
Improving Generalization of Deep Networks for Inverse Reconstruction of Image Sequences
Deep learning networks have shown stateoftheart performance in many i...
read it

NeRP: Implicit Neural Representation Learning with Prior Embedding for Sparsely Sampled Image Reconstruction
Image reconstruction is an inverse problem that solves for a computation...
read it

A survey of computational frameworks for solving the acoustic inverse problem in threedimensional photoacoustic computed tomography
Photoacoustic computed tomography (PACT), also known as optoacoustic tom...
read it

Highly Scalable Image Reconstruction using Deep Neural Networks with Bandpass Filtering
To increase the flexibility and scalability of deep neural networks for ...
read it

GeometryAware Neighborhood Search for Learning Local Models for Image Reconstruction
Local learning of sparse image models has proven to be very effective to...
read it

NonEuclidean SelfOrganizing Maps
SelfOrganizing Maps (SOMs, Kohonen networks) belong to neural network m...
read it

A GeometryInformed Deep Learning Framework for UltraSparse 3D Tomographic Image Reconstruction
Deep learning affords enormous opportunities to augment the armamentariu...
read it
Learning GeometryDependent and PhysicsBased Inverse Image Reconstruction
Deep neural networks have shown great potential in image reconstruction problems in Euclidean space. However, many reconstruction problems involve imaging physics that are dependent on the underlying nonEuclidean geometry. In this paper, we present a new approach to learn inverse imaging that exploit the underlying geometry and physics. We first introduce a nonEuclidean encodingdecoding network that allows us to describe the unknown and measurement variables over their respective geometrical domains. We then learn the geometrydependent physics in between the two domains by explicitly modeling it via a bipartite graph over the graphical embedding of the two geometry. We applied the presented network to reconstructing electrical activity on the heart surface from bodysurface potential. In a series of generalization tasks with increasing difficulty, we demonstrated the improved ability of the presented network to generalize across geometrical changes underlying the data in comparison to its Euclidean alternatives.
READ FULL TEXT
Comments
There are no comments yet.