Deep Learning in Medical and Biological Image and Video Segmentation

 · 1 min read

Second MLC Meeting on November 8

Our second seminar about machine learning will be focused on deep learning in medical and biological image and video segmentation. Therefore, we are happy to welcome Walter de Back from the Institute for Medical Informatics and Biometry who is an experienced scientist in this field. Walter will lead the half-day event including an introduction and small lecture as well as a hand on tutorial. The event will be suitable for beginners and advanced users in deep learning.

When: 8th of November from 9 am to 2 pm (including 1 hour lunch break)

Where: Toepler-Bau of TU Dresden, Mommsenstra├če 12 in room TOE 203

Abstract of the seminar

Deep learning has a great potential for biomedical image analysis. In particular, convolutional neural networks are an important class of machine learning techniques that can be trained to classify, detect, localize, and segment objects or abnormalities in different image modalities by learning to extract relevant image features. Recent studies have shown that deep learning-based systems can outperform human radiologists in accuracy on a variety of tasks, at only a fraction of the time and cost. In this seminar, I will provide a gentle introduction to convolutional neural networks and their application to biomedical image analysis tasks, highlighting both their strengths and limitations. In addition, we will have a hands-on session training a neural network to segment biomedical images, based on keras.

About Walter de Back

Walter de Back has an MSc in Artificial Intelligence from Utrecht University (NL) and obtained a PhD in Computational Biology from TU Dresden for studies on pattern formation in tissues using multi-scale computational modeling. Walter now works as a postdoc data scientist at the Faculty of Medicine (TU Dresden) where he uses deep learning in a number of projects including cell segmentation in live cell microscopy, dental age estimation from panoramic radiographs, and tumor tissue classification based on mass spectrometry imaging data. More information on: