Find out how Artificial Intelligence tackles the huge diversity in the most complex challenges that exist in the medical field.
We’re happy to meet again on December 5 to discuss everything that’s new in Computer Vision, Deep Learning & Artificial Intelligence.
At this Meetup we will be discussing AI in Medical Imaging, with two great speakers:
Moshe Safran – CEO at RSIP Vision USA
Moshe has been with RSIP Vision for over ten years and in 2018 became the head of research and development, providing new ways for the company to solve complex AI technology challenges. Moshe has led the implementation of a wide variety of computer vision projects, from 3D reconstruction of heart chambers (patented) using parametric modeling, to semiconductor precise measurements, deep learning, microscopy, and precise agriculture. He is currently leading multiple RSIP Vision teams simultaneously. His duties include customer communication, driving the focus for the projects, providing professional guidance in algorithm development, recruiting and managing employees, and planning and executing new projects with potential customers. Moshe joined RSIP Vision in 2008 as an algorithm developer before serving almost eight years as algorithm team leader and research scientist. He earned his BSc degree in physics (summa cum laude) from The Hebrew University of Jerusalem after which he spent three years in a computational neuroscience PhD program, conducting research in the fly visual system and in medical image processing.
Mehdi Moradi – Research manager at IBM Research, Almaden Research Lab
Mehdi Moradi is a research staff member and manager at IBM Almaden Research Center. He leads the deep learning/machine learning team focusing on medical image analysis at Almaden. He graduated from the School of Computing at Queen’s University, Kingston, Ontario, Canada, in 2008. After completing research fellowships at the University of British Columbia (2008-2011) and at Harvard Medical School (2011-2012), he joined the Department of Electrical and Computer Engineering at the University of British Columbia as an assistant professor in 2012 and IBM Research in 2014. In his research, he addresses the difficulties of deep learning in data limited scenarios of medical imaging were datasets are small and annotations are expensive. He has served as a program committee member of MICCAI and SPIE Medical Imaging and associate editor of several journals in the area, and extensively published in medical image analysis literature.
18:00-18:30 Arrival, mingling, and pizza
18:30-19:00 Mehdi Moradi (IBM Research): .
In recent years, deep neural networks have improved the benchmarks of learning in areas such as vision and speech recognition. This improvement comes with a big price tag. Deep neural networks are very large supervised models and need huge quantities of labelled data at the time of training. In medical image analysis, labeling data is expensive. In certain imaging modalities, such as MRI and CT, 3D analysis and segmentation are required which increases the size of networks and limit our ability to use transfer learning from 2D models of the mainstream computer vision community. I and my team have worked towards reducing the burden of data labeling and improving 3D networks that are specific to medical imaging. Mehdi will describe some of our contributions in these domains.
19:00-19:30 Moshe Safran (RSIP Vision USA):
Best Practices in Applied Deep Learning for Digital Pathology Multiplex Image Analysis.
We present a comprehensive, neural network approach to multiplex digital pathology image analysis. Tasks include nuclei detection, nuclei segmentation, tumor region of interest segmentation, key marker segmentation, cellular colocalization (classification), and results integration. Our network overcomes a wide variety of qualitative challenges that are difficult if not impossible to address robustly using classical image processing methods, including variations in size, shape, intensity, hollow vs filled, and merging and overlapping nuclei. Our algorithm outperforms classical solutions in the relevant quantitative measures, achieving 94% F1 score for nuclei segmentation, and 94%-99% accuracy in cell classification, in a challenging multiplex image analysis task.
19:30-20:00 – Q&A and further networking
Venue: Quinlan Community Center 10185 N Stelling Rd – Cupertino, CA. This meetup is made possible thanks to sponsorship by RSIP Vision. Please register below!