Extracting and Learning Fine-Grained Labels from Chest Radiographs.

Chest radiographs are the most common diagnostic exam in emergency rooms and intensive care units today. Recently, a number of researchers have begun working on large chest X-ray datasets to develop deep learning models for recognition of a handful of coarse finding classes such as opacities, masses and nodules. In this paper, we focus on extracting and learning fine-grained labels for chest X-ray images. Specifically we develop a new method of extracting fine-grained labels from radiology reports by combining vocabulary-driven concept extraction with phrasal grouping in dependency parse trees for association of modifiers with findings. A total of457finegrained labels depicting the largest spectrum of findings to date were selected and sufficiently large datasets acquired to train a new deep learning model designed for fine-grained classification. We show results that indicate a highly accurate label extraction process and a reliable learning of fine-grained labels. The resulting network, to our knowledge, is the first to recognize fine-grained descriptions offindings in images covering over nine modifiers including laterality, location, severity, size and appearance.

Inter-model consistency and complementarity: learning from ex-vivo imaging and electrophysiological data towards an integrated understanding of cardiac physiology.

Computational models of the heart at various scales and levels of complexity have been independently developed, parameterised and validated using a wide range of experimental data for over four decades. However, despite remarkable progress, the lack of coordinated efforts to compare and combine these computational models has limited their impact on the numerous open questions in cardiac physiology. To address this issue, a comprehensive dataset has previously been made available to the community that contains the cardiac anatomy and fibre orientations from magnetic resonance imaging as well as epicardial transmembrane potentials from optical mapping measured on a perfused ex-vivo porcine heart. This data was used to develop and customize four models of cardiac electrophysiology with different level of details, including a personalized fast conduction Purkinje system, a maximum a posteriori estimation of the 3D distribution of transmembrane potential, the personalization of a simplified reaction-diffusion model, and a detailed biophysical model with generic conduction parameters. This study proposes the integration of these four models into a single modelling and simulation pipeline, after analyzing their common features and discrepancies. The proposed integrated pipeline demonstrates an increase prediction power of depolarization isochrones in different pacing conditions.