東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back to Search results for [ null ]

Switch To: Labeled | MARC Mode | ISBD

Complex-Valued Deep Learning With Ap...

Virtue, Patrick M.

Linked to FindBook

Google Book

Amazon

博客來

Complex-Valued Deep Learning With Applications to Magnetic Resonance Image Synthesis.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Complex-Valued Deep Learning With Applications to Magnetic Resonance Image Synthesis./
Author:	Virtue, Patrick M.
Published:	Ann Arbor : ProQuest Dissertations & Theses, : 2019,
Description:	153 p.
Notes:	Source: Dissertations Abstracts International, Volume: 81-05, Section: B.
Contained By:	Dissertations Abstracts International81-05B.
Subject:	Computer science. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=22619704
ISBN:	9781392585603

Complex-Valued Deep Learning With Applications to Magnetic Resonance Image Synthesis.
Virtue, Patrick M.

Complex-Valued Deep Learning With Applications to Magnetic Resonance Image Synthesis. - Ann Arbor : ProQuest Dissertations & Theses, 2019 - 153 p.

Source: Dissertations Abstracts International, Volume: 81-05, Section: B.

Thesis (Ph.D.)--University of California, Berkeley, 2019.

This item must not be sold to any third party vendors.

Magnetic resonance imaging (MRI) has the ability to produce a series of images that each have different visual contrast between tissues, allowing clinicians to qualitatively assess pathologies that may be visible in one contrast-weighted image but not others. Unfortunately, these standard contrast-weighted images do not contain quantitative values, producing challenges for post-processing, assessment, and longitudinal studies. MR fingerprinting is a recent technique that produces quantitative tissue maps from a single pseudorandom acquisition, but it relies on computationally heavy nearest neighbor algorithms to solve the associated nonlinear inverse problem. In this dissertation, we present our deep learning methods to speed up quantitative MR fingerprinting and synthesize the standard contrast-weighted images directly from the same MR fingerprinting scan.Adapting deep learning methodologies to MR image synthesis presents two specific challenges: 1) complex-valued data and 2) the presence of noise while undersampling.MRI signals are inherently complex-valued, as they are measurements of rotating magnetization within the body. However, modern neural networks are not designed to support complex values. As an example, the pervasive ReLU activation function is undefined for complex numbers. This limitation curtails the impact of deep learning for complex data applications, such as MRI, radio frequency modulation identification, and target recognition in synthetic-aperture radar images. In this dissertation, we discuss the motivation for complex-valued networks, the changes that we have made to implement complex backpropagation, and our new complex cardioid activation function that made it possible to outperform real-valued networks for MR fingerprinting image synthesis.In Fourier-based medical imaging, undersampling results in an underdetermined system, in which a linear reconstruction will exhibit artifacts. Another consequence is lower signal-to-noise ratio (SNR) because of fewer acquired measurements. The coupled effects of low SNR and underdetermined system during reconstruction makes it difficult to model the signal and analyze image reconstruction algorithms. We demonstrate that neural networks trained only with a Gaussian noise model fail to process in vivo MR fingerprinting data, while our proposed empirical noise model allows neural networks to successfully synthesize quantitative images. Additionally, to better understand the impact of noise on undersampled imaging systems, we present an image quality prediction process that reconstructs fully sampled, fully determined data with noise added to simulate the SNR loss induced by a given undersampling pattern. The resulting prediction image empirically shows the effects of noise in undersampled image reconstruction without any effect from an underdetermined system, allowing MR pulse sequence and reconstruction developers to determine if low SNR, rather than the underdetermined system, is the limiting factor for a successful reconstruction.

ISBN: 9781392585603Subjects--Topical Terms:

523869
Computer science.
Subjects--Index Terms:

Complex values

Complex-Valued Deep Learning With Applications to Magnetic Resonance Image Synthesis.
LDR:04248nmm a2200373 4500 001 2268753
005 20200824072444.5
008 220629s2019 ||||||||||||||||| ||eng d
020 $a 9781392585603
035 $a (MiAaPQ)AAI22619704
035 $a AAI22619704
040 $a MiAaPQ $c MiAaPQ
100 1 $a Virtue, Patrick M. $3 3546053
245 1 0 $a Complex-Valued Deep Learning With Applications to Magnetic Resonance Image Synthesis.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2019
300 $a 153 p.
500 $a Source: Dissertations Abstracts International, Volume: 81-05, Section: B.
500 $a Advisor: Lustig, Michael;Yu, Stella X.
502 $a Thesis (Ph.D.)--University of California, Berkeley, 2019.
506 $a This item must not be sold to any third party vendors.
520 $a Magnetic resonance imaging (MRI) has the ability to produce a series of images that each have different visual contrast between tissues, allowing clinicians to qualitatively assess pathologies that may be visible in one contrast-weighted image but not others. Unfortunately, these standard contrast-weighted images do not contain quantitative values, producing challenges for post-processing, assessment, and longitudinal studies. MR fingerprinting is a recent technique that produces quantitative tissue maps from a single pseudorandom acquisition, but it relies on computationally heavy nearest neighbor algorithms to solve the associated nonlinear inverse problem. In this dissertation, we present our deep learning methods to speed up quantitative MR fingerprinting and synthesize the standard contrast-weighted images directly from the same MR fingerprinting scan.Adapting deep learning methodologies to MR image synthesis presents two specific challenges: 1) complex-valued data and 2) the presence of noise while undersampling.MRI signals are inherently complex-valued, as they are measurements of rotating magnetization within the body. However, modern neural networks are not designed to support complex values. As an example, the pervasive ReLU activation function is undefined for complex numbers. This limitation curtails the impact of deep learning for complex data applications, such as MRI, radio frequency modulation identification, and target recognition in synthetic-aperture radar images. In this dissertation, we discuss the motivation for complex-valued networks, the changes that we have made to implement complex backpropagation, and our new complex cardioid activation function that made it possible to outperform real-valued networks for MR fingerprinting image synthesis.In Fourier-based medical imaging, undersampling results in an underdetermined system, in which a linear reconstruction will exhibit artifacts. Another consequence is lower signal-to-noise ratio (SNR) because of fewer acquired measurements. The coupled effects of low SNR and underdetermined system during reconstruction makes it difficult to model the signal and analyze image reconstruction algorithms. We demonstrate that neural networks trained only with a Gaussian noise model fail to process in vivo MR fingerprinting data, while our proposed empirical noise model allows neural networks to successfully synthesize quantitative images. Additionally, to better understand the impact of noise on undersampled imaging systems, we present an image quality prediction process that reconstructs fully sampled, fully determined data with noise added to simulate the SNR loss induced by a given undersampling pattern. The resulting prediction image empirically shows the effects of noise in undersampled image reconstruction without any effect from an underdetermined system, allowing MR pulse sequence and reconstruction developers to determine if low SNR, rather than the underdetermined system, is the limiting factor for a successful reconstruction.
590 $a School code: 0028.
650 4 $a Computer science. $3 523869
650 4 $a Medical imaging. $3 3172799
650 4 $a Electrical engineering. $3 649834
653 $a Complex values
653 $a Deep learning
653 $a Image processing
653 $a Magnetic resonance imaging
653 $a Neural networks
690 $a 0984
690 $a 0574
690 $a 0544
710 2 $a University of California, Berkeley. $b Electrical Engineering & Computer Sciences. $3 1671057
773 0 $t Dissertations Abstracts International $g 81-05B.
790 $a 0028
791 $a Ph.D.
792 $a 2019
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=22619704