Image Search and Pneumothorax Detection - A PathFinder Project
Learn about an innovative approach to pneumothorax detection using image search technology in this 37-minute conference talk from the Toronto Machine Learning Series. Explore how automated detection systems can assist radiologists in identifying collapsed lungs on chest X-ray images. Discover the potential of using deep features and image matching to provide a "virtual second opinion" for diagnosis. Gain insights into the development of the Autoencoding Thorax Net (AutoThorax-Net) for searching a repository of over 550,000 chest X-ray images. Understand the limitations of deep learning classifiers in clinical practice and the advantages of image search as a triaging tool. Presented by Professor Hamid Reza Tizhoosh from the University of Waterloo, director of Kimia Lab and principal investigator for a PathFinder project.