When: Wednesday, April 25, 2018
Where: PGH 563
Time: 11:00 AM

Detection of Touching Elongated Cells Using a Stack of Conditional Random Fields

Speaker: Ali Memariani

Automated detection of touching cells in images with inhomogeneous illumination is a challenging problem. A detection framework using a stack of two conditional random fields is proposed to detect touching elongated cells in scanning electron microscopy images with inhomogeneous illumination. The first conditional random field employs shading information to segment the cells where the effect of inhomogeneous illumination is reduced. The second conditional random field estimates the cell walls using their estimated cell wall probability. The method is evaluated using a dataset of Clostridium difficile cells. Finally, the method is compared with two region-based cell detection methods, CellDetect and DeTEC, improving the F-score by at least 20%.

Bio:

Ali Memariani is a PhD student at Computational Biomedicine Lab (CBL). Ali has obtained a B.Sc. in industrial engineering from Tehran Polytechnic, Iran in 2009. He obtained a M.Sc. in computer science from University of Malaya (UM), Malaysia in 2012. He then continued his work as a research assistant in UM for one year working in the area of machine learning and computer vision. Ali’s research in CBL is focused on Analysis of Scanning Electron Microscopy (SEM) images. He has quantified and compared the effect of treatments in deforming the shape of C. difficile cells. Furthermore, Ali has developed an algorithm using conditional random fields to detect touching elongated cells. His current research is focused on detection of cells using deep neural networks.

Education and Training Applications of AR and VR

Speaker: Brian Holtkamp

As virtual reality (VR) and augmented reality (AR) devices become cheaper and more common, the possibility of using them in education and training grows. We're exploring both an AR application for exposure therapy with Autism Spectrum Disorder (ASD) patients and a VR application focused on fire safety training. Exposure therapy is when a therapist recreates a one-on- one social scenario with a ASD patient that attempts to trigger social anxiety. One limitation to the therapy is the inability to change the therapist's appearance to the desired gender, appearance, or age to cause the patient’s anxiety. We’re currently using the Microsoft Hololens to present the ASD patient an avatar that meets their anxiety characteristics. The avatar's actions, appearance, and voice are controlled by the therapist to enhance their exposure therapy session. We're also using VR to enhance the Community Emergency Response Team (CERT) training's fire module. CERT is a program designed to train volunteers in disaster preparedness. One module covered in training is focused on fire safety, focused on how fires form, how to stay safe near fire, and how to extinguish small fires. We’re building a VR application that can reiterate the steps taught in the course on how to properly use a fire extinguisher and can safely simulate a fire inside of a building. This allows the trainee to practice their training and experience the dangers of smoke and fire inside of enclosed spaces that can’t be safely recreated in real life.

Bio:

Brian Holtkamp is a fifth-year PhD candidate at the University of Houston. His advisors are Dr. Chang Yun and Dr. Jaspal Subhlok. He is looking to utilize innovative technologies such as virtual reality and augmented reality to enhance how we teach, learn, and create educational material.

Finding Community Structure in Graphs by Random Walks

Speaker: Reza Fathi

Community detection in graphs is a well-known and studied problem. In general, a community is defined a group of nodes having a higher conductance (higher edges inside a group compared to the members of other groups). It is a hard problem to face because it is an NP-hard problem in general, there is no universal agreement on its definition, and even comparison of different algorithms' result and performance.

In this talk, I will describe our random walk algorithm for finding communities which utilize local mixing time and will share some of our results.

Bio:

eza Fathi is a Fifth year Ph.D. student working with Dr. Gopal Pandurangan. His research interests include Algorithm Design, Big Data Analytics, and Computer Security. His current research is focused on the design of distributed graph algorithms.