A5053558899- Radiomics and Machine Learning in Medical Imaging
- AI in cancer detection
- Digital Radiography and Breast Imaging
- Artificial Intelligence in Healthcare and Education
- Image Retrieval and Classification Techniques
- Monoclonal and Polyclonal Antibodies Research
- Global Cancer Incidence and Screening
- vaccines and immunoinformatics approaches
- Renal cell carcinoma treatment
- Computational Drug Discovery Methods
- Advanced Neural Network Applications
- Influenza Virus Research Studies
- Protein Structure and Dynamics
- Text and Document Classification Technologies
- DNA and Biological Computing
- Biomedical Text Mining and Ontologies
- Domain Adaptation and Few-Shot Learning
- Cell Image Analysis Techniques
- Artificial Intelligence in Healthcare
- Multimodal Machine Learning Applications
- Diatoms and Algae Research
- Microbial Natural Products and Biosynthesis
- Adversarial Robustness in Machine Learning
- Advanced X-ray and CT Imaging
- Machine Learning and Data Classification
IBM Research - Haifa
2017-2025
Johns Hopkins Medicine
2022
University of Haifa
2019-2022
Johns Hopkins University
2022
Maccabi Health Care Services
2019
Assuta Medical Center
2019
Background Computational models on the basis of deep neural networks are increasingly used to analyze health care data. However, efficacy traditional computational in radiology is a matter debate. Purpose To evaluate accuracy and efficiency combined machine learning approach for early breast cancer detection applied linked set digital mammography images electronic records. Materials Methods In this retrospective study, 52 936 were collected 13 234 women who underwent at least one mammogram...
Background Digital breast tomosynthesis (DBT) has higher diagnostic accuracy than digital mammography, but interpretation time is substantially longer. Artificial intelligence (AI) could improve reading efficiency. Purpose To evaluate the use of AI to reduce workload by filtering out normal DBT screens. Materials and Methods The retrospective study included 13 306 examinations from 9919 women performed between June 2013 November 2018 two health care networks. cohort was split into training,...
Importance An accurate and robust artificial intelligence (AI) algorithm for detecting cancer in digital breast tomosynthesis (DBT) could significantly improve detection accuracy reduce health care costs worldwide. Objectives To make training evaluation data the development of AI algorithms DBT analysis available, to develop well-defined benchmarks, create publicly available code existing methods. Design, Setting, Participants This diagnostic study is based on a multi-institutional...
Monoclonal antibodies (mAbs) represent one of the most prevalent FDA-approved modalities for treating autoimmune diseases, infectious and cancers. However, discovery development therapeutic remains a time-consuming expensive process. Recent advancements in machine learning (ML) artificial intelligence (AI) have shown significant promise revolutionizing antibody optimization. In particular, models that predict biological activity enable in-silico evaluation binding functional properties; such...
This paper presents the challenge report for 2021 Kidney and Tumor Segmentation Challenge (KiTS21) held in conjunction with international conference on Medical Image Computing Computer Assisted Interventions (MICCAI). KiTS21 is a sequel to its first edition 2019, it features variety of innovations how was designed, addition larger dataset. A novel annotation method used collect three separate annotations each region interest, these were performed fully transparent setting using web-based...
Machine Learning is at the forefront of scientific progress in Healthcare and Medicine.To accelerate discovery, it important to have tools that allow iterations be collaborative, reproducible, reusable easily built upon without "reinventing wheel" for each task.FuseMedML, or fuse, a Python framework designed accelerated (ML) based discovery medical domain.It highly flexible easy collaboration, encouraging code reuse.Flexibility enabled by generic data object design where kept nested...
Drug discovery typically consists of multiple steps, including identifying a target protein key to disease's etiology, validating that interacting with this could prevent symptoms or cure the disease, discovering small molecule biologic therapeutic interact it, and optimizing candidate through complex landscape required properties. related tasks often involve prediction generation while considering entities potentially interact, which poses challenge for typical AI models. For purpose we...
Recent work showed that active site rather than full-protein-sequence information improves predictive performance in kinase-ligand binding affinity prediction. To refine the notion of an "active site", we here propose and compare multiple definitions. We report significant evidence our novel definition is superior to previous definitions better models ATP-noncompetitive inhibitors. Moreover, leverage discontiguity sequence motivate protein-sequence augmentation strategies find combining them...
Medical image classification involves thresholding of labels that represent malignancy risk levels. Usually, a task defines single threshold, and when developing computer-aided diagnosis tools, network is trained per such e.g. as screening out healthy (very low risk) patients to leave possibly sick ones for further analysis (low threshold), or trying find malignant cases among those marked non-risk by the radiologist ("second reading", high threshold). We propose way rephrase problem in...
Deep neural networks have demonstrated impressive performance in various machine learning tasks. However, they are notoriously sensitive to changes data distribution. Often, even a slight change the distribution can lead drastic reduction. Artificially augmenting may help some extent, but most cases, fails achieve model invariance Some examples where this sub-class of domain adaptation be valuable imaging modalities such as thermal imaging, X-ray, ultrasound, and MRI, acquisition parameters...
Bioactivity data plays a key role in drug discovery and repurposing. The resource-demanding nature of \textit{in vitro} vivo} experiments, as well the recent advances data-driven computational biochemistry research, highlight importance silico} target interaction (DTI) prediction approaches. While numerous large public bioactivity sources exist, research field could benefit from better standardization existing resources. At present, different works that share similar goals are often...
As machine learning algorithms continue to improve, there is an increasing need for explaining why a model produces certain prediction input. In recent years, several methods interpretability have been developed, aiming provide explanation of which subset regions the input main reason prediction. parallel, significant research community effort occurring in years developing adversarial example generation fooling models, while not altering true label input,as it would classified by human...