Automated Data Quality Control in FDOPA brain PET Imaging using Deep Learning

Antonella Pontoriero; Giovanna Nordio; Rubaida Easmin; Alessio Giacomel; Barbara Santangelo; Sameer Jauhar; Ilaria Bonoldi; Maria Rogdaki; Federico Turkheimer; Oliver Howes; Mattia Veronese

doi:10.1016/j.cmpb.2021.106239

Automated Data Quality Control in FDOPA brain PET Imaging using Deep Learning

Antonella Pontoriero, Giovanna Nordio, Rubaida Easmin, Alessio Giacomel, Barbara Santangelo, Sameer Jauhar, Ilaria Bonoldi, Maria Rogdaki, Federico Turkheimer, Oliver Howes, Mattia Veronese

School of Biomedical Engineering and Imaging Sciences

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

19 Citations (Scopus)

Abstract

INTRODUCTION: With biomedical imaging research increasingly using large datasets, it becomes critical to find operator-free methods to quality control the data collected and the associated analysis. Attempts to use artificial intelligence (AI) to perform automated quality control (QC) for both single-site and multi-site datasets have been explored in some neuroimaging techniques (e.g. EEG or MRI), although these methods struggle to find replication in other domains. The aim of this study is to test the feasibility of an automated QC pipeline for brain [18F]-FDOPA PET imaging as a biomarker for the dopamine system.

METHODS: Two different Convolutional Neural Networks (CNNs) were used and combined to assess spatial misalignment to a standard template and the signal-to-noise ratio (SNR) relative to 200 static [18F]-FDOPA PET images that had been manually quality controlled from three different PET/CT scanners. The scans were combined with an additional 400 scans, in which misalignment (200 scans) and low SNR (200 scans) were simulated. A cross-validation was performed, where 80% of the data were used for training and 20% for validation. Two additional datasets of [18F]-FDOPA PET images (50 and 100 scans respectively with at least 80% of good quality images) were used for out-of-sample validation.

RESULTS: The CNN performance was excellent in the training dataset (accuracy for motion: 0.86 ± 0.01, accuracy for SNR: 0.69 ± 0.01), leading to 100% accurate QC classification when applied to the two out-of-sample datasets. Data dimensionality reduction affected the generalizability of the CNNs, especially when the classifiers were applied to the out-of-sample data from 3D to 1D datasets.

CONCLUSIONS: This feasibility study shows that it is possible to perform automatic QC of [18F]-FDOPA PET imaging with CNNs. The approach has the potential to be extended to other PET tracers in both brain and non-brain applications, but it is dependent on the availability of large datasets necessary for the algorithm training.

Original language	English
Article number	106239
Pages (from-to)	106239
Journal	Computer Methods and Programs in Biomedicine
Volume	208
Early online date	22 Jun 2021
DOIs	https://doi.org/10.1016/j.cmpb.2021.106239
Publication status	Published - Sept 2021

Access to Document

10.1016/j.cmpb.2021.106239

Cite this

@article{6b929c1c257f4875a8c914b1c1ff56ac,

title = "Automated Data Quality Control in FDOPA brain PET Imaging using Deep Learning",

abstract = "INTRODUCTION: With biomedical imaging research increasingly using large datasets, it becomes critical to find operator-free methods to quality control the data collected and the associated analysis. Attempts to use artificial intelligence (AI) to perform automated quality control (QC) for both single-site and multi-site datasets have been explored in some neuroimaging techniques (e.g. EEG or MRI), although these methods struggle to find replication in other domains. The aim of this study is to test the feasibility of an automated QC pipeline for brain [18F]-FDOPA PET imaging as a biomarker for the dopamine system.METHODS: Two different Convolutional Neural Networks (CNNs) were used and combined to assess spatial misalignment to a standard template and the signal-to-noise ratio (SNR) relative to 200 static [18F]-FDOPA PET images that had been manually quality controlled from three different PET/CT scanners. The scans were combined with an additional 400 scans, in which misalignment (200 scans) and low SNR (200 scans) were simulated. A cross-validation was performed, where 80% of the data were used for training and 20% for validation. Two additional datasets of [18F]-FDOPA PET images (50 and 100 scans respectively with at least 80% of good quality images) were used for out-of-sample validation.RESULTS: The CNN performance was excellent in the training dataset (accuracy for motion: 0.86 ± 0.01, accuracy for SNR: 0.69 ± 0.01), leading to 100% accurate QC classification when applied to the two out-of-sample datasets. Data dimensionality reduction affected the generalizability of the CNNs, especially when the classifiers were applied to the out-of-sample data from 3D to 1D datasets.CONCLUSIONS: This feasibility study shows that it is possible to perform automatic QC of [18F]-FDOPA PET imaging with CNNs. The approach has the potential to be extended to other PET tracers in both brain and non-brain applications, but it is dependent on the availability of large datasets necessary for the algorithm training.",

author = "Antonella Pontoriero and Giovanna Nordio and Rubaida Easmin and Alessio Giacomel and Barbara Santangelo and Sameer Jauhar and Ilaria Bonoldi and Maria Rogdaki and Federico Turkheimer and Oliver Howes and Mattia Veronese",

note = "Funding Information: Dr Veronese is funded by the National Institute for Health Research Biomedical Research Centre at South London and Maudsley National Health Service Foundation Trust and King's College London, and by the Wellcome Trust Digital Award 215747/Z/19/Z. Dr Howes is funded by Medical Research Council grant MC- A656-5QD30, Maudsley Charity grant 666, support from the US Brain & Behavior Research Foundation, and Wellcome Trust grant 094849/Z/10/Z to Dr Howes and the National Institute for Health Research Biomedical Research Centre at South London and Maudsley National Health Service Foundation Trust and King's College London. Dr Jauhar is funded by the National Institute for Health Research Biomedical Research Centre at South London and Maudsley National Health Service Foundation Trust and King's College London, and a JMAS SIM Fellowship from the Royal College of Physicians, Edinburgh. Dr Bonoldi is supported by the National Institute for Health Research Biomedical Research Centre at South London, Maudsley National Health Service Foundation Trust, and King's College London. Funding Information: This study was funded by Medical Research Council-UK (no. MC_U120097115), Maudsley Charity (no. 666), Brain and Behavior Research Foundation, and Wellcome Trust (no. 094849/Z/10/Z) grants to Dr Howes and the National Institute for Health Research (NIHR) Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King's College London. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health. Publisher Copyright: {\textcopyright} 2021 Copyright: Copyright 2021 Elsevier B.V., All rights reserved.",

year = "2021",

month = sep,

doi = "10.1016/j.cmpb.2021.106239",

language = "English",

volume = "208",

pages = "106239",

journal = "Computer Methods and Programs in Biomedicine",

issn = "0169-2607",

publisher = "Elsevier Ireland Ltd",

}

TY - JOUR

T1 - Automated Data Quality Control in FDOPA brain PET Imaging using Deep Learning

AU - Pontoriero, Antonella

AU - Nordio, Giovanna

AU - Easmin, Rubaida

AU - Giacomel, Alessio

AU - Santangelo, Barbara

AU - Jauhar, Sameer

AU - Bonoldi, Ilaria

AU - Rogdaki, Maria

AU - Turkheimer, Federico

AU - Howes, Oliver

AU - Veronese, Mattia

N1 - Funding Information: Dr Veronese is funded by the National Institute for Health Research Biomedical Research Centre at South London and Maudsley National Health Service Foundation Trust and King's College London, and by the Wellcome Trust Digital Award 215747/Z/19/Z. Dr Howes is funded by Medical Research Council grant MC- A656-5QD30, Maudsley Charity grant 666, support from the US Brain & Behavior Research Foundation, and Wellcome Trust grant 094849/Z/10/Z to Dr Howes and the National Institute for Health Research Biomedical Research Centre at South London and Maudsley National Health Service Foundation Trust and King's College London. Dr Jauhar is funded by the National Institute for Health Research Biomedical Research Centre at South London and Maudsley National Health Service Foundation Trust and King's College London, and a JMAS SIM Fellowship from the Royal College of Physicians, Edinburgh. Dr Bonoldi is supported by the National Institute for Health Research Biomedical Research Centre at South London, Maudsley National Health Service Foundation Trust, and King's College London. Funding Information: This study was funded by Medical Research Council-UK (no. MC_U120097115), Maudsley Charity (no. 666), Brain and Behavior Research Foundation, and Wellcome Trust (no. 094849/Z/10/Z) grants to Dr Howes and the National Institute for Health Research (NIHR) Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King's College London. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health. Publisher Copyright: © 2021 Copyright: Copyright 2021 Elsevier B.V., All rights reserved.

PY - 2021/9

Y1 - 2021/9

N2 - INTRODUCTION: With biomedical imaging research increasingly using large datasets, it becomes critical to find operator-free methods to quality control the data collected and the associated analysis. Attempts to use artificial intelligence (AI) to perform automated quality control (QC) for both single-site and multi-site datasets have been explored in some neuroimaging techniques (e.g. EEG or MRI), although these methods struggle to find replication in other domains. The aim of this study is to test the feasibility of an automated QC pipeline for brain [18F]-FDOPA PET imaging as a biomarker for the dopamine system.METHODS: Two different Convolutional Neural Networks (CNNs) were used and combined to assess spatial misalignment to a standard template and the signal-to-noise ratio (SNR) relative to 200 static [18F]-FDOPA PET images that had been manually quality controlled from three different PET/CT scanners. The scans were combined with an additional 400 scans, in which misalignment (200 scans) and low SNR (200 scans) were simulated. A cross-validation was performed, where 80% of the data were used for training and 20% for validation. Two additional datasets of [18F]-FDOPA PET images (50 and 100 scans respectively with at least 80% of good quality images) were used for out-of-sample validation.RESULTS: The CNN performance was excellent in the training dataset (accuracy for motion: 0.86 ± 0.01, accuracy for SNR: 0.69 ± 0.01), leading to 100% accurate QC classification when applied to the two out-of-sample datasets. Data dimensionality reduction affected the generalizability of the CNNs, especially when the classifiers were applied to the out-of-sample data from 3D to 1D datasets.CONCLUSIONS: This feasibility study shows that it is possible to perform automatic QC of [18F]-FDOPA PET imaging with CNNs. The approach has the potential to be extended to other PET tracers in both brain and non-brain applications, but it is dependent on the availability of large datasets necessary for the algorithm training.

AB - INTRODUCTION: With biomedical imaging research increasingly using large datasets, it becomes critical to find operator-free methods to quality control the data collected and the associated analysis. Attempts to use artificial intelligence (AI) to perform automated quality control (QC) for both single-site and multi-site datasets have been explored in some neuroimaging techniques (e.g. EEG or MRI), although these methods struggle to find replication in other domains. The aim of this study is to test the feasibility of an automated QC pipeline for brain [18F]-FDOPA PET imaging as a biomarker for the dopamine system.METHODS: Two different Convolutional Neural Networks (CNNs) were used and combined to assess spatial misalignment to a standard template and the signal-to-noise ratio (SNR) relative to 200 static [18F]-FDOPA PET images that had been manually quality controlled from three different PET/CT scanners. The scans were combined with an additional 400 scans, in which misalignment (200 scans) and low SNR (200 scans) were simulated. A cross-validation was performed, where 80% of the data were used for training and 20% for validation. Two additional datasets of [18F]-FDOPA PET images (50 and 100 scans respectively with at least 80% of good quality images) were used for out-of-sample validation.RESULTS: The CNN performance was excellent in the training dataset (accuracy for motion: 0.86 ± 0.01, accuracy for SNR: 0.69 ± 0.01), leading to 100% accurate QC classification when applied to the two out-of-sample datasets. Data dimensionality reduction affected the generalizability of the CNNs, especially when the classifiers were applied to the out-of-sample data from 3D to 1D datasets.CONCLUSIONS: This feasibility study shows that it is possible to perform automatic QC of [18F]-FDOPA PET imaging with CNNs. The approach has the potential to be extended to other PET tracers in both brain and non-brain applications, but it is dependent on the availability of large datasets necessary for the algorithm training.

UR - http://www.scopus.com/inward/record.url?scp=85110441377&partnerID=8YFLogxK

U2 - 10.1016/j.cmpb.2021.106239

DO - 10.1016/j.cmpb.2021.106239

M3 - Article

C2 - 34289438

SN - 0169-2607

VL - 208

SP - 106239

JO - Computer Methods and Programs in Biomedicine

JF - Computer Methods and Programs in Biomedicine

M1 - 106239

ER -

Automated Data Quality Control in FDOPA brain PET Imaging using Deep Learning

Abstract

Access to Document

Other files and links

Fingerprint

Cite this