A5102845929- Advanced Vision and Imaging
- Computer Graphics and Visualization Techniques
- Head and Neck Cancer Studies
- 3D Shape Modeling and Analysis
- Advanced Data Storage Technologies
- Adversarial Robustness in Machine Learning
- Radiomics and Machine Learning in Medical Imaging
- Molecular Biology Techniques and Applications
- Receptor Mechanisms and Signaling
- Head and Neck Surgical Oncology
- Salivary Gland Tumors Diagnosis and Treatment
- Advanced Image Processing Techniques
- Privacy-Preserving Technologies in Data
- Advanced Optical Sensing Technologies
- Cancer Diagnosis and Treatment
- Medical Imaging Techniques and Applications
- Image Processing Techniques and Applications
- Neuroscience and Neuropharmacology Research
- Advanced Memory and Neural Computing
- Neurotransmitter Receptor Influence on Behavior
- Patient-Provider Communication in Healthcare
- Brain Metastases and Treatment
- Radiation Effects in Electronics
- Brain Tumor Detection and Classification
- Generative Adversarial Networks and Image Synthesis
Dartmouth College
2024
Einstein Medical Center Philadelphia
2022-2024
Thomas Jefferson University
2024
Rutgers, The State University of New Jersey
2023
Nvidia (United States)
2023
Borough of Manhattan Community College
2021
City University of New York
2021
Virginia Tech
2019
University of Illinois Urbana-Champaign
2019
National Center for Supercomputing Applications
2019
Gan, Tong J.; Diemunsch, Pierre; Habib, Ashraf S.; Kovac, Anthony; Kranke, Peter; Meyer, Tricia A.; Watcha, Mehermoor; Chung, Frances; Angus, Shane; Apfel, Christian C.; Bergese, Sergio D.; Candiotti, Keith Chan, Matthew; Davis, Peter Hooper, Vallire Lagoo-Deenadayalan, Sandhya; Myles, Paul; Nezat, Greg; Philip, Beverly K.; Tramèr, Martin R. Author Information
We present a one-shot method to infer and render photorealistic 3D representation from single unposed image (e.g., face portrait) in real-time. Given RGB input, our encoder directly predicts canonical triplane of neural radiance field for 3D-aware novel view synthesis via volume rendering. Our is fast (24 fps) on consumer hardware, produces higher quality results than strong GAN-inversion baselines that require test-time optimization. To train pipeline, we use only synthetic data, showing...
This study evaluates population-based outcomes of patients with squamous cell carcinoma (SCC) the nasal cavity treated in British Columbia.A retrospective review SCC from 1984 to 2014 was performed (n = 159). Locoregional recurrence (LRR) and overall survival (OS) were evaluated.The 3-year OS 74.2% for radiation alone, 75.8% surgery 78.4% ( P 0.16). The LRR 28.4% 28.2% 22.6% 0.21). On multivariable analysis, postoperative relative alone associated a lower risk (hazard ratio: 0.36, 0.03)....
The purpose of this case series was to review a rare subset tumors known as gastric lipomas, which are typically found incidentally. motivation for study arose from the identification 2 cases within our institution in short period.
Abstract Summary Elemental imaging provides detailed profiling of metal bioaccumulation, offering more precision than bulk analysis by targeting specific tissue areas. However, accurately identifying comparable regions from elemental maps is challenging, requiring the integration hematoxylin and eosin (H&E) slides for effective comparison. Facilitating streamlined co-registration Whole Slide Images (WSI) maps, TRACE enhances abundance in various pathological conditions. Through an...
Abstract Colorectal Cancer (CRC) accounts for nearly 10% of all cancer cases and is the second-leading cause cancer-related deaths worldwide. The increased incidence CRC in younger demographics underscores need improved screening prognostication. These enhancements are crucial to augment staging practices which focus on accurate evaluation local lymph node involvement distant metastasis. Essential trace elements such as copper iron within tumor microenvironment (TME) present novel...
Abstract Essential trace elements are crucial in various biological processes, including tumor growth, migration, and metastasis. For instance, copper ions involved mitochondrial metabolism, cell proliferation, metastasis proangiongenic pathways. Localizing where metal ion homeostasis is disrupted vital for understanding these processes beyond what bulk analysis can reveal. Elemental imaging offers high-resolution, quantifiable multi-elemental distribution maps, providing a more detailed...
Human error is the main cause of many road traffic accidents which every year take lives millions people and injure more. Driving safety thus a major concern leading to research in autonomous driving systems. One project at centre for computational intelligence (C <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> i), NTU, aims using fuzzy neural architectures such as GenSoFNN-Yager realize intelligent Le., learn autonomously park, make...
3D-aware Generative Adversarial Networks (GANs) have shown remarkable progress in learning to generate multi-view-consistent images and 3D geometries of scenes from collections 2D via neural volume rendering. Yet, the significant memory computational costs dense sampling rendering forced GANs adopt patch-based training or employ low-resolution with post-processing super resolution, which sacrifices multiview consistency quality resolved geometry. Consequently, not yet been able fully resolve...
Recent breakthroughs in single-image 3D portrait reconstruction have enabled telepresence systems to stream videos from a single camera real-time, potentially democratizing telepresence. However, per-frame exhibits temporal inconsistency and forgets the user's appearance. On other hand, self-reenactment methods can render coherent portraits by driving personalized prior, but fail faithfully reconstruct appearance (e.g., facial expressions lighting). In this work, we recognize need maintain...
Point-spread-function (PSF) engineering is a well-established computational imaging technique that uses phase masks and other optical elements to embed extra information (e.g., depth) into the images captured by conventional CMOS image sensors. To date, however, PSF-engineering has not been applied neuromorphic event cameras; powerful new sensing technology responds changes in log-intensity of light. This paper establishes theoretical limits (Cram\'er Rao bounds) on 3D point localization...
Abstract The complex interplay between metal abundance, transport mechanisms, cell distribution, and tumor progression-related biological pathways (e.g., metabolism, collagen remodeling) remains poorly understood. Traditionally, genes metals have been studied in isolation, limiting insights into their interactions. Recent advances spatial transcriptomics elemental profiling now enable comprehensive exploration of tissue-wide metal-gene interactions, though integration challenging. In this...
Recent breakthroughs in single-image 3D portrait reconstruction have enabled telepresence systems to stream videos from a single camera real-time, democratizing telepresence. However, per-frame exhibits temporal inconsistency and forgets the user's appearance. On other hand, self-reenactment methods can render coherent portraits by driving avatar built reference image, but fail faithfully preserve appearance (e.g., instantaneous facial expression lighting). As result, none of these two...
This talk explores the fundamental limits of 3D tracking with PSF engineering and event cameras. We demonstrate existing phase masks are already near-optimal for localizing flashing point sources, but suboptimal moving objects. Full-text article not available; see video presentation
Elemental imaging provides detailed profiling of metal bioaccumulation, offering more precision than bulk analysis by targeting specific tissue areas. However, accurately identifying comparable regions from elemental maps is challenging, requiring the integration hematoxylin and eosin (H&E) slides for effective comparison. Facilitating streamlined co-registration whole slide images (WSI) maps, TRACE enhances abundance in various pathological conditions. Through an interactive containerized...
High dynamic range (HDR) images are important for a of tasks, from navigation to consumer photography. Accordingly, host specialized HDR sensors have been developed, the most successful which based on capturing variable per-pixel exposures. In essence, these methods capture an entire exposure bracket sequence at once in single shot. This paper presents straightforward but highly effective approach turning off-the-shelf polarization camera into high-performance camera. By placing linear...