LAPNet: non-rigid registration derived in k-space for magnetic resonance imaging

Thomas Kustner; Jiazhen Pan; Haikun Qi; Gastao Cruz; Christopher Gilliam; Thierry Blu; Bin Yang; Sergios Gatidis; Rene Botnar; Claudia Prieto

doi:10.1109/TMI.2021.3096131

LAPNet: non-rigid registration derived in k-space for magnetic resonance imaging

Thomas Kustner^*, Jiazhen Pan, Haikun Qi, Gastao Cruz, Christopher Gilliam, Thierry Blu, Bin Yang, Sergios Gatidis, Rene Botnar, Claudia Prieto

^*Corresponding author for this work

Electronic, Electrical and Systems Engineering

Research output: Contribution to journal › Article › peer-review

1 Citation (Scopus)

51 Downloads (Pure)

Abstract

Physiological motion, such as cardiac and respiratory motion, during Magnetic Resonance (MR) image acquisition can cause image artifacts. Motion correction techniques have been proposed to compensate for these types of motion during thoracic scans, relying on accurate motion estimation from undersampled motion-resolved reconstruction. A particular interest and challenge lie in the derivation of reliable non-rigid motion fields from the undersampled motion-resolved data. Motion estimation is usually formulated in image space via diffusion, parametric-spline, or optical flow methods. However, image-based registration can be impaired by remaining aliasing artifacts due to the undersampled motion-resolved reconstruction. In this work, we describe a formalism to perform non-rigid registration directly in the sampled Fourier space, i.e. k-space. We propose a deep-learning based approach to perform fast and accurate non-rigid registration from the undersampled k-space data. The basic working principle originates from the Local All-Pass (LAP) technique, a recently introduced optical flow-based registration. The proposed LAPNet is compared against traditional and deep learning image-based registrations and tested on fully-sampled and highly-accelerated (with two undersampling strategies) 3D respiratory motion-resolved MR images in a cohort of 40 patients with suspected liver or lung metastases and 25 healthy subjects. The proposed LAPNet provided consistent and superior performance to image-based approaches throughout different sampling trajectories and acceleration factors.

Original language	English
Pages (from-to)	3686-3697
Number of pages	12
Journal	IEEE Transactions on Medical Imaging
Volume	40
Issue number	12
Early online date	9 Jul 2021
DOIs	https://doi.org/10.1109/TMI.2021.3096131
Publication status	Published - Dec 2021

Bibliographical note

Funding Information:
Manuscript received May 31, 2021; accepted July 7, 2021. Date of publication July 9, 2021; date of current version November 30, 2021. This work was supported in part by the Deutsche Forschungsgemeinschaft (DFG) through the Germany’s Excellence Strategy by EXC 2180 under Grant 390900677 and EXC 2064/1 under Grant 390727645; in part by EPSRC under Grant EP/P032311/1, Grant EP/P001009/1, and Grant EP/P007619/1; and in part by the Wellcome EPSRC Centre for Medical Engineering under Grant NS/A000049/1. (Corresponding author: Thomas Küstner.) This work involved human subjects. Approval of all ethical and experimental procedures and protocols was granted by the Local Ethics Committee under Application No. 721/2012BO1, and performed in line with the Helsinki Declaration of 1964 and its later amendments, and the EU Good Clinical Practice Directive 2005/28/EC, the EU Clinical Trial Regulation No. 536/2014, and the EU Regulation No. 746/2017 on medical devices.

Publisher Copyright:
© 1982-2012 IEEE.

Keywords

deep learning registration
Magnetic resonance imaging
motion correction
non-rigid registration

ASJC Scopus subject areas

Software
Radiological and Ultrasound Technology
Computer Science Applications
Electrical and Electronic Engineering

Access to Document

10.1109/TMI.2021.3096131

KüstnerT2021LAPNet
T. Küstner et al., "LAPNet: Non-Rigid Registration Derived in k-Space for Magnetic Resonance Imaging," in IEEE Transactions on Medical Imaging, vol. 40, no. 12, pp. 3686-3697, Dec. 2021, doi: 10.1109/TMI.2021.3096131. © 2021 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
Accepted author manuscript, 2.9 MBLicence: Other (please specify with Rights Statement)

Cite this

@article{36020b4676eb4967ab9ee881f5177052,

title = "LAPNet: non-rigid registration derived in k-space for magnetic resonance imaging",

abstract = "Physiological motion, such as cardiac and respiratory motion, during Magnetic Resonance (MR) image acquisition can cause image artifacts. Motion correction techniques have been proposed to compensate for these types of motion during thoracic scans, relying on accurate motion estimation from undersampled motion-resolved reconstruction. A particular interest and challenge lie in the derivation of reliable non-rigid motion fields from the undersampled motion-resolved data. Motion estimation is usually formulated in image space via diffusion, parametric-spline, or optical flow methods. However, image-based registration can be impaired by remaining aliasing artifacts due to the undersampled motion-resolved reconstruction. In this work, we describe a formalism to perform non-rigid registration directly in the sampled Fourier space, i.e. k-space. We propose a deep-learning based approach to perform fast and accurate non-rigid registration from the undersampled k-space data. The basic working principle originates from the Local All-Pass (LAP) technique, a recently introduced optical flow-based registration. The proposed LAPNet is compared against traditional and deep learning image-based registrations and tested on fully-sampled and highly-accelerated (with two undersampling strategies) 3D respiratory motion-resolved MR images in a cohort of 40 patients with suspected liver or lung metastases and 25 healthy subjects. The proposed LAPNet provided consistent and superior performance to image-based approaches throughout different sampling trajectories and acceleration factors.",

keywords = "deep learning registration, Magnetic resonance imaging, motion correction, non-rigid registration",

author = "Thomas Kustner and Jiazhen Pan and Haikun Qi and Gastao Cruz and Christopher Gilliam and Thierry Blu and Bin Yang and Sergios Gatidis and Rene Botnar and Claudia Prieto",

note = "Funding Information: Manuscript received May 31, 2021; accepted July 7, 2021. Date of publication July 9, 2021; date of current version November 30, 2021. This work was supported in part by the Deutsche Forschungsgemeinschaft (DFG) through the Germany{\textquoteright}s Excellence Strategy by EXC 2180 under Grant 390900677 and EXC 2064/1 under Grant 390727645; in part by EPSRC under Grant EP/P032311/1, Grant EP/P001009/1, and Grant EP/P007619/1; and in part by the Wellcome EPSRC Centre for Medical Engineering under Grant NS/A000049/1. (Corresponding author: Thomas K{\"u}stner.) This work involved human subjects. Approval of all ethical and experimental procedures and protocols was granted by the Local Ethics Committee under Application No. 721/2012BO1, and performed in line with the Helsinki Declaration of 1964 and its later amendments, and the EU Good Clinical Practice Directive 2005/28/EC, the EU Clinical Trial Regulation No. 536/2014, and the EU Regulation No. 746/2017 on medical devices. Publisher Copyright: {\textcopyright} 1982-2012 IEEE.",

year = "2021",

month = dec,

doi = "10.1109/TMI.2021.3096131",

language = "English",

volume = "40",

pages = "3686--3697",

journal = "IEEE Transactions on Medical Imaging",

issn = "0278-0062",

publisher = "Institute of Electrical and Electronics Engineers (IEEE)",

number = "12",

}

TY - JOUR

T1 - LAPNet

T2 - non-rigid registration derived in k-space for magnetic resonance imaging

AU - Kustner, Thomas

AU - Pan, Jiazhen

AU - Qi, Haikun

AU - Cruz, Gastao

AU - Gilliam, Christopher

AU - Blu, Thierry

AU - Yang, Bin

AU - Gatidis, Sergios

AU - Botnar, Rene

AU - Prieto, Claudia

N1 - Funding Information: Manuscript received May 31, 2021; accepted July 7, 2021. Date of publication July 9, 2021; date of current version November 30, 2021. This work was supported in part by the Deutsche Forschungsgemeinschaft (DFG) through the Germany’s Excellence Strategy by EXC 2180 under Grant 390900677 and EXC 2064/1 under Grant 390727645; in part by EPSRC under Grant EP/P032311/1, Grant EP/P001009/1, and Grant EP/P007619/1; and in part by the Wellcome EPSRC Centre for Medical Engineering under Grant NS/A000049/1. (Corresponding author: Thomas Küstner.) This work involved human subjects. Approval of all ethical and experimental procedures and protocols was granted by the Local Ethics Committee under Application No. 721/2012BO1, and performed in line with the Helsinki Declaration of 1964 and its later amendments, and the EU Good Clinical Practice Directive 2005/28/EC, the EU Clinical Trial Regulation No. 536/2014, and the EU Regulation No. 746/2017 on medical devices. Publisher Copyright: © 1982-2012 IEEE.

PY - 2021/12

Y1 - 2021/12

N2 - Physiological motion, such as cardiac and respiratory motion, during Magnetic Resonance (MR) image acquisition can cause image artifacts. Motion correction techniques have been proposed to compensate for these types of motion during thoracic scans, relying on accurate motion estimation from undersampled motion-resolved reconstruction. A particular interest and challenge lie in the derivation of reliable non-rigid motion fields from the undersampled motion-resolved data. Motion estimation is usually formulated in image space via diffusion, parametric-spline, or optical flow methods. However, image-based registration can be impaired by remaining aliasing artifacts due to the undersampled motion-resolved reconstruction. In this work, we describe a formalism to perform non-rigid registration directly in the sampled Fourier space, i.e. k-space. We propose a deep-learning based approach to perform fast and accurate non-rigid registration from the undersampled k-space data. The basic working principle originates from the Local All-Pass (LAP) technique, a recently introduced optical flow-based registration. The proposed LAPNet is compared against traditional and deep learning image-based registrations and tested on fully-sampled and highly-accelerated (with two undersampling strategies) 3D respiratory motion-resolved MR images in a cohort of 40 patients with suspected liver or lung metastases and 25 healthy subjects. The proposed LAPNet provided consistent and superior performance to image-based approaches throughout different sampling trajectories and acceleration factors.

AB - Physiological motion, such as cardiac and respiratory motion, during Magnetic Resonance (MR) image acquisition can cause image artifacts. Motion correction techniques have been proposed to compensate for these types of motion during thoracic scans, relying on accurate motion estimation from undersampled motion-resolved reconstruction. A particular interest and challenge lie in the derivation of reliable non-rigid motion fields from the undersampled motion-resolved data. Motion estimation is usually formulated in image space via diffusion, parametric-spline, or optical flow methods. However, image-based registration can be impaired by remaining aliasing artifacts due to the undersampled motion-resolved reconstruction. In this work, we describe a formalism to perform non-rigid registration directly in the sampled Fourier space, i.e. k-space. We propose a deep-learning based approach to perform fast and accurate non-rigid registration from the undersampled k-space data. The basic working principle originates from the Local All-Pass (LAP) technique, a recently introduced optical flow-based registration. The proposed LAPNet is compared against traditional and deep learning image-based registrations and tested on fully-sampled and highly-accelerated (with two undersampling strategies) 3D respiratory motion-resolved MR images in a cohort of 40 patients with suspected liver or lung metastases and 25 healthy subjects. The proposed LAPNet provided consistent and superior performance to image-based approaches throughout different sampling trajectories and acceleration factors.

KW - deep learning registration

KW - Magnetic resonance imaging

KW - motion correction

KW - non-rigid registration

UR - http://www.scopus.com/inward/record.url?scp=85110908521&partnerID=8YFLogxK

U2 - 10.1109/TMI.2021.3096131

DO - 10.1109/TMI.2021.3096131

M3 - Article

C2 - 34242163

AN - SCOPUS:85110908521

SN - 0278-0062

VL - 40

SP - 3686

EP - 3697

JO - IEEE Transactions on Medical Imaging

JF - IEEE Transactions on Medical Imaging

IS - 12

ER -

LAPNet: non-rigid registration derived in k-space for magnetic resonance imaging

Abstract

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this