Search

Your search keyword '"Mina, Yair"' showing total 3 results
Start Over Author "Mina, Yair" Topic brain
3 results on '"Mina, Yair"'

1. Association of White Matter Hyperintensities With HIV Status and Vascular Risk Factors.

Author

Mina Y, Wu T, Hsieh HC, Hammoud DA, Shah S, Lau CY, Ham L, Snow J, Horne E, Ganesan A, Rapoport SI, Tramont EC, Reich DS, Agan BK, Nath A, and Smith BR

Subjects
Adult, Cross-Sectional Studies, Female, Humans, Leukoaraiosis epidemiology, Longitudinal Studies, Magnetic Resonance Imaging, Male, Middle Aged, Risk Factors, Brain pathology, HIV Infections pathology, Leukoaraiosis pathology, White Matter pathology
Abstract

Objective: To test the hypothesis that brain white matter hyperintensities (WMH) are more common in people living with HIV (PLWH), even in the setting of well-controlled infection, and to identify clinical measures that correlate with these abnormalities., Methods: Research brain MRI scans, acquired within longitudinal studies evaluating neurocognitive outcomes, were reviewed to determine WMH load using the Fazekas visual rating scale in PLWH with well-controlled infection (antiretroviral therapy for at least 1 year and plasma viral load <200 copies/mL) and in sociodemographically matched controls without HIV (CWOH). The primary outcome measure of this cross-sectional analysis was increased WMH load, determined by total Fazekas score ≥2. Multiple logistic regression analysis was performed to evaluate the effect of HIV serostatus on WMH load and to identify MRI, CSF, and clinical variables that associate with WMH in the PLWH group., Results: The study included 203 PLWH and 58 CWOH who completed a brain MRI scan between April 2014 and March 2019. The multiple logistic regression analysis, with age and history of tobacco use as covariates, showed that the adjusted odds ratio of the PLWH group for increased WMH load is 3.7 (95% confidence interval 1.8-7.5; p = 0.0004). For the PLWH group, increased WMH load was associated with older age, male sex, tobacco use, hypertension, and hepatitis C virus coinfection, and also with the presence of measurable tumor necrosis factor α in CSF., Conclusion: Our results suggest that HIV serostatus affects the extent of brain WMH. This effect is mainly associated with aging and modifiable comorbidities., (© 2021 American Academy of Neurology.)

Published
2021
Full Text
View/download PDF

2. Atypical duration of epileptic and psychogenic nonepileptic events.

Author

Anis S, Fahoum F, Korczyn AD, Sverdlov D, Abramovici S, Mina Y, Neufeld MY, and Kipervasser S

Subjects
Adolescent, Adult, Anticonvulsants, Cohort Studies, Electroencephalography methods, Female, Humans, Male, Middle Aged, Psychophysiologic Disorders diagnosis, Retrospective Studies, Seizures diagnosis, Seizures psychology, Time Factors, Video Recording methods, Young Adult, Brain physiopathology, Electroencephalography trends, Psychophysiologic Disorders physiopathology, Seizures physiopathology, Video Recording trends
Abstract

Objectives: We aimed to estimate the frequency of epileptic seizures (ES) and psychogenic nonepileptic seizures (PNES) with atypical duration in our epilepsy monitoring unit (EMU), in order to raise awareness of atypical durations of both types of events., Materials & Methods: We retrospectively reviewed all consecutive video-electroencephalogram (vEEG) recordings in our medical center's EMU from January 2013 to December 2017 and identified patients with seizures with atypical duration. Short PNES were defined as those lasting fewer than 2 min and long ES as those lasting for more than 5 min., Results: The files of 830 adult (age >16 years) patients were reviewed, of whom 26 patients (3.1%, mean age: 33.3 ± 9.8 years, 12 females) were diagnosed as having an unusual seizure duration. Among 432 patients with ES during monitoring, fourteen patients [3.2% (95% confidence interval (CI): 1.5%-5.0%), mean age: 33.0 ± 12.2, 5 females [had long ES durations (exceeding 5 min). In 64% of patients with long ES, the events were provoked by antiepileptic drug (AED) withdrawal during vEEG, 62% had focal lesion on brain imaging, and 64% had a frontotemporal or a temporal seizure focus. Among 223 patients diagnosed with PNES, 12 patients [5.4% (95% CI: 2.2%-8.6%), mean age: 33.6 ± 6.6, 7 females] had short PNES durations (less than 2 min) and demonstrated motor (9/12, 75%), altered responsiveness (6/12, 50%), and vocalization (5/12, 42%) as the most prominent clinical features., Conclusions: The data from our case files highlight two main considerations in the diagnosis of paroxysmal events: prolonged event can be due to ES, while short events can be psychogenic., Competing Interests: Declaration of competing interest None., (Copyright © 2020 Elsevier Inc. All rights reserved.)

Published
2020
Full Text
View/download PDF

3. Logistic regression-based model is more efficient than U-Net model for reliable whole brain MRI segmentation

Author

Dieckhaus, Henry, Meijboom, Rozanna, Okar, Serhat, Wu, Tianxia, Parvathaneni, Prasanna, Mina, Yair, Chandran, Siddharthan, Waldman, Adam D., Reich, Daniel S., and Nair, Govind

Subjects
Deep Learning, Logistic Models, Brain Neoplasms, Image Processing, Computer-Assisted, Brain, Humans, HIV Infections, Magnetic Resonance Imaging, Article
Abstract

OBJECTIVES: Automated whole brain segmentation from magnetic resonance images is of great interest for development of clinically relevant volumetric markers for various neurological diseases. While deep learning methods have demonstrated remarkable potential in this area, they may perform poorly in non-optimal conditions, such as with limited training data. Manual whole brain segmentation is an incredibly tedious process, so minimizing the dataset size required for training segmentation algorithms may be of wide interest. The purpose of this study was to compare the performance of the prototypical deep learning segmentation architecture (U-Net) with a previously published atlas-free traditional machine learning method, Classification using Derivative-based Features (C-DEF) for whole brain segmentation, in the setting of limited training data. MATERIALS AND METHODS: C-DEF and U-Net models were evaluated after training on manually curated data from 5, 10, and 15 participants in two research cohorts – (i) people living with clinically diagnosed HIV infection and (ii) relapsing-remitting Multiple Sclerosis (MS), each acquired at separate institutions, and between 5 and 295 participant data using a large, publicly available and annotated dataset of glioblastoma and lower grade glioma (BraTS). Statistics was performed on Dice Similarity Coefficient (DSC) using repeated-measures ANOVA and Dunnett-Hsu pairwise comparison. RESULTS: C-DEF produced better segmentation than U-Net in lesion (29.2 – 38.9%) and cerebrospinal fluid classes (5.3% – 11.9%) when trained with data from 15 or fewer participants. Unlike C-DEF, U-Net showed significant improvement when increasing the size of the training data (24 – 30% higher than baseline). In the BraTS dataset, C-DEF produced equivalent or better segmentations than U-Net for enhancing tumor and peritumoral edema regions across all training data sizes explored. However, U-Net was more effective than C-DEF for segmentation of necrotic tumor tissue when trained on 10 or more participants, probably due to the inconsistent signal intensity of the tissue class. CONCLUSIONS: These results demonstrate that classical machine learning methods can produce more accurate brain segmentation than the far more complex deep learning methods when only small or moderate amounts of training data are available (n ≤ 15). The magnitude of this advantage varies by tissue and by cohort, while U-Net may be preferable for deep grey matter and necrotic tumor segmentation, particularly with larger training datasets (n ≥ 20). The magnitude of this advantage varies by tissue and by cohort. Given that segmentation models often need to be re-trained for application to novel imaging protocols or pathology, the bottleneck associated with large-scale manual annotation could be avoided with classical machine learning algorithms such as C-DEF.

Published
2022

Catalog

Books, media, physical & digital resources
See catalog results