Developing Deep Learning-Based Solutions for Radiology
Driven by advances in deep learning, medical image analysis offers the promise to augment and improve the workflow of radiologists. Despite this auspicious outlook, there are numerous challenges that must be overcome. I will speak about the end-to-end process of developing deep-learning based medical image analysis solutions, including: the importance of understanding clinical workflow, and how algorithm-based predictions (correct and incorrect) impact patient management; the importance of non-subjective ground truth; and research to improve transfer learning for large-scale datasets from other domains.
David Richmond is a Senior Scientist at Watson Health Imaging, IBM, where he leads algorithm development on an AI-based medical image analysis solution. Prior to this position, he was Deputy Director of the Image and Data Analysis Core at Harvard Medical School, and he received his PhD from UC Berkeley.