På svenska

Presentation Information     2018-01-29 (14:15)   •  4307

Speaker Anindya Gupta  (CBA)
Title Deep neural networks for the classification and denoising of medical and biomedical images
Abstract As soon as it became possible to scan and load medical or biomedical images into a computer, researchers have built systems for automated analysis. Since the inception of deep learning, research has seen a shift from rule-based, problem-specific solutions to increasingly generic, problem agnostic methods that rely on training data. In particular, convolutional neural networks have rapidly become a primary choice for many computer-aided detection (CAD) workflows due to its astonishing results in multiple tasks, e.g., classification, denoising, etc. Given the prevalence of deep neural networks, I will highlight the applications of neural networks in medical and biomedical image analysis for two tasks. Firstly, classification of pulmonary nodules in CT images and classification of cilia in low-magnification TEM images. Secondly, denoising of short exposure high-magnification TEM images for ultrastructural enhancement.