Skip to content

200223-200229-papers-reading.md

Latest commit

 

History

History
103 lines (70 loc) · 11.7 KB

200223-200229-papers-reading.md

File metadata and controls

103 lines (70 loc) · 11.7 KB

200223-200229 论文阅读总结

[TOC]

本周一共读了 46 篇文章。

论文列表

  1. SUPER Learning: A Supervised-Unsupervised Framework for Low-Dose CT Image Reconstruction: CT

  2. BCD-Net for Low-dose CT Reconstruction: Acceleration, Convergence, and Generalization: CT

  3. Scan‐specific robust artificial‐neural‐networks for k‐space interpolation (RAKI) reconstruction: Database‐free deep learning for fast imaging: MRI

  4. A Structural Oriented Training Method for GAN Based Fast Compressed Sensing MRI: MRI

  5. Undersampled MR image reconstruction using an enhanced recursive residual network: MRI

  6. Learning the invisible A hybrid deep learning-shearlet framework for limited angle computed tomography: CT

  7. Convolutional regularization methods for 4D, x-ray CT reconstruction: CT

  8. Robust Compressive Sensing MRI Reconstruction using Generative Adversarial Networks: MRI

  9. Denoising Prior Driven Deep Neural Network for Image Restoration: inverse problem

  10. Deep Neural Networks for Sparse-View Filtered Backprojection Imaging: CT

  11. Online MR image reconstruction for compressed sensing acquisition in T2* imaging: MRI

  12. Promising Generative Adversarial Network Based Sinogram Inpainting Method for Ultra-Limited-Angle Computed Tomography Imaging: CT

  13. Deep Back Projection for Sparse-view CT Reconstruction: CT

  14. What do AI algorithms actually learn? – On false structures in deep learning: theory

  15. Learning Converged Propagations with Deep Prior Ensemble for Image Enhancement: inverse problem

  16. Deep Plug-and-Play Prior for Parallel MRI Reconstruction: MRI

  17. Consensus Neural Network for Medical Imaging Denoising with Only Noisy Training Samples: denoise

  18. Deep learning in medical imaging and radiation therapy: Review

  19. Parallel imaging in time‐of‐flight magnetic resonance angiography using deep multistream convolutional neural networks: MRI

  20. Highly Accelerated Multishot EPI through Synergistic Machine Learning and Joint Reconstruction: EPI

  21. ETER-net: End to End MR Image Reconstruction Using Recurrent Neural Network: MRI

  22. InversionNet: An Efficient and Accurate Data-driven Full Waveform Inversion: Waveform Inversion

  23. Plug-and-Play Methods for Magnetic Resonance Imaging: MRI

  24. Nonlinear Dipole Inversion (NDI) enables Quantitative Susceptibility Mapping (QSM) without parameter tuning: QSM

  25. LORAKI: Autocalibrated Recurrent Neural Networks for Autoregressive MRI Reconstruction in k-Space: MRI

  26. Deep Learning Formulation of ECGI for Data-Driven Integration of Spatiotemporal Correlations and Imaging Information: ECGI

  27. Universal Deep Beamformer for Variable Rate Ultrasound Imaging: US

  28. Learning the Sampling Pattern for MRI: MRI

  29. Exploiting Motion for Deep Learning Reconstruction of Extremely-Undersampled Dynamic MRI: MRI

  30. Model-based Free-Breathing Cardiac MRI Reconstruction Using Deep Learned & Storm Priors: MODL-STORM: MRI

  31. Data consistency networks for (calibration-less) accelerated parallel MR image reconstruction: MRI

  32. A hierarchical approach to deep learning and its application to tomographic reconstruction: CT

  33. Accelerated Coronary MRI Using 3D SPIRIT-RAKI with Sparsity Regularization: MRI

  34. Accelerated Simultaneous Multi-Slice MRI using Subject-Specific Convolutional Neural Networks: MRI

  35. Σ-net: Ensembled Iterative Deep Neural Networks for Accelerated Parallel MR Image Reconstruction: MRI

  36. Variational Deep Learning for Low-dose Computed Tomography: CT

  37. Unsupervised Deep Basis Pursuit: Learning inverse problems without ground-truth data: MRI

  38. Solving Inverse Computational Imaging Problems using Deep Pixel-level Prior: inverse problem

  39. Complex Fully Convolutional Neural Networks for MR Image Reconstruction: MRI

  40. Multi-Scale Learned Iterative Reconstruction: inverse problem

  41. SANTIS Sampling-Augmented Neural neTwork with Incoherent Structure for MR image reconstruction: MRI

  42. On the statistical rate of nonlinear recovery in generative models with heavy-tailed data: theory

  43. Deep Learning Enables Reduced Gadolinium Dose for Contrast-Enhanced Brain MRI: MRI

  44. Cardiac MR Segmentation from Undersampled k-space Using Deep Latent Representation Learning: MRI

  45. A Deep Ensemble Network for Compressed Sensing MRI: MRI

  46. A Prior Learning Network for Joint Image and Sensitivity Estimation in Parallel MR Imaging: MRI

  47. BCD-Net for Low-dose CT Reconstruction: Acceleration, Convergence, and Generalization: CT

  48. Scan‐specific robust artificial‐neural‐networks for k‐space interpolation (RAKI) reconstruction: Database‐free deep learning for fast imaging: MRI

  49. A Structural Oriented Training Method for GAN Based Fast Compressed Sensing MRI: MRI

  50. Undersampled MR image reconstruction using an enhanced recursive residual network: MRI

  51. Learning the invisible A hybrid deep learning-shearlet framework for limited angle computed tomography: CT

  52. Convolutional regularization methods for 4D, x-ray CT reconstruction: CT

  53. Robust Compressive Sensing MRI Reconstruction using Generative Adversarial Networks: MRI

  54. Denoising Prior Driven Deep Neural Network for Image Restoration: inverse problem

  55. Deep Neural Networks for Sparse-View Filtered Backprojection Imaging: CT

  56. Online MR image reconstruction for compressed sensing acquisition in T2* imaging: MRI

  57. Promising Generative Adversarial Network Based Sinogram Inpainting Method for Ultra-Limited-Angle Computed Tomography Imaging:

Ideas from papers

  • 在 MRI 采样的同时进行重建： Online MR image reconstruction for compressed sensing acquisition in T2* imaging <<<<<<< HEAD
  • 理论探讨——深度学习学到的结构是 false structure：What do AI algorithms actually learn? – On false structures in deep learning
  • 把 DIP 类型的损失函数作为一般的无监督训练的损失函数：Unsupervised Deep Basis Pursuit: Learning inverse problems without ground-truth data
  • 对于某些具有 scale invariance 特点的测量矩阵，可以通过不同粒度（或者说不同的离散化程度）在迭代过程中逐步细化，来达到减少计算量的目标：Multi-Scale Learned Iterative Reconstruction

个人思考

  • 发现不少文章都会对图像进行一个变换作为网络的输入。这就和分类或分割相反，在神经网络的表达中，如何实现不同域之间的转换，空间划分这个角度能否用来解释，另外一个问题是两个域在网络的分界点怎么确定？

  • What do AI algorithms actually learn? – On false structures in deep learning 这篇文章让我思考如何用划分子空间的方式来解释这个问题。虽然作者认为 false structure 这个结论只是一个假设，但我认为一定存在某种对深度学习这种方法的刻画，这种刻画应当是和 false structure 是对应的。