A5003185083- Glioma Diagnosis and Treatment
- Radiomics and Machine Learning in Medical Imaging
- Brain Tumor Detection and Classification
- Immune cells in cancer
- MRI in cancer diagnosis
- Nanoparticle-Based Drug Delivery
- Misinformation and Its Impacts
- Advanced MRI Techniques and Applications
- Social Media and Politics
- Medical Imaging Techniques and Applications
- AI in cancer detection
- S100 Proteins and Annexins
- Brain Metastases and Treatment
- Artificial Intelligence in Healthcare and Education
- Nanoplatforms for cancer theranostics
- Retinal Imaging and Analysis
- COVID-19 diagnosis using AI
- Neuroinflammation and Neurodegeneration Mechanisms
- Hate Speech and Cyberbullying Detection
- Barrier Structure and Function Studies
- Lung Cancer Diagnosis and Treatment
- Machine Learning in Healthcare
- Retinopathy of Prematurity Studies
- Cell Image Analysis Techniques
- Generative Adversarial Networks and Image Synthesis
Massachusetts General Hospital
2017-2023
Athinoula A. Martinos Center for Biomedical Imaging
2017-2023
University of Washington
2020-2023
Center for Neuro-Oncology
2020
Harvard University
2017-2018
Massachusetts Institute of Technology
2017-2018
Bellingham Technical College
2018
University of the Pacific
2001
Retinopathy of prematurity (ROP) is a leading cause childhood blindness worldwide. The decision to treat primarily based on the presence plus disease, defined as dilation and tortuosity retinal vessels. However, clinical diagnosis disease highly subjective variable.To implement validate an algorithm deep learning automatically diagnose from photographs.A convolutional neural network was trained using data set 5511 photographs. Each image previously assigned reference standard (RSD) consensus...
Purpose: Isocitrate dehydrogenase (IDH) mutations in glioma patients confer longer survival and may guide treatment decision making. We aimed to predict the IDH status of gliomas from MR imaging by applying a residual convolutional neural network preoperative radiographic data.Experimental Design: Preoperative was acquired for 201 Hospital University Pennsylvania (HUP), 157 Brigham Women's (BWH), 138 The Cancer Imaging Archive (TCIA) divided into training, validation, testing sets. trained...
Deep learning has become a promising approach for automated support clinical diagnosis. When medical data samples are limited, collaboration among multiple institutions is necessary to achieve high algorithm performance. However, sharing patient often limitations due technical, legal, or ethical concerns. In this study, we propose methods of distributing deep models as an attractive alternative data.We simulate the distribution across 4 using various training heuristics and compare results...
Longitudinal measurement of glioma burden with MRI is the basis for treatment response assessment. In this study, we developed a deep learning algorithm that automatically segments abnormal fluid attenuated inversion recovery (FLAIR) hyperintensity and contrast-enhancing tumor, quantitating tumor volumes as well product maximum bidimensional diameters according to Response Assessment in Neuro-Oncology (RANO) criteria (AutoRANO). Two cohorts patients were used study. One consisted 843...
Abstract Misinformation online poses a range of threats, from subverting democratic processes to undermining public health measures. Proposed solutions encouraging more selective sharing by individuals removing false content and accounts that create or promote it. Here we provide framework evaluate interventions aimed at reducing viral misinformation both in isolation when used combination. We begin deriving generative model spread, inspired research on infectious disease. By applying this...
Performance of models highly depend not only on the used algorithm but also data set it was applied to. This makes comparison newly developed tools to previously published approaches difficult. Either researchers need implement others' algorithms first, establish an adequate benchmark their data, or a direct new and old techniques is infeasible. The Ischemic Stroke Lesion Segmentation (ISLES) challenge, which has ran now consecutively for 3 years, aims address this problem comparability....
Abstract Purpose: Targeting tumor blood vessels is an attractive therapy in glioblastoma (GBM), but the mechanism of action these agents and how they modulate delivery concomitant chemotherapy are not clear humans. We sought to elucidate bevacizumab modulates vasculature impact those vascular changes have on drug patients with recurrent GBM. Experimental Design: Temozolomide was labeled [11C], serial PET-MRI scans were performed GBM treated daily temozolomide. prior first dose, 1 day after...
Generative adversarial networks (GANs) are a class of unsupervised machine learning algorithms that can produce realistic images from randomly-sampled vectors in multi-dimensional space. Until recently, it was not possible to generate high-resolution using GANs, which has limited their applicability medical contain biomarkers only detectable at native resolution. Progressive growing GANs is an approach wherein image generator trained initially synthesize low resolution synthetic (8x8...
Purpose To determine the influence of preprocessing on repeatability and redundancy radiomics features extracted using a popular open-source software package in scan-rescan glioblastoma MRI study. Materials Methods In this study, secondary analysis T2-weighted fluid-attenuated inversion recovery (FLAIR) T1-weighted postcontrast images from 48 patients (mean age, 56 years [range, 22–77 years]) diagnosed with were included two prospective studies (ClinicalTrials.gov NCT00662506 [2009–2011]...
Understanding the spread of online rumors is a pressing societal challenge and an active area research across domains. In context 2022 U.S. midterm elections, one influential social media platform for sharing information — including that may be false, misleading, or unsubstantiated was Twitter (now renamed X). To increase understanding dynamics about we present analyze dataset 1.81 million posts corresponding to 135 distinct which during election season (September 5 December 1, 2022). We...
The World Wide Web is a complex interconnected digital ecosystem, where information and attention flow between platforms communities throughout the globe. These interactions co-construct how we understand world, reflecting shaping public discourse. Unfortunately, researchers often struggle to circulates evolves across web because platform-specific data siloed restricted by linguistic barriers. To address this gap, present comprehensive, multilingual dataset capturing all Wikipedia links...
Including uncertainty information in the assessment of a segmentation pathologic structures on medical images, offers potential to increase trust into deep learning algorithms for analysis imaging. Here, we examine options extract from models and influence choice cost functions these measures. To this end train conventional UNets without dropout, UNet ensembles, Monte-Carlo (MC) dropout segment lung nodules low dose CT using either soft Dice or weighted categorical cross-entropy (wcc) as...
Lung cancer is by far the leading cause of death in US. Recent studies have demonstrated effectiveness screening using low dose CT (LDCT) reducing lung related mortality. While nodules are detected with a high rate sensitivity, this exam has specificity and it still difficult to separate benign malignant lesions. The ISBI 2018 Nodule Malignancy Prediction Challenge, developed team from Quantitative Imaging Network National Cancer Institute, was focused on prediction nodule malignancy two...
We evaluated the efficacy of bavituximab-a mAb with anti-angiogenic and immunomodulatory properties-in newly diagnosed patients glioblastoma (GBM) who also received radiotherapy temozolomide. Perfusion MRI myeloid-related gene transcription inflammatory infiltrates in pre-and post-treatment tumor specimens were studied to evaluate on-target effects (NCT03139916).Thirty-three adults IDH--wild-type GBM 6 weeks concurrent chemoradiotherapy, followed by cycles temozolomide (C1-C6). Bavituximab...
Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) provides quantitative metrics (e.g. Ktrans, ve) via pharmacokinetic models. We tested inter-algorithm variability in these with 11 published DCE-MRI algorithms, all implementing Tofts-Kermode or extended Tofts Digital reference objects (DROs) known Ktrans and ve values were used to assess performance at varying noise levels. Additionally, data from 15 head neck squamous cell carcinoma patients over 3 time-points during...
Deep learning has quickly become the weapon of choice for brain lesion segmentation. However, few existing algorithms pre-configure any biological context their chosen segmentation tissues, and instead rely on neural network's optimizer to develop such associations de novo. We present a novel method applying deep networks problem glioma tissue that takes into account structured nature gliomas - edematous surrounding mutually-exclusive regions enhancing non-enhancing tumor. trained multiple...
Retinopathy of prematurity (ROP) is a disease that affects premature infants, where abnormal growth the retinal blood vessels can lead to blindness unless treated accordingly. Infants considered at risk severe ROP are monitored for symptoms plus disease, characterized by arterial tortuosity and venous dilation posterior pole, with standard photographic definition. Disagreement among experts in diagnosing has driven development computer-based methods classify images based on hand-crafted...
Relative cerebral blood volume (rCBV) cannot be used as a response metric in clinical trials, part, because of variations biomarker consistency and associated interpretation across sites, stemming from differences image acquisition postprocessing methods (PMs). This study leveraged dynamic susceptibility contrast magnetic resonance imaging digital reference object to characterize rCBV 12 sites participating the Quantitative Imaging Network (QIN), specifically focusing on site-specific...
Supervised deep learning algorithms have enabled significant performance gains in medical image classification tasks. But these methods rely on large labeled datasets that require resource-intensive expert annotation. Semi-supervised generative adversarial network (GAN) approaches offer a means to learn from limited data alongside larger unlabeled datasets, but not been applied discern fine-scale, sparse or localized features define abnormalities. To overcome limitations, we propose...
This paper introduces and presents a first analysis of uniquely curated dataset misinformation, disinformation, rumors spreading on Twitter about the 2020 U.S. election. Previous research misinformation—an umbrella term for false misleading content—has largely focused either broad categories, using finite set keywords to cover complex topic, or few, case studies, with increased precision but limited scope. Our approach, by comparison, leverages real-time reports collected from September...