scholarly journals Conditional Invertible Neural Networks for Medical Imaging

2021 ◽  
Vol 7 (11) ◽  
pp. 243
Author(s):  
Alexander Denker ◽  
Maximilian Schmidt ◽  
Johannes Leuschner ◽  
Peter Maass
Keyword(s):  
Neural Networks ◽  
Medical Imaging ◽  
Inverse Problems ◽  
Low Dose ◽  
Data Driven ◽  
Resonance Imaging ◽  
Point Estimates ◽  
Ill Posed ◽  
Low Dose Computed Tomography

Over recent years, deep learning methods have become an increasingly popular choice for solving tasks from the field of inverse problems. Many of these new data-driven methods have produced impressive results, although most only give point estimates for the reconstruction. However, especially in the analysis of ill-posed inverse problems, the study of uncertainties is essential. In our work, we apply generative flow-based models based on invertible neural networks to two challenging medical imaging tasks, i.e., low-dose computed tomography and accelerated medical resonance imaging. We test different architectures of invertible neural networks and provide extensive ablation studies. In most applications, a standard Gaussian is used as the base distribution for a flow-based model. Our results show that the choice of a radial distribution can improve the quality of reconstructions.

Low-dose computed tomography: A solution for in vivo medical imaging and accurate patient-specific 3D bone modeling?

2006 ◽  
Vol 21 (9) ◽  
pp. 992-998 ◽  
Author(s):  
Serge Van Sint Jan ◽  
Stéphane Sobzack ◽  
Pierre-Michel Dugailly ◽  
Véronique Feipel ◽  
Philippe Lefèvre ◽  
...  
Keyword(s):  
Computed Tomography ◽  
Medical Imaging ◽  
Low Dose ◽  
Patient Specific ◽  
Bone Modeling ◽  
Low Dose Computed Tomography

scholarly journals Low-Dose Computed Tomography Filtering Using Geodesic Distances

2020 ◽  
Author(s):  
Daniel A. Góes ◽  
Nelson D. A. Mascarenhas
Keyword(s):  
Computed Tomography ◽  
Low Dose ◽  
Convex Sets ◽  
Projection Onto Convex Sets ◽  
Local Means ◽  
Gamma Distributions ◽  
Non Local ◽  
Geodesic Distances ◽  
Low Dose Computed Tomography

Due to the concerns related to patient exposure to X-ray, the dosage used in computed tomography must be reduced (Low-dose Computed Tomography - LDCT). One of the effects of LDCT is the degradation in the quality of the final reconstructed image. In this work, we propose a method of filtering LDCT sinograms that are subject to signal-dependent Poisson noise. To filter this type of noise, we use a Bayesian approach, changing the Non-local Means (NLM) algorithm to use geodesic stochastic distances for Gamma distribution, the conjugate prior to Poisson, as a similarity metric between each projection point. Among the geodesic distances evaluated, we found a closed solution for the Shannon entropy for Gamma distributions. We compare our method with the following methods based on NLM: PoissonNLM, Stochastic Poisson NLM, Stochastic Gamma NLM and the original NLM after Anscombe transform. We also compare with BM3D after Anscombe transform. Comparisons are made on the final images reconstructed by the Filtered-Back Projection (FBP) and Projection onto Convex Sets (POCS) methods using the metrics PSNR and SSIM.

scholarly journals Emphysema Quantification in Inflation-Fixed Lungs Using Low-Dose Computed Tomography and 3He Magnetic Resonance Imaging

2010 ◽  
Vol 34 (5) ◽  
pp. 773-779 ◽  
Author(s):  
David S. Gierada ◽  
Jason C. Woods ◽  
Richard E. Jacob ◽  
Andrew J. Bierhals ◽  
Cliff K. Choong ◽  
...  
Keyword(s):  
Magnetic Resonance Imaging ◽  
Computed Tomography ◽  
Magnetic Resonance ◽  
Low Dose ◽  
Resonance Imaging ◽  
Low Dose Computed Tomography

scholarly journals Whole Body Low Dose Computed Tomography (WBLDCT) Can Be Comparable to Whole-Body Magnetic Resonance Imaging (WBMRI) in the Assessment of Multiple Myeloma

Diagnostics ◽  
2021 ◽  
Vol 11 (5) ◽  
pp. 857
Author(s):  
Davide Ippolito ◽  
Teresa Giandola ◽  
Cesare Maino ◽  
Davide Gandola ◽  
Maria Ragusi ◽  
...  
Keyword(s):  
Magnetic Resonance Imaging ◽  
Computed Tomography ◽  
Multiple Myeloma ◽  
Magnetic Resonance ◽  
Low Dose ◽  
Whole Body ◽  
Resonance Imaging ◽  
Lesion Number ◽  
Low Dose Computed Tomography ◽  
Ct And Mri

Aim of the study is to compare the agreement between whole-body low-dose computed tomography (WBLDCT) and magnetic resonance imaging (WBMRI) in the evaluation of bone marrow involvement in patients with multiple myeloma (MM). Patients with biopsy-proven MM, who underwent both WBLDCT and WBMRI were retrospectively enrolled. After identifying the presence of focal bone involvement (focal infiltration pattern), the whole skeleton was divided into five anatomic districts (skull, spine, sternum and ribs, pelvis, and limbs). Patients were grouped according to the number and location of the lytic lesions (<5, 5–20, and >20) and Durie and Salmon staging system. The agreement between CT and MRI regarding focal pattern, staging, lesion number, and distribution was assessed using the Cohen Kappa statistics. The majority of patients showed focal involvement. According to the distribution of the focal lesions and Durie Salmon staging, the agreement between CT and MRI was substantial or almost perfect (all κ > 0.60). The agreement increased proportionally with the number of lesions in the pelvis and spine (κ = 0.373 to κ = 0.564, and κ = 0.469–0.624), while for the skull the agreement proportionally decreased without reaching a statistically significant difference (p > 0.05). In conclusion, WBLDCT showed an almost perfect agreement in the evaluation of focal involvement, staging, lesion number, and distribution of bone involvement in comparison with WBMRI.

scholarly journals DEEP LEARNING FOR LOW-DOSE CT NOISE REMOVAL USING DILATED CONVOLUTION AND PERCEPTUAL LOSS

2021 ◽  
Author(s):  
Maryam Gholizadeh-Ansari
Keyword(s):  
Deep Neural Network ◽  
Low Dose ◽  
State Of The Art ◽  
Noise Removal ◽  
Ct Image ◽  
Low Dose Ct ◽  
Dilated Convolution ◽  
Residual Learning ◽  
Low Dose Computed Tomography

Low-dose computed tomography has been recommended to reduce the radiation risks of CT scans for patients. However, the reconstructed CT image will be considerably degraded because of photon starvation. Both traditional noise removal techniques and neural networks have been used to enhance the quality of low-dose CT images. In this study, a deep neural network is proposed to mitigate this problem. The network employs dilated convolution, batch normalization, and residual learning. Moreover, a nontrainable edge detection layer is proposed helping to produce sharper edges in the output image without introducing additional complexity. This network is optimized by a combination of mean-square error and perceptual loss to preserve textural details in the CT image that are critical for diagnosis. This objective function solves the over-smoothing problem and grid-like artifacts caused by per-pixel loss and perceptual loss, respectively. The experiments demonstrate the effects of each modification to the network and confirm that the proposed network achieves better performance relative to the state of the art methods.

On a characteristic property of conditionally well-posed problems

2015 ◽  
Vol 23 (3) ◽  
Author(s):  
Mikhail Y. Kokurin
Keyword(s):  
Inverse Problems ◽  
Input Data ◽  
Characteristic Property ◽  
Optimization Problems ◽  
Data Driven ◽  
Error Models ◽  
Regularizing Algorithms ◽  
Stochastic Error ◽  
Ill Posed ◽  
Well Posed

AbstractThe aim of this paper is to discuss and illustrate the fact that conditionally well-posed problems stand out among all ill-posed problems as being regularizable via an operator independent of the level of errors in input data. We give examples of corresponding purely data driven regularizing algorithms for various classes of conditionally well-posed inverse problems and optimization problems in the context of deterministic and stochastic error models.

scholarly journals Solving inverse problems using data-driven models

Acta Numerica ◽  
2019 ◽  
Vol 28 ◽  
pp. 1-174 ◽  
Author(s):  
Simon Arridge ◽  
Peter Maass ◽  
Ozan Öktem ◽  
Carola-Bibiane Schönlieb
Keyword(s):  
Inverse Problems ◽  
Industrial Applications ◽  
Data Driven ◽  
Analytical Models ◽  
Survey Paper ◽  
Domain Specific ◽  
Combining Data ◽  
Domain Specific Knowledge ◽  
Ill Posed ◽  
Using Data

Recent research in inverse problems seeks to develop a mathematically coherent foundation for combining data-driven models, and in particular those based on deep learning, with domain-specific knowledge contained in physical–analytical models. The focus is on solving ill-posed inverse problems that are at the core of many challenging applications in the natural sciences, medicine and life sciences, as well as in engineering and industrial applications. This survey paper aims to give an account of some of the main contributions in data-driven inverse problems.

Sign in / Sign up

Export Citation Format

Share Document