cohort.

The detection of contrast-enhancing lesions (CELs) is fundamental for the diagnosis and monitoring of patients with multiple sclerosis (MS). This task is time-consuming and suffers from high intra- and inter-rater variability in clinical practice. However, only a few studies proposed automatic approaches for CEL detection. This study aimed to develop a deep learning model that automatically detects and segments CELs in clinical Magnetic Resonance Imaging (MRI) scans. A 3D UNet-based network was trained with clinical MRI from the Swiss Multiple Sclerosis Cohort. The dataset comprised 372 scans from 280 MS patients: 162 showed at least one CEL, while 118 showed no CELs. The input dataset consisted of T1-weighted before and after gadolinium injection, and FLuid Attenuated Inversion Recovery images. The sampling strategy was based on a white matter lesion mask to confirm the existence of real contrast-enhancing lesions. To overcome the dataset imbalance, a weighted loss function was implemented. The Dice Score Coefficient and True Positive and False Positive Rates were 0.76, 0.93, and 0.02, respectively. Based on these results, the model developed in this study might well be considered for clinical decision support.