Short Acquisition Time PET/MR Pharmacokinetic Modelling Using CNNs

Catherine J. Scott; Jieqing Jiao; M. Jorge Cardoso; Kerstin Kläser; Andrew Melbourne; Pawel J. Markiewicz; Jonathan M. Schott; Brian F. Hutton; Sébastien Ourselin

doi:10.1007/978-3-030-00928-1_6

Short Acquisition Time PET/MR Pharmacokinetic Modelling Using CNNs

Catherine J. Scott^*, Jieqing Jiao, M. Jorge Cardoso, Kerstin Kläser, Andrew Melbourne, Pawel J. Markiewicz, Jonathan M. Schott, Brian F. Hutton, Sébastien Ourselin

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference paper › peer-review

4 Citations (Scopus)

Abstract

Standard quantification of Positron Emission Tomography (PET) data requires a long acquisition time to enable pharmacokinetic (PK) model fitting, however blood flow information from Arterial Spin Labelling (ASL) Magnetic Resonance Imaging (MRI) can be combined with simultaneous dynamic PET data to reduce the acquisition time. Due the difficulty of fitting a PK model to noisy PET data with limited time points, such ‘fixed- R ₁ ’ techniques are constrained to a 30 min minimum acquisition, which is intolerable for many patients. In this work we apply a deep convolutional neural network (CNN) approach to combine the PET and MRI data. This permits shorter acquisition times as it avoids the noise sensitive voxelwise PK modelling and facilitates the full modelling of the relationship between blood flow and the dynamic PET data. This method is compared to three fixed- R ₁ PK methods, and the clinically used standardised uptake value ratio (SUVR), using 60 min dynamic PET PK modelling as the gold standard. Testing on 11 subjects participating in a study of pre-clinical Alzheimer’s Disease showed that, for 30 min acquisitions, all methods which combine the PET and MRI data have comparable performance, however at shorter acquisition times the CNN approach has a significantly lower mean square error (MSE) compared to fixed- R ₁ PK modelling (p=0.001). For both acquisition windows, SUVR had a significantly higher MSE than the CNN method (p ࣘ 0.003). This demonstrates that combining simultaneous PET and MRI data using a CNN can result in robust PET quantification within a scan time which is tolerable to patients with dementia.

Original language	English
Title of host publication	Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings
Editors	Julia A. Schnabel, Christos Davatzikos, Carlos Alberola-López, Gabor Fichtinger, Alejandro F. Frangi
Publisher	Springer Verlag
Pages	48-56
Number of pages	9
Volume	11070
ISBN (Print)	978-3-030-00927-4
DOIs	https://doi.org/10.1007/978-3-030-00928-1_6
Publication status	Published - 2018
Event	21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018 - Granada, Spain Duration: 16 Sept 2018 → 20 Sept 2018

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11070 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018
Country/Territory	Spain
City	Granada
Period	16/09/2018 → 20/09/2018

Access to Document

10.1007/978-3-030-00928-1_6

Cite this

Scott, C. J., Jiao, J., Cardoso, M. J., Kläser, K., Melbourne, A., Markiewicz, P. J., Schott, J. M., Hutton, B. F., & Ourselin, S. (2018). Short Acquisition Time PET/MR Pharmacokinetic Modelling Using CNNs. In J. A. Schnabel, C. Davatzikos, C. Alberola-López, G. Fichtinger, & A. F. Frangi (Eds.), Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings (Vol. 11070, pp. 48-56). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11070 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-00928-1_6

Scott, Catherine J. ; Jiao, Jieqing ; Cardoso, M. Jorge et al. / Short Acquisition Time PET/MR Pharmacokinetic Modelling Using CNNs. Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings. editor / Julia A. Schnabel ; Christos Davatzikos ; Carlos Alberola-López ; Gabor Fichtinger ; Alejandro F. Frangi. Vol. 11070 Springer Verlag, 2018. pp. 48-56 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inbook{664e5b53cae7464da92a64ae8124a337,

title = "Short Acquisition Time PET/MR Pharmacokinetic Modelling Using CNNs",

abstract = " Standard quantification of Positron Emission Tomography (PET) data requires a long acquisition time to enable pharmacokinetic (PK) model fitting, however blood flow information from Arterial Spin Labelling (ASL) Magnetic Resonance Imaging (MRI) can be combined with simultaneous dynamic PET data to reduce the acquisition time. Due the difficulty of fitting a PK model to noisy PET data with limited time points, such {\textquoteleft}fixed- R 1 {\textquoteright} techniques are constrained to a 30 min minimum acquisition, which is intolerable for many patients. In this work we apply a deep convolutional neural network (CNN) approach to combine the PET and MRI data. This permits shorter acquisition times as it avoids the noise sensitive voxelwise PK modelling and facilitates the full modelling of the relationship between blood flow and the dynamic PET data. This method is compared to three fixed- R 1 PK methods, and the clinically used standardised uptake value ratio (SUVR), using 60 min dynamic PET PK modelling as the gold standard. Testing on 11 subjects participating in a study of pre-clinical Alzheimer{\textquoteright}s Disease showed that, for 30 min acquisitions, all methods which combine the PET and MRI data have comparable performance, however at shorter acquisition times the CNN approach has a significantly lower mean square error (MSE) compared to fixed- R 1 PK modelling (p=0.001). For both acquisition windows, SUVR had a significantly higher MSE than the CNN method (p ࣘ 0.003). This demonstrates that combining simultaneous PET and MRI data using a CNN can result in robust PET quantification within a scan time which is tolerable to patients with dementia. ",

author = "Scott, {Catherine J.} and Jieqing Jiao and Cardoso, {M. Jorge} and Kerstin Kl{\"a}ser and Andrew Melbourne and Markiewicz, {Pawel J.} and Schott, {Jonathan M.} and Hutton, {Brian F.} and S{\'e}bastien Ourselin",

year = "2018",

doi = "10.1007/978-3-030-00928-1_6",

language = "English",

isbn = "978-3-030-00927-4",

volume = "11070",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "48--56",

editor = "Schnabel, {Julia A.} and Christos Davatzikos and Carlos Alberola-L{\'o}pez and Gabor Fichtinger and Frangi, {Alejandro F.}",

booktitle = "Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings",

address = "Germany",

note = "21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018 ; Conference date: 16-09-2018 Through 20-09-2018",

}

Scott, CJ, Jiao, J, Cardoso, MJ, Kläser, K, Melbourne, A, Markiewicz, PJ, Schott, JM, Hutton, BF & Ourselin, S 2018, Short Acquisition Time PET/MR Pharmacokinetic Modelling Using CNNs. in JA Schnabel, C Davatzikos, C Alberola-López, G Fichtinger & AF Frangi (eds), Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings. vol. 11070, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11070 LNCS, Springer Verlag, pp. 48-56, 21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018, Granada, Spain, 16/09/2018. https://doi.org/10.1007/978-3-030-00928-1_6

Short Acquisition Time PET/MR Pharmacokinetic Modelling Using CNNs. / Scott, Catherine J.; Jiao, Jieqing; Cardoso, M. Jorge et al.
Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings. ed. / Julia A. Schnabel; Christos Davatzikos; Carlos Alberola-López; Gabor Fichtinger; Alejandro F. Frangi. Vol. 11070 Springer Verlag, 2018. p. 48-56 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11070 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference paper › peer-review

TY - CHAP

T1 - Short Acquisition Time PET/MR Pharmacokinetic Modelling Using CNNs

AU - Scott, Catherine J.

AU - Jiao, Jieqing

AU - Cardoso, M. Jorge

AU - Kläser, Kerstin

AU - Melbourne, Andrew

AU - Markiewicz, Pawel J.

AU - Schott, Jonathan M.

AU - Hutton, Brian F.

AU - Ourselin, Sébastien

PY - 2018

Y1 - 2018

N2 - Standard quantification of Positron Emission Tomography (PET) data requires a long acquisition time to enable pharmacokinetic (PK) model fitting, however blood flow information from Arterial Spin Labelling (ASL) Magnetic Resonance Imaging (MRI) can be combined with simultaneous dynamic PET data to reduce the acquisition time. Due the difficulty of fitting a PK model to noisy PET data with limited time points, such ‘fixed- R 1 ’ techniques are constrained to a 30 min minimum acquisition, which is intolerable for many patients. In this work we apply a deep convolutional neural network (CNN) approach to combine the PET and MRI data. This permits shorter acquisition times as it avoids the noise sensitive voxelwise PK modelling and facilitates the full modelling of the relationship between blood flow and the dynamic PET data. This method is compared to three fixed- R 1 PK methods, and the clinically used standardised uptake value ratio (SUVR), using 60 min dynamic PET PK modelling as the gold standard. Testing on 11 subjects participating in a study of pre-clinical Alzheimer’s Disease showed that, for 30 min acquisitions, all methods which combine the PET and MRI data have comparable performance, however at shorter acquisition times the CNN approach has a significantly lower mean square error (MSE) compared to fixed- R 1 PK modelling (p=0.001). For both acquisition windows, SUVR had a significantly higher MSE than the CNN method (p ࣘ 0.003). This demonstrates that combining simultaneous PET and MRI data using a CNN can result in robust PET quantification within a scan time which is tolerable to patients with dementia.

AB - Standard quantification of Positron Emission Tomography (PET) data requires a long acquisition time to enable pharmacokinetic (PK) model fitting, however blood flow information from Arterial Spin Labelling (ASL) Magnetic Resonance Imaging (MRI) can be combined with simultaneous dynamic PET data to reduce the acquisition time. Due the difficulty of fitting a PK model to noisy PET data with limited time points, such ‘fixed- R 1 ’ techniques are constrained to a 30 min minimum acquisition, which is intolerable for many patients. In this work we apply a deep convolutional neural network (CNN) approach to combine the PET and MRI data. This permits shorter acquisition times as it avoids the noise sensitive voxelwise PK modelling and facilitates the full modelling of the relationship between blood flow and the dynamic PET data. This method is compared to three fixed- R 1 PK methods, and the clinically used standardised uptake value ratio (SUVR), using 60 min dynamic PET PK modelling as the gold standard. Testing on 11 subjects participating in a study of pre-clinical Alzheimer’s Disease showed that, for 30 min acquisitions, all methods which combine the PET and MRI data have comparable performance, however at shorter acquisition times the CNN approach has a significantly lower mean square error (MSE) compared to fixed- R 1 PK modelling (p=0.001). For both acquisition windows, SUVR had a significantly higher MSE than the CNN method (p ࣘ 0.003). This demonstrates that combining simultaneous PET and MRI data using a CNN can result in robust PET quantification within a scan time which is tolerable to patients with dementia.

UR - http://www.scopus.com/inward/record.url?scp=85054075789&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-00928-1_6

DO - 10.1007/978-3-030-00928-1_6

M3 - Conference paper

AN - SCOPUS:85054075789

SN - 978-3-030-00927-4

VL - 11070

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 48

EP - 56

BT - Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings

A2 - Schnabel, Julia A.

A2 - Davatzikos, Christos

A2 - Alberola-López, Carlos

A2 - Fichtinger, Gabor

A2 - Frangi, Alejandro F.

PB - Springer Verlag

T2 - 21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018

Y2 - 16 September 2018 through 20 September 2018

ER -

Scott CJ, Jiao J, Cardoso MJ, Kläser K, Melbourne A, Markiewicz PJ et al. Short Acquisition Time PET/MR Pharmacokinetic Modelling Using CNNs. In Schnabel JA, Davatzikos C, Alberola-López C, Fichtinger G, Frangi AF, editors, Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings. Vol. 11070. Springer Verlag. 2018. p. 48-56. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-00928-1_6

Short Acquisition Time PET/MR Pharmacokinetic Modelling Using CNNs

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this