Fig. 1: Integrating multiple 2D tomographic images of a 3D structure.
From: Multiview deep learning improves detection of major cardiac conditions from echocardiography
Each single imaging view contains detailed 2D information about a slice (tomograph) of a 3D structure, such as the heart. Multiple 2D tomographic views often contain distinct and complementary information about different anatomic structures contained within that view. To obtain the most comprehensive assessment of any single diagnosis within a complex organ such as the heart, it is necessary to simultaneously consider information from multiple views at once. This is routinely performed by cardiologists who interpret echocardiograms of the heart. This practice of simultaneously considering differential information from multiple imaging views provides the conceptual basis for the development of the multiview DNN architecture. Images copyright Atif Qasim; reproduced with permission.
