Introduction to Medical Image Analysis Using Convolutional Neural Networks

Level: Introductory Length: 7 hours Format: In-Person Lecture Intended Audience: Students, researchers, and engineers from academia and industry, who seek to obtain first practical working knowledge in deep learning. Description: Segmentation, detection, and classification are major tasks in medical image analysis and image understanding. Medical imaging researchers heavily use the results of recent developments in machine learning approaches, and with deep learning methods they achieve significantly better results in many real-world problems compared to previous solutions. The course aims to enable students and professionals to apply deep learning methods to their data and problem. Using an interactive programming environment, participants of the course will practically explore all required steps and learn the tools and techniques from data preparation to result interpretation. We will work on example data and train models to segment anatomical structures, to detect abnormalities, and to classify them. We will also give brief insights into adjunct fields, like interpretation of trained models, generative models, image improvement, and image reconstruction from raw data. Participants will work in dockerized environments providing selected deep learning toolkit installations, example data, and teaching Jupyter notebooks. The Docker containers provided are also the course documentation and are meant to be taken home for further use and research. Learning Outcomes: This course will enable you to: - identify the commonly used Deep Learning frameworks (Theano, TensorFlow, CNTK, Caffe, Torch, Lasagne, Keras) and their respective strengths - describe the state of the art of deep learning methods in medical applications - construct computing pipeline using Python based infrastructure, using the above frameworks - select suitable deep learning network architecture for a given problem and implement it - explain and interpret learning progress using appropriate metrics - interpret the model performance using visual analytics Instructor(s): Markus Thorsten Wenzel works on machine learning methods for medical applications since 2005 and has published more than 30 conference and journal papers on the subject. He received his PhD for his work on decision support systems for breast care. At Fraunhofer MEVIS, he is a senior scientist for cognitive medical computing. He is a funded member of the Fraunhofer Society research class "Cognitive Machines" and is experiencend in teaching and lecturing for academia and industry. He has acquired and led several international research projects. Hans Meine is a senior scientist who has been using machine learning for image analysis since 2002, and focused on various medical applications at Fraunhofer MEVIS since 2011. Since early 2016, he is organizing the internal training and coaching of Fraunhofer MEVIS staff for the new methodologies in Deep Learning, and now leads the "Image and Data Analysis" competence area that incorporates both image and non-image data. Recently, his team scored top positions in the “Liver and Tumor Segmentation” challenges at ISBI and MICCAI 2017 using Deep Learning. Event: SPIE Medical Imaging 2018 Course Held: 10 February 2018