T2-FLAIR Images Research Articles

External Validation Of A Radiomics-Based Machine Learning Model For Distinguishing Radiation Necrosis From Progression Of Brain Metastases Treated With Stereotactic Radiosurgery

Radiation necrosis (RN) is common and potentially debilitating after stereotactic radiosurgery (SRS) for brain metastases (BM). The goal of this study is to validate a previously reported radiomics signature for distinguishing RN from true progression (TP) using an independent image set. Patients with BM were treated with Gammaknife SRS at Wake Forest University between 2004 and 2012 (WF dataset). Those who developed radiographic evidence of progression in the treated lesion and subsequently had pathologic confirmation were included. The lesions were identified on T1 post-contrast (T1c) and T2 fluid attenuated inversion recovery (T2 FLAIR) MRI. 51 T1c and 51 T2 FLAIR radiomics features were extracted using a software developed in house. An IsoSVM machine learning classifier model trained using our institutional data (JH dataset) was tested using WF dataset. Fifty-three lesions from 43 patients were included in this study, of which all had T1c and 33 had T2 FLAIR images available. Seven lesions were pure RN pathologically, while the TP group included 37 pure tumor and 9 mixed tumor and necrosis. Twenty-two patients (51%) underwent whole brain radiotherapy with a median dose of 35Gy. Median SRS marginal dose was 18Gy (range 10-22Gy). There was no statistically significant difference among baseline characteristics between RN and TP. Of the 6 T1c radiomic features from the JH dataset, 4 demonstrated the same trend in the WF dataset (neighborhood grey tone difference matrix (NGTDM) texture strength, NGTDM coarseness, grey level run length matrix (GLRLM) grey level nonuniformity, and GLRLM run percentage). NGTDM texture strength (mean RN 161.13 vs TP 70.86, p = 0.06), and NGTDM coarseness (mean RN 0.03 vs TP 0.02, p = 0.07) approached significance. Of the 4 FLAIR features, 3 showed similar trend (kurtosis, minimum, NGTDM texture strength). The original IsoSVM model trained with all 6 T1c and 4 FLAIR features showed sensitivity of 65%, specificity of 87%, and AUC of 0.81. Due to the limited number of cases with FLAIR available, we re-trained the model with JH dataset using only the 4 T1c features showing the same trend between the two datasets, resulting in sensitivity and specificity of 80% and 73%, with AUC of 0.8. This model when tested on the WF dataset showed sensitivity and specificity of 71% and 52%, with AUC of 0.62. When the model was re-trained with the 4 T1c and 3 FLAIR features trending similarly between the two datasets, the sensitivity and specificity were 50% and 100%, and the AUC improved to 0.84. Radiomics profiling of post-SRS T1c and T2 FLAIR MRI showed promising trends between RN and TP in 4 T1c features across two independent datasets. Radiomics-based IsoSVM model can distinguish RN vs TP in individual patients. Statistical significance and model performance were inferior in the validation (WF) dataset due to smaller sample size and class imbalance with fewer RN cases. Further studies with larger datasets are needed to optimize and validate the model.

MR-Linac Radiotherapy in Year One: Experience's in Imaging, Patient Treatment and Data Collection

To present our institutions initial clinical experience with a high field MR-Linac with regards to patients treated, data collected, functional imaging performed and general performance of the machine. From Aug 15, 2019 to Feb 7, 2020, thirty-three patients were treated on our 1.5 T MR-Linac with all patients enrolled in the MOMENTUM study (ClinicalTrials.gov, identifier: NCT04075305). Two types of workflows were used: one accounting for daily patient shifts only (adapt to position or ATP) and the other involving online plan adaptation through re-contouring and re-optimization based on daily MRI images (adapt to shape or ATS). Of the thirty-three patients, thirty were CNS brain patients (15 GBM and 15 other) treated using ATP and three were prostate cancer patients treated using ATS and SBRT to a dose of 40 Gy in 5 fractions. Imaging involved T1 (brain ATP) or T2 (prostate ATS) weighted scans captured online for registration/planning. Functional imaging sequences for brain ATP (DWI, MT, BOLD, CEST) were taken before, during and post irradiation in addition to T2-FLAIR. DWI was done for the prostate cases before, during and post irradiation. Timing was recorded for the various stages of the workflows. Machine performance was characterized in terms of unscheduled downtime events. A total of 540 fractions were delivered on the MR-Linac over its initial 25 weeks of use. The average session time (n = 200 fractions) for the brain ATP workflow (excluding post imaging) was 26.9 minutes (3.7 min setup, 5.5 min pre-tx imaging, 5.1 min image registration, 4.5 min plan optimization, 2.0 min physics QA, and 6.1 min beam on). The minimum session time was 24 minutes for the brain ATP. The average session time for the prostate ATS workflow (n = 10 fractions) was 47.7 minutes (6.6 min setup, 3.2 min pre-tx imaging, 3.5 min image registration, 15.1 min contouring, 5.6 min plan optimization, 1.5 min physics QA, 2.9 min verification imaging and 9.3 min beam on). Additional post beam research scans took, on average, 12.5 minutes for the prostate cases. The following multi-parametric imaging maps were generated for the brain patients: T1 and T2, ADC, T2-FLAIR, CEST asymmetry and Amide MTR. The brain T2-FLAIR signal, in some cases, reduced over the treatment course. Changes in ADC maps were observed for both brain and prostate over the treatment. In some instances, brain re-planning was done offline based on daily online T2-FLAIR imaging. In terms of downtime, patients were transferred to a standard Linac for 7.1 % of fractions due to either a magnetron change, modulator tank replacement, or a cooling issue. An MR-Linac program has been implemented at our institution involving a multi-disciplinary group with both machine data and patient data being captured. The current status, clinical and technical considerations, and most recent image-based findings will be presented.

Postcontrast Fluid-Attenuated Inversion Recovery (FLAIR) Sequence MR Imaging in Detecting Intracranial Pathology.

Background Imaging sequences for detection of meningeal and parenchymal lesions are critical in intracranial pathology. Our study analysed FLAIR MRI sequence for evaluating postcontrast enhancement. Objectives FLAIR imaging sequences have been used in evaluation of enhancement in the brain. We conducted a study of FLAIR imaging sequences to better delineate postcontrast enhancement. Materials and Methods In this prospective hospital-based observational study, postcontrast T1 MTC and delayed postcontrast T2 FLAIR and T1 FLAIR images of 66 patients with intracranial pathology were assessed by experienced radiologists from November 2017 to November 2019. Results 28 cases of meningeal enhancement were identified in delayed postcontrast T2 FLAIR images. Low-grade gliomas included in the study showed postcontrast enhancement on postcontrast T1 MTC images. Multiple sclerosis lesions were better seen on postcontrast T1 FLAIR. In extraaxial lesions of 11 cases of meningioma, brighter enhancement was seen on delayed postcontrast T2 FLAIR images. Conclusion We found that delayed postcontrast T2 FLAIR was better in detection of meningeal enhancement in infectious meningitis and in meningitis carcinomatosis than T1 MTC images. In delayed postcontrast T2 FLAIR images, intra-axial parenchyma lesions appeared more conspicuous or similar to T1 MTC images. Delayed postcontrast T1 FLAIR images provided better anatomic delineation of intra-axial lesions.

Optimization of Contrast Agent Dosage on Contrast-Enhanced T2 Fluid-Attenuated Inversion Recovery: An In Vitro and In Vivo Study.

This study aimed to ascertain the minimum gadolinium dosage on contrast-enhanced (CE) T2 fluid-attenuated inversion recovery (FLAIR) at appropriate imaging time. Different dosages of gadodiamide were imaged with a 3.0-T magnetic resonance scanner for T2-FLAIR and T1WI. Twenty glioma-induced rat models were randomly assigned into 4 groups (1/2, 1/4, 1/6, 1/8 of routine dosage) and imaged for T2-FLAIR and T1WI preinjection and postinjection of gadodiamide. Contrast-enhanced T2-FLAIR was acquired for 8 repetitions postinjection. Enhancement effects were assessed by calculating contrast-noise ratio and contrast ratio using Kruskal-Wallis and Mann-Whitney rank sum test. The in vitro experiment showed that gadodiamide at 1/4 of the T1WI dosage presented the best contrast on CE-T2-FLAIR. For in vivo study, the best enhancement effect on CE-T2-FLAIR was achieved at 1/2 of the routine dosage at 8 to 12 minutes of delayed scanning. Compared with CE-T1WI at routine dosage, CE-T2-FLAIR at 1/2 gadodiamide dosage presented similar enhancement effects with no statistical difference (P = 0.244 and 0.090 for contrast-noise ratio and contrast ratio, respectively). Contrast-enhanced T2-FLAIR imaging with half of T1WI routine gadodiamide dosage can produce similar enhancement effects to CE-T1WI when characterizing brain gliomas. The cut-down of contrast agent dosage may help reduce gadolinium accumulation in certain tissues.

Olfactory Bulb Signal Abnormality in Patients with COVID-19 Who Present with Neurologic Symptoms.

Unique among the acute neurologic manifestations of Severe Acute Respiratory Syndrome coronavirus 2, the virus responsible for the coronavirus disease 2019 (COVID-19) pandemic, is chemosensory dysfunction (anosmia or dysgeusia), which can be seen in patients who are otherwise oligosymptomatic or even asymptomatic. The purpose of this study was to determine if there is imaging evidence of olfactory apparatus pathology in patients with COVID-19 and neurologic symptoms. A retrospective case-control study compared the olfactory bulb and olfactory tract signal intensity on thin-section T2WI and postcontrast 3D T2 FLAIR images in patients with COVID-19 and neurologic symptoms, and age-matched controls imaged for olfactory dysfunction. There was a significant difference in normalized olfactory bulb T2 FLAIR signal intensity between the patients with COVID-19 and the controls with anosmia (P = .003). Four of 12 patients with COVID-19 demonstrated intraneural T2 signal hyperintensity on postcontrast 3D T2 FLAIR compared with none of the 12 patients among the controls with anosmia (P = .028). Olfactory bulb 3D T2 FLAIR signal intensity was greater in the patients with COVID-19 and neurologic symptoms compared with an age-matched control group with olfactory dysfunction, and this was qualitatively apparent in 4 of 12 patients with COVID-19. Analysis of these preliminary finding suggests that olfactory apparatus vulnerability to COVID-19 might be supported on conventional neuroimaging and may serve as a noninvasive biomarker of infection.

Abstract 5487: A biologically-motivated mathematical model for forecasting chemoradiationresponse in high grade gliomas: Initial results

Abstract Introduction: Using patient-specific imaging data, we aimed to calibrate biologically-based mathematical models of tumor growth and treatment response to forecast the progression of high-grade glioma on an individual patient basis. Methods: Four patients with partially resected or biopsy only, histologically confirmed high grade gliomas were imaged with T2-FLAIR, diffusion weighted (DW-) MRI, and post-contrast T1-weighted MRI at baseline (prior to chemoradiation), and at 1-, 3-, and 5-months following chemoradiation. Patients received standard of care radiotherapy (RT) with concurrent and adjuvant temozolomide. The gross tumor volume (GTV) defined as enhancing tumor and clinical tumor volume (CTV) defined as non-enhancing, T2-hyperintense region were expertly segmented from the post-contrast T1-weighted and T2-FLAIR images, respectively. The apparent diffusion coefficient (ADC) estimated from DW-MRI was used to assign tumor cell number within the GTV. Within the CTV, we interpolated the tumor cell number from 2% of the carrying capacity (at the periphery of the CTV) to the number of cells calculated at the periphery of the GTV. Using this data, a 3D mechanically-coupled, reaction-diffusion model was developed to describe the spatiotemporal evolution of tumor cell number due to proliferation, diffusion, and death due to chemoradiation on an individual patient basis. For each patient, model parameters were calibrated using their cellularity measurements from the entire time course (to evaluate how well the model describes the data) and separately from a subset of the time course (to evaluate how well the model predicts future data). For the prediction scenario, we then used the calibrated parameters in a forward evaluation of the model to predict future tumor growth at the final imaging visit. The model estimates of tumor cellularity where compared directly to the measured tumor distributions at the volume and voxel levels. Results: For the model fit scenario, a strong level of voxel-wise correlation (Pearson correlation coefficients, PCC, ranging from 0.68 to 0.94) and agreement (concordance correlation coefficients, CCC, 0.76 to 0.99) were observed between the calibrated model and the measured cellularity for each patient. Additionally, Dice coefficients (measuring the degree of overlap between calibrated and observed values) ranged from 0.76 to 0.92 at all time points. For the prediction scenario, a strong of voxel-wise correlation (PCC ranging from 0.61 to 0.90) and agreement (CCC ranging from 0.72 to 0.92) were observed between the predicted and estimated cellularity for each patient. A strong level of spatial agreement was also observed with Dice values ranging from 0.65 to 0.84 at all time points. Conclusions: This preliminary study demonstrates that patient-specific imaging data can be used to calibrate a mathematical model that can accurately predict the spatio-temporal distribution of enhancing and non-enhancing glioma. Funding: NCI U01 CA174706, CPRIT RR160005. AAPM Grant 2018 Citation Format: David Andrew Hormuth, Karine Al Feghali, Andrew M. Elliot, Caroline Chung, Thomas E. Yankeelov. A biologically-motivated mathematical model for forecasting chemoradiationresponse in high grade gliomas: Initial results [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 5487.

U-net combined with CRF and anatomical based spatial features to segment white matter hyperintensities.

White matter hyperintensities (WMH) are important biomarkers for cerebral small vessel disease and closely associated with other neurodegenerative process. In this paper, we proposed a fully automatic WMH segmentation method based on U-net architecture. CRF were combined with U-net to refine segmentation results. We used a new anatomical based spatial feature produced by brain tissue segmentation based on T1 image, along with intensities of T1 and T2-FLAIR images to train our neural network. We compared 8 forms of automated WMH segmentation methods, range from traditional statistical learnng methods to deep learning based methods, with different architecture and used different features. Results showed our proposed method achieved best performance in terms of most metrics, and the inclusion of anatomical based spatial features strongly increase the segmentation performance.

BOLD signal within and around white matter lesions distinguishes multiple sclerosis and non-specific white matter disease: a three-dimensional approach.

Multiple sclerosis (MS) diagnostic criteria are based upon clinical presentation and presence of white matter hyperintensities on two-dimensional magnetic resonance imaging (MRI) views. Such criteria, however, are prone to false-positive interpretations due to the presence of similar MRI findings in non-specific white matter disease (NSWMD) states such as migraine and microvascular disease. The coexistence of age-related changes has also been recognized in MS patients, and this comorbidity further poses a diagnostic challenge. In this study, we investigated the physiologic profiles within and around MS and NSWMD lesions and their ability to distinguish the two disease states. MS and NSWMD lesions were identified using three-dimensional (3D) T2-FLAIR images and segmented using geodesic active contouring. A dual-echo functional MRI sequence permitted near-simultaneous measurement of blood-oxygen-level-dependent signal (BOLD) and cerebral blood flow (CBF). BOLD and CBF were calculated within lesions and in 3D concentric layers surrounding each lesion. BOLD slope, an indicator of lesion metabolic capacity, was calculated as the change in BOLD from a lesion through its surrounding perimeters. We observed sequential BOLD signal reductions from the lesion towards the perimeters for MS, while no such decreases were observed for NSWMD lesions. BOLD slope was significantly lower in MS compared to NSWM lesions, suggesting decreased metabolic activity in MS lesions. Furthermore, BOLD signal within and around lesions significantly distinguished MS and NSWMD lesions. These results suggest that this technique shows promise for clinical utility in distinguishing NSWMD or MS disease states and identifying NSWMD lesions occurring in MS patients.

Primary Categorizing and Masking Cerebral Small Vessel Disease Based on "Deep Learning System".

ObjectiveTo supply the attending doctor’s diagnosis of the persisting of cerebral small vessel disease and speed up their work effectively, we developed a “deep learning system (DLS)” for cerebral small vessel disease predication. The reliability and the disease area segmentation accuracy, of the proposed DLS, was also investigated.MethodsA deep learning model based on the convolutional neural network was designed and trained on 1,010 DWI b1000 images from 1010 patients diagnosed with segmentation of subcortical infarction, 359 T2∗ images from 359 patients diagnosed with segmentation of cerebral microbleed, as well as 824 T1-weighted and T2-FLAIR images from 824 patients diagnosed with segmentation of lacune and WMH. Dicw accuracy, recall, and f1-score were calculated to evaluate the proposed deep learning model. Finally, we also compared the DLS prediction capability with that of 6 doctors with 3 to 18 years’ clinical experience (8 ± 6 years).ResultsThe results support that an appropriately trained DLS can achieve a high-level dice accuracy, 0.598 in the training section over all these four classifications on 30 patients (0.576 for young neuroradiologists), validation accuracy is 0.496 in lacune, 0.666 in WMH, 0.728 in subcortical infarction, and 0.503 in cerebral microbleeds. It is comparable to attending doctor with a few years of experience, regardless of whether the emphasis is placed on the segmentation or detection of lesions with less time-spending compared with manual analysis, about 4.4 s/case, which is dramatically less than doctors about 634 s/case.ConclusionThe results of our comparison lend support to the case that an appropriately trained DLS can be trusted to the same extent as one would trust an attending doctor with a few years of experience, regardless of whether the emphasis is placed on the segmentation or detection of lesions.

Resection of a Posterior Fossa Endodermal Cyst With Exoscopic Assistance: 2-Dimensional Operative Video.

Neurenteric cysts are rare benign congenital tumors of endodermal origin that most commonly occur in the cervical and upper thoracic spine, with only about 10% to 18% of the reported cases occurring intracranially.1 A definitive preoperative diagnosis is complicated by the variable appearance of neurenteric cysts on magnetic resonance (MR) imaging.2 The recommended treatment of neurenteric cysts is complete surgical resection when possible.3,4 We present a case of a posterior fossa neurenteric cyst. A 33-yr-old man without medical history presented with left-sided headache and mild left-sided facial numbness and weakness. Admission MR imaging revealed a nonenhancing mass, which was hyperintense on T1-weighted MR images, compressing the brainstem anteriorly. The lesion was isointense on T2 FLAIR images and hypointense on diffusion-weighted imaging, initially read as possible epidermoid cyst. The patient underwent a left-sided retrosigmoid craniotomy via far lateral transcondylar approach. The tumor was adjacent to both vertebral arteries, the left PICA, and cranial nerves (CN) VII-XII with superior extension to CN V. The cyst was encased in a thin capsule, and its contents were yellowish in color and ranged from thick liquid to colloidal and caseous consistency. The cyst also contained heavily calcified portions, which were excised using sharp dissection. Images of the cyst wall show that it is focally lined with ciliated columnar epithelium with intracellular mucin confirming an endodermal or neurenteric cyst. After the operation, the patient's symptoms resolved, and he was discharged on postoperative day 4. Postoperative MR images confirmed gross total resection. The patient consented to video production.

Value of Magnetic Resonance Spectroscopy MRS in Prolonged Febrile Convulsion in Children

Abstract Background: Febrile Convulsion (FC) or Febrile Seizure (FS) is the most common type of seizures that occur in children between the ages of 6 months to 5 years with body temperature over 38oC and without any infection such as meningitis and encephalitis in the central nervous system. The incidence of FC is 5 cases per 1000 children annually. Aim of Study: To determine metabolic changes that occur in temporal lobe and hippocampus after prolonged febrile seizures, by using Magnetic Resonance Spectroscopy (MRS). Patients and Methods: Thirty patients with prolonged febrile seizures more than 15 minutes were included. Control group including thirty child age and gender healthy matched. EEG was done within 72 hour of febrile seizure to all case and control group in the study using Pediatric Neurology Department EEG machine. Also, MRS (Magnetic Resonance Spectroscopy) was done within 48h of febrile seizure to the cases and was done to the control group, by using achieva 1.5t Philips. The spectra were all acquired in conjunction with an MRS study of the brain that included coronal T2, axial & coronal FLAIR images through the temporal lobes, and axial T1 and T2 SE images through the entire brain. To obtain MRS technique, scout imaging of the brain in coronal and sagittal orientations was performed with T1 WI. The MRS was per-formed using multi-voxel technique (PRESS-CSI sequence with TE/TR=135ms/1690ms, 12 averages, FOV 120 X 120mm2). The region of interest was placed through whole temporal lobi with the voxel layer position adjusted based on the localization of hippocampi in order to examine the whole hippocampi at long distance (voxel size set to 10 X 10 X 15mm3). The resonances of major metabolites detected were as follows: The N Acetyl Aspartate (NAA) peak at 2.02ppm, the creatine (Cr) and phosphocreatine peak at 3.02ppm, and the choline (Cho) peak at 3.20ppm. Results: We found that there was high significant difference between case and control groups regarding MRS. As in the case group there was reduction of NAA and slightly increase in Cho & Cr, which lead to changes in metabolic ratios (NAA/Cr, Cho/NAA & Cho/Cr). While in the control group was normal.The optimal cut off value of NAA/Cr was less than 2.61, The sensitivity and specificity of NAA/Cr was 96.7% and 93.3% respectively and accuracy was 95%. While, the optimal cut off value of Cho/NAA more than 0.785, the sensitivity and specificity of Cho/NAA was 96.7% and 80% respectively and accuracy was 88.3%. While, the optimal cut off value of Cho/Cr less than 2.115, The sensitivity and specificity of Cho/Cr was 93.3% and 93.3% respectively and accuracy was 93.3%. Conclusion: MR spectroscopy is a very sensitive guiding tool in predicting Temporal Lobe Epilepsy (TLE) and the side of involvement in patients with TLE even in patients with MR negative studies. Also, it helps in detecting abnormal spectra of various brain metabolites at temporal lobe and hippocampus. Further more, it can detect hippocampal injury earlier and with more accurate results than EEG.

Identification of expanded repeats in NOTCH2NLC in neurodegenerative dementias

Recently, the (GGC)n repeat expansion in the NOTCH2NLC gene has been identified to be associated with neuronal intranuclear inclusion disease (NIID). Given the clinical overlap of dementia-dominant NIID with neurodegenerative dementia, we therefore hypothesized that the NOTCH2NLC repeat expansion might also contribute to these diseases. In the present study, repeat primed polymerase chain reaction (RP-PCR) and GC-rich PCR were conducted to detect the repeats of NOTCH2NLC in a cohort of 1004 patients with neurodegenerative dementias from mainland China. As a result, 4 sporadic patients were found to carry the NOTCH2NLC repeats expansion, totally accounting for 0.4% of all dementia individuals, and the accurate repeated sizes were 110, 133,120 and 76 respectively. Of 4 mutation carriers, three and one were clinically diagnosed Alzheimer's disease (AD) and frontotemporal dementia (FTD) respectively. In addition, 3 out of them revealed leukoencephalopathy in T2-Flair imaging. This study revealed that although rare, the NOTCH2NLC repeat expansions may be associated with AD or FTD-like phenotype as well as leukoencephalopathy.

Perfusion magnetic resonance imaging in contouring of glioblastoma patients: Preliminary experience from a single institution.

T1-contrast and T2-flair images of magnetic resonance imaging (MRI) are commonly fused with computed tomography (CT) and used for delineation of postoperative residual tumor and bed after surgery in patients with glioblastoma multiforme (GBM). Our prospective study was aimed to see the feasibility of incorporating perfusion MRI in delineation of brain tumor for radiotherapy planning and its implication on treatment volumes. Twenty-four patients with histopathologically proven GBM were included in the study. All patients underwent radiotherapy planning with a contrast CT scan. In addition to radiotherapy (RT) planning protocol, T1-perfusion MRI was also done in all patients in the same sitting. Perfusion imaging was processed on the in-house-developed JAVA-based software. The images of CT and MRI were sent to the iPlan planning system (Brainlab AG, GmbH) using a Digital Imaging and Communications in Medicine - Radiation Therapy (DICOM-RT) protocol. A structure of gross tumor volume (GTV)-perfusion (GTV-P) was delineated based only on the MRI perfusion images. Subsequently, GTV-P and GTV were fused together to make GTV-summated (GTV-S). Using existing guidelines, GTV-S was expanded to form clinical target volume-summated (CTV-S) and planning target volume-summated (PTV-S). The increment in each of the summated volumes as compared to baseline volume was noted. The common overlap volume (GTVO) between GTV and GTV-P was calculated using intersection theory (GTV n GTV-P = GTVO [Overlap]). Mean ± standard deviation (cc) for GTV, GTV-P, and GTVO was 46.3 ± 33.4 cc (range: 5.2 cc-108.0 cc), 26.0 ± 26.2 (range: 6.6 cc-10.3.0 cc), and 17.5 ± 22.3 cc (range: 10.0 cc-92 cc), respectively. Median volume (cc) for GTV, GTV-P, and GTVO was 40.8 cc, 17.2 cc, and 8.0 cc, respectively. Mean absolute and relative increments from GTV to that of GTV-S were 8.5 ± 8.2 cc and 27.2 ± 30.9%, respectively. Average CTV volume (cc) was 230.4 ± 115.3 (range: 80.8 cc-442.0 cc). Mean and median CTV-S volumes were 262.0 ± 126.3 cc (range: 80.8 cc-483.0 cc) and 221.0 cc, respectively. The increment in the mean CTV volume (with respect to CTV created from GTV-S) was 15.2 ± 15.9%. Mean and median PTV volumes created on the summated CTV were 287.1 ± 134.0 cc (range: 118.9 cc-576.0 cc) and 258.0 cc, respectively. Absolute and relative increments in PTV volume, while incorporating the perfusion volume, were 31.3 ± 28.9 cc and 12.5 ± 13.3%, respectively. Out of the total of 24 patients, perfusion scanning did not do any increment in GTV in five patients. Our study is the first to present the feasibility and the outcome of contouring on perfusion imaging and its overlay on regular MRI images. The implications of this on long-term outcome and control rates of glioblastoma patients need to be seen in future studies.

Usefulness of Contrast-Enhanced 3D-FLAIR MR Imaging for Differentiating Rathke Cleft Cyst from Cystic Craniopharyngioma.

Because it can be difficult to discriminate between a Rathke cleft cyst and cystic craniopharyngioma by conventional MR imaging alone, we investigated whether contrast-enhanced 3D T2-FLAIR MR imaging at 3T helps to distinguish a Rathke cleft cyst from a cystic craniopharyngioma. We evaluated pre- and postcontrast T1-weighted and 3D T2-FLAIR images of 17 patients with pathologically confirmed Rathke cleft cyst (n = 10) or cystic craniopharyngioma (n = 7). All underwent 3T MR imaging studies before surgery. Two neuroradiologists independently recorded the enhancement grade of the lesion wall as grade 2 (most of the wall enhanced), grade 1 (some of the wall enhanced), and grade 0 (none of the wall enhanced). One neuroradiologist performed a blinded reading study of conventional MR images with/without 3D T2-FLAIR images. Interobserver agreement was determined by calculating the κ coefficient. Statistical analyses, including receiver operating characteristic curve analysis were performed. Interobserver agreement for postcontrast T1WI and 3D T2-FLAIR images was excellent (κ = 0.824 and κ = 0.867, respectively). Although the difference in the mean enhancement grade of Rathke cleft cysts and cystic craniopharyngiomas was not significant on postcontrast T1WIs, it was significant on postcontrast 3D T2-FLAIR images (P = .0011). The area under the receiver operating characteristic curve of the conventional MR alone and conventional MR with 3D T2-FLAIR readings was 0.879 and 1.0, respectively, though there was no significant difference in the area under the curve between the 2 readings. Contrast-enhanced 3D T2-FLAIR imaging at 3T helps to distinguish a Rathke cleft cyst from cystic craniopharyngioma.

NIMG-45. 3D SINGLE-SCAN FOR MULTI-CONTRAST-WEIGHTED MRI FOR LESION DETECTION IN PATIENTS WITH GLIOMAS

Abstract INTRODUCTION Multi-contrast-weighted (MCW) MRI allows derivation of T1-weighted, T2-weighted, and T2-FLAIR images and their parametric maps. Under current practice, these high-resolution 3D MR images are acquired separately with a total scan time is approximately 30 minutes. The goal of this study was to develop a novel sequence and processing strategy for MCW MRI and lesion detection of gliomas with 1mm isotropic resolution in 6 minutes, which can significantly improve clinical imaging workflow. The proposed whole brain, single-scan method enables the generation of segmented tumor regions that correspond the hyperintense lesion on T2-weighted or T2-FLAIR images (T2L). METHODS Our strategy involves using a highly accelerated 3D MRI sequence to achieve whole brain, MCW images that can be jointly segmented with modified k-means clustering. Six patients with enhancing gliomas were scanned with this sequence before the injection of gadolinium, 4 of which also received a repeat scan post-gadolinium. Our automatic segmentation of the T2L and manual definition of the contrast-enhancing lesion (CEL) was compared against manually segmented lesions drawn from conventional anatomical imaging. RESULTS Whole-brain MCW MRI was achieved with 1mm isotropic resolution in 6 mins. Grey-matter, white-matter, CSF, and the T2L could be automatically segmented with our clustering method. The resulting automatically-segmented T2L volume was not significantly different between our new sequence and clinically defined regions (average Dice Coefficient=0.81(+/-0.05); p=0.95). CONCLUSIONS Our novel, 6-minute scan and processing strategy was able to generate MCW anatomical images and automatically segment grey-matter, white-matter, and T2L volumes that were similar to manual segmentation of conventional anatomical imaging. We are currently extending the automatic segmentation to the CEL. This approach has the potential to improve clinical workflow by allowing time for therapy-specific advanced imaging and providing a noninvasive tool for monitoring tumor evolution. Validation is currently being performed in a larger cohort.

NIMG-55. TO ASSESS THE UTILITY OF ADVANCED QUANTITATIVE DIFFUSION MRI DERIVED FROM MULTI B VALUE ACQUISITIONS IN THE ASSESSMENT OF TREATMENT RESPONSE, USING A SPATIALLY-INDEPENDENT APPROACH

Abstract PURPOSE To assess the utility of advanced diffusion MRI derived from multi b value acquisitions in the assessment of treatment response, using a spatially-independent approach. METHOD AND MATERIALS 13 GBM patients were enrolled into our multicentre study. All patients completed RT with TMZ. Imaging was performed pre-RT and mid RT. The MRI protocol included a ‘low’ b value acquisition (b= 0s/mm, 50s/mm, 150s/mm, 200s/mm, 500s/mm, 1000s/mm) from which monoexponential diffusion indices ADC and biexponential indices, IVIM parameters D*, D and f were calculated. A ‘high b value’ acquisition (b=0 s/mm, 500s/mm, 1000s/mm, 1500s/mm, 2000s/mm, 2500s/mm, 3000s/mm, 3500s/mm, 4000s/mm) was acquired to allow stretched exponential diffusion indices, DDC and alpha to be derived. FLAIR sequences were used to define ROI and clinical assessment of mid-treatment and end-treatment response using RANO criteria. Histograms were generated from voxels located within manually segmented ROIs defined by increased signal on T2 FLAIR images. Changes in histogram percentile profiles were evaluated across the two timepoints and compared with RANO assessment at the mid treatment and end treatment timepoints. RESULTS Following completion of treatment, 5 patients had PD, 4 SD and 4 CR. Patients with PD showed a histogram shift to the left across all diffusion models, in keeping with increasing diffusion restriction and implying increased cellularity. Patients with SD or CR showed little or no shift in the histogram. DDC and f are the most predictive of progression against RANO assessment, and appear superior to routine ADC. Reduction in 75th centile (f) and 95th centile (DDC) are the most sensitive histogram metrics for predicting early progressive disease. CONCLUSION Results suggest association between early changes in specific diffusion components and subsequent treatment response. Spatially-independent diffusion parameter comparisons provide unbiased sampling of tumour heterogeneity and abrogate the confound of voxel-to-voxel misregistration due to tumour growth/shrinkage.

Rare manifestations of lyme neuroborreliosis

Lyme disease is an endemic infection in many European countries and Russia, including North-West region. Neurological manifestations are reported in about 10%-25% of patients. Lyme neuroborreliosis (LNB) can includes polyradiculoneuritis, meningoradiculitis, meningitis and encephalitis in adults. Two cases of LNB are presented. Patient one: a 46-yrs old man was admitted with a history of dizziness, fever and walking disorder after four months of fishing in endemic ( tick-borne infections) area. On examination patient showed dysarthria, ataxia and hemianopsia. CSF showed mild pleocytosis and elevated total protein. Serological methods (including Western blot test) were positive for LNB. Brain MRI showed subacute ischemic lesions on T2-weighted and T2-FLAIR images on vertebrobasilar territory. Patient two was a 52-yrs old man, hospitalized with general stiffness after two months after tick bite without erythema migrans and fever. Neurological examination revealed a dysarthria, bradykinesia and right sided rigidity. Serological studies were positive for LNB. The CSF total protein content was 0,89 g/l. MRI images showed multifocal lesions of white matter (demyelination, typically seen in inflammatory). Both patients were diagnosed and treated successfully for LNB: they received a 21-day course of 2 g per day ceftriaxone, then benzathine benzylpenicillin every 14 days for 3 months and every 28 days for another 3 months. Corticosteroids and platelet aggregation inhibitors in case one and dopaminergic medications (case two) were also prescribed. A year after treatment patients were free of neurological symptoms. The cases suggest that LNB can manifest rare clinical forms as cerebral vasculitis and parkinsonism.

Plasma Lipid Profiling Identifies Biomarkers of Cerebral Microvascular Disease.

Brain-specific sphingolipids (SLs) may serve as effective biomarkers of white matter hyperintensities (WMH). Here, we investigate the efficacy of SLs as a novel fluid-based biomarker to identify WMH reflective of chronic ischemia. Patients presenting to our stroke center for evaluation of acute neurological deficits were enrolled in the Advanced Serum Profiling in Recent Stroke (ASPIRE) study. From this cohort of 202 individuals, 58 patients who underwent an MRI and did not have a clinical stroke event were included in this study. Plasma samples were collected at the time of MRI, and targeted SL profiling was performed by HPLC/tandem mass spectrometry. T2 FLAIR imaging was evaluated for WMH and scored according to the Fazekas scoring (FS) method and manually quantified. Twenty two SLs were definitively identified, consisting of ceramide (Cer) and sphingomyelin (SM) species. Of these, two sphingolipids, SM 38:1 and Cer 34:1, significantly correlated with high FS (r = 0.287, p = 0.029, and r = 0.356, p = 0.006, respectively) and were used in subsequent analysis. SM 38:1 (OR 1.129, 95% CI 1.032, 1.236, p = 0.008) and Cer 34:1 (OR 1.118, 95% CI 1.031, 1.212, p = 0.007), accurately differentiated between FS 0–2 vs. 2.5–6 in regression analysis. A combined lipid score demonstrated fair discrimination in ROC analysis (AUC = 0.729, p = 0.003) and was cross-validated using leave-one-out analysis. Plasma levels of brain-specific SLs may serve as effective biomarkers of subacute white matter disease.

Examining the identification of age-related atrophy between T1 and T1\u2009+\u2009T2-FLAIR cortical thickness measurements

Cortical thickness is traditionally derived from T1-weighted MRI images. Recent studies have shown an improvement in segmentation with the combination of T1 + T2-FLAIR images. MRI data from 54 adults (mean: 71 years, 65–81 years, 48% females) that are part of an ongoing cohort study were analyzed to investigate whether T1 + T2-FLAIR cortical thickness measurements were superior to those derived from T1-weighted images in identifying age-related atrophy. T1-weighted and T2-FLAIR MRI images were processed through FreeSurfer v6.0. Data was extracted using the Desikan-Killiany (DKT) atlas. FreeSurfer’s GUI QDEC examined age-related atrophy. Nonparametric tests, effect sizes, and Pearson correlations examined differences between T1-only and T1 + T2-FLAIR cortical thickness data. These analyses demonstrated that T1 + T2-FLAIR processed images significantly improved the segmentation of gray matter (chi-square x2, p < 0.05) and demonstrated significantly thicker cortical thickness means (p < 0.05) with medium to large effect sizes. Significant regions of age-related cortical atrophy were identified within the T1 + T2-FLAIR data (FDR corrected, p < 0.05). This is in contrast to the T1-only data where no regions survived FDR correction. In summary, T1 + T2-FLAIR data were associated with significant improvement in cortical segmentation and the identification of age-related cortical atrophy. Future studies should consider employing this imaging strategy to obtain cortical thickness measurements sensitive to age-related changes.

Improving survival prediction of high-grade glioma via machine learning techniques based on MRI radiomic, genetic and clinical risk factors

ObjectivesTo develop a radiomic signature to predict overall survival (OS) for high-grade glioma (HGG), and construct a nomogram by combining selected radiomic, genetic and clinical risk factors to further improve the performance of the risk model. Materials and Methods147 cases of HGG with MRI images, genetic data, clinical data were studied, wherein 112 patients were used as training cohort, and 35 patients were as independent test cohort. Radiomics features were extracted from tumor area and peritumoral edema area on CE-T1WI and T2FLAIR images. Association between radiomics signature, genetic, clinical risk factors and OS was explored by Kaplan-Meier survival analysis and log rank test. The multivariate Cox regression analysis was trained with radiomic features along with selected genetic and clinical risk factors, which was presented as a nomogram. ResultsThe radiomic signature constructed by 11 radiomics features stratified patients into low- and high-risk groups, and the C-Index for OS prediction was 0.707 and 0.711 in training and test cohorts, respectively. The multivariable Cox regression analysis identified radiomics signature (hazard ratio (HR): 2.18, P = 0.005), IDH (HR: 0.490, P = 0.007) and age (HR: 1.039, P = 0.005) as independent risk factors. A nomogram combining these independent risk factors further improved the performance for OS estimation (C-index = 0.764 and 0.758 in training and test cohorts, respectively). ConclusionThe radiomics signature is a new prognostic biomarker for HGG. A nomogram incorporating radiomics signature, IDH and age improved the performance of OS estimation, which might be a new complement to the treatment guidelines of glioma.