Applications of Deep Learning in Biomedical Datasets

Jahandar Jahanipour

Department of Electrical and Computer Engineering, University of Houston, Houston, TX

Abstract:

Machine learning methods have long been employed to automatically analyze biomedicalmdata. In the past years, large amounts of such biological and clinical data have been collected at an unrivaled speed and scale. That is when deep learning algorithms (along with GPU-accelerated machines) replaced many other machine learning tools to avoid the creation of hand-engineering features and allow us scale analyses to unprecedented amounts of data. TensorFlow, nowadays, is considered as the most popular deep learning platform. In this seminar I will show the applications of deep learning in medical imaging using the most popular framework in hand, TensorFlow. I demonstrate how to implement famous models (such as Neural Network, Convolutional Neural Network, AutoEncoder, …) in order to solve medical-related problems. You will learn about the basics of coding in TensorFlow and creating simple models in order to analyze your data with deep learning architectures.

Bio:

Jahandar Jahanipour is Ph.D. student in the Electrical and Computer Engineering at University of Houston. He is Research Assistant at Bio-Analytics Lab and fellow at the Center for Advanced Computing and Data Systems (CACDS) at University of Houston. His current research involves analysis of whole rat brain slices that have been imaged in a uniquely deep manner – with a panel of 50 molecular markers at an optical resolution of 250nm per pixel. He is developing a unique set of deep learning and clustering based methods to reveal previously unknown cellular differences in cells and detect homogeneity among cells in brain. His research interests include machine learning, deep learning and computer
vision.