Learning Iterative Image Reconstruction
نویسنده
چکیده
Successful image reconstruction requires the recognition of a scene and the generation of a clean image of that scene. We propose to use recurrent neural networks for both analysis and synthesis. The networks have a hierarchical architecture that represents images in multiple scales with different degrees of abstraction. The mapping between these representations is mediated by a local connection structure. We supply the networks with degraded images and train them to reconstruct the originals iteratively. This iterative reconstruction makes it possible to use partial results as context information to resolve ambiguities. We demonstrate the power of the approach using three examples: superresolution, fill in of occluded parts, and noise removal / contrast enhancement.
منابع مشابه
Fast System Matrix Calculation in CT Iterative Reconstruction
Introduction: Iterative reconstruction techniques provide better image quality and have the potential for reconstructions with lower imaging dose than classical methods in computed tomography (CT). However, the computational speed is major concern for these iterative techniques. The system matrix calculation during the forward- and back projection is one of the most time- cons...
متن کاملComparing IDREAM as an Iterative Reconstruction Algorithm against In Filtered Back Projection in Computed Tomography
Introduction: Recent studies of Computed Tomography (CT) conducted on patient dose reduction have recommended using an iterative reconstruction algorithm and mA (mili-Ampere) dose modulation. The current study aimed to evaluate Iterative Dose Reduction Algorithm (IDREAM) as an iterative reconstruction algorithm. Material and Methods: Two CT p...
متن کاملDeep Learning-Guided Image Reconstruction from Incomplete Data
An approach to incorporate deep learning within an iterative image reconstruction framework to reconstruct images from severely incomplete measurement data is presented. Specifically, we utilize a convolutional neural network (CNN) as a quasi-projection operator within a least squares minimization procedure. The CNN is trained to encode high level information about the class of images being ima...
متن کاملDR2-Net: Deep Residual Reconstruction Network for Image Compressive Sensing
Most traditional algorithms for compressive sensing image reconstruction suffer from the intensive computation. Recently, deep learning-based reconstruction algorithms have been reported, which dramatically reduce the time complexity than iterative reconstruction algorithms. In this paper, we propose a novel Deep Residual Reconstruction Network (DRNet) to reconstruct the image from its Compress...
متن کاملDeep Learning for Photoacoustic Tomography from Sparse Data
The development of fast and accurate image reconstruction algorithms is a central aspect of computed tomography. In this paper we investigate this issue for the sparse data problem in photoacoustic tomography (PAT). We develop a direct and highly efficient reconstruction algorithm based on deep learning. In our approach image reconstruction is performed with a deep convolutional neural network ...
متن کاملAssessment of simulated patient motion and its effect on myocardial perfusion SPECT using two reconstruction methods (Filtered Backprojection;FBP and Iterative method) [Persian]
Introduction: Motion of the patient during myocardial perfusion SPECT could potentially results in false perfusion defects. The effect of different reconstruction methods on these artifacts is not studied. Clarification of the relation between the extent, severity and duration of motion with the resultant artifacts may be helpful in designing special soft wares for motion correction. This...
متن کامل