A5029260213- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- MRI in cancer diagnosis
- Medical Image Segmentation Techniques
- Advanced MRI Techniques and Applications
- Digital Imaging for Blood Diseases
- Cell Image Analysis Techniques
- Medical Imaging Techniques and Applications
- COVID-19 diagnosis using AI
- Medical Imaging and Analysis
- Artificial Intelligence in Healthcare and Education
- Colorectal Cancer Screening and Detection
- Advanced Neuroimaging Techniques and Applications
- Advanced Neural Network Applications
- Digital Radiography and Breast Imaging
- Gene expression and cancer classification
- Brain Tumor Detection and Classification
- Cerebrovascular and Carotid Artery Diseases
- Single-cell and spatial transcriptomics
- Venous Thromboembolism Diagnosis and Management
- Acute Ischemic Stroke Management
- Diagnosis and Treatment of Venous Diseases
- Domain Adaptation and Few-Shot Learning
- Spinal Fractures and Fixation Techniques
- Generative Adversarial Networks and Image Synthesis
University of Toronto
2016-2025
Sunnybrook Health Science Centre
2016-2025
Sunnybrook Hospital
2012-2024
Sunnybrook Research Institute
2015-2024
Health Sciences Centre
2008-2024
Canada Research Chairs
2020
Martel
2020
Vector Institute
2019
Bellingham Technical College
2016-2018
McMaster University
2017
Background— Thromboembolic disease secondary to complicated carotid atherosclerotic plaque is a major cause of cerebral ischemia. Clinical management relies on the detection significant (>70%) stenosis. A large proportion patients suffer irreversible ischemia as result lesser degrees Diagnostic techniques that can identify nonstenotic high-risk would therefore be beneficial. High-risk defined histologically if it contains hemorrhage/thrombus. Magnetic resonance direct thrombus imaging...
Unsupervised learning has been a long-standing goal of machine and is especially important for medical image analysis, where the can compensate scarcity labeled datasets. A promising subclass unsupervised self-supervised learning, which aims to learn salient features using raw input as signal. In this work, we tackle issue domain-specific without any supervision improve multiple task performances that are interest digital histopathology community. We apply contrastive method by collecting...
Background: Current magnetic resonance techniques generate high signal from venous blood and show thrombi as filling defects. Magnetic direct thrombus imaging (MRDTI) directly visualizes acute thrombus. Objective: To determine the accuracy of MRDTI for diagnosis symptomatic deep thrombosis (DVT) below above knee. Design: Prospective, blinded study. Setting: A 1355-bed university hospital. Patients: 101 patients with suspected DVT who had routine venography. Participants were recruited a...
Completely labeled pathology datasets are often challenging and time-consuming to obtain. Semi-supervised learning (SSL) methods able learn from fewer data points with the help of a large number unlabeled points. In this paper, we investigated possibility using clustering analysis identify underlying structure space for SSL. A cluster-then-label method was proposed high-density regions in which were then used supervised SVM finding decision boundary. We have compared our other...
The number of biomedical image analysis challenges organized per year is steadily increasing. These international competitions have the purpose benchmarking algorithms on common data sets, typically to identify best method for a given problem. Recent research, however, revealed that practice related challenge reporting does not allow adequate interpretation and reproducibility results. To address discrepancy between impact quality (control), Biomedical Image Analysis ChallengeS (BIAS)...
Abstract Background Local response prediction for brain metastases (BM) after stereotactic radiosurgery (SRS) is challenging, particularly smaller BM, as existing criteria are based solely on unidimensional measurements. This investigation sought to determine whether radiomic features provide additional value routinely available clinical and dosimetric variables predict local recurrence following SRS. Methods Analyzed were 408 BM in 87 patients treated with A total of 440 extracted from the...
Background— It is recognized that complicated plaque largely accounts for the morbidity and mortality from atherosclerosis. Ideally, investigation of symptomatic asymptomatic patients would identify atheromatous plaques independently stenosis. We have previously shown a magnetic resonance direct thrombus imaging (MRDTI) technique demonstrates atheroma as high signal within carotid arterial wall. used this to examine prevalence in vivo ipsilateral arteries recently with suspected artery...
Early detection of breast cancer is one the most important factors in determining prognosis for women with malignant tumors. Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) has been shown to be sensitive modality screening high-risk women. Computer-aided diagnosis (CAD) systems have potential assist radiologists early cancer. A key component development such a CAD system will selection an appropriate classification function responsible separating and benign lesions. The...
Pathologists often look at whole slide images (WSIs) low magnification to find potentially important regions and then zoom in higher perform more sophisticated analysis of the tissue structures. Many automated methods WSI attempt preprocess down-sampled image order select salient which are further analyzed by a computationally intensive step full magnification. Although it can greatly reduce processing times, this process may lead small being overlooked We propose texture technique ease H&E...
Abstract Hippocampal volumetry is a critical biomarker of aging and dementia, it widely used as predictor cognitive performance; however, automated hippocampal segmentation methods are limited because the algorithms (a) not publicly available, (b) subject to error with significant brain atrophy, cerebrovascular disease lesions, and/or (c) computationally expensive or require parameter tuning. In this study, we trained 3D convolutional neural network using 259 bilateral manually delineated...
Abstract Pathology Image Informatics Platform (PIIP) is an NCI/NIH sponsored project intended for managing, annotating, sharing, and quantitatively analyzing digital pathology imaging data. It expands on existing, freely available image viewer, Sedeen. The goal of this to develop embed some commonly used analysis applications into the Sedeen viewer create a resource cancer research communities. Thus far, new plugins have been developed incorporated platform out focus detection, region...
Increasing evidence shows that flaws in machine learning (ML) algorithm validation are an underestimated global problem. Particularly automatic biomedical image analysis, chosen performance metrics often do not reflect the domain interest, thus failing to adequately measure scientific progress and hindering translation of ML techniques into practice. To overcome this, our large international expert consortium created Metrics Reloaded, a comprehensive framework guiding researchers...