TY - JOUR
T1 - Adopting transfer learning for neuroimaging
T2 - a comparative analysis with a custom 3D convolution neural network model
AU - Soliman, Amira
AU - Chang, Jose R.
AU - Etminani, Kobra
AU - Byttner, Stefan
AU - Davidsson, Anette
AU - Martínez-Sanchis, Begoña
AU - Camacho, Valle
AU - Bauckneht, Matteo
AU - Stegeran, Roxana
AU - Ressner, Marcus
AU - Agudelo-Cifuentes, Marc
AU - Chincarini, Andrea
AU - Brendel, Matthias
AU - Rominger, Axel
AU - Bruffaerts, Rose
AU - Vandenberghe, Rik
AU - Kramberger, Milica G.
AU - Trost, Maja
AU - Nicastro, Nicolas
AU - Frisoni, Giovanni B.
AU - Lemstra, Afina W.
AU - Berckel, Bart N. M. van
AU - Pilotto, Andrea
AU - Padovani, Alessandro
AU - Morbelli, Silvia
AU - the Alzheimer’s Disease Neuroimaging Initiative
AU - Aarsland, Dag
AU - Nobili, Flavio
AU - Garibotto, Valentina
AU - Ochoa-Figueroa, Miguel
N1 - Funding Information: This research study was conducted retrospectively using data obtained from European DLB (EDLB) Consortium. Local institutional ethics committee approvals for the retrospective analyses were available for all centers in Europe, including the transfer of fully anonymized imaging data. Regarding the data from Linköping’s University Hospital, informed consent was waived for this retrospective assessment and additionally, all patients were informed by letter that their medical data can be rendered anonymous and used for scientific purposes. All patients from the rest of the centers gave informed written consent for the imaging procedure and radiopharmaceutical application. The study has approval by the Swedish Ethical Review Authority (Etikprövningsmyndigheten) with approval number: 2019-00526. Part of data collection and sharing for this project was funded by the Alzheimer’s Disease Neuroimaging Initiative (ADNI) (National Institutes of Health Grant U01 AG024904) and DOD ADNI (Department of Defense award number W81XWH-12-2-0012) Funding Information: This study was part of a collaborative project between Center for Applied Intelligent System Research (CAISR) at Halmstad University, Sweden, and Department of Clinical Physiology, Department of Radiology and the Center for Medical Imaging Visualization (CMIV) at Linköping University Hospital, Sweden, and the European DLB consortium, which was funded by Analytic Imaging Diagnostics Arena (AIDA) initiative, jointly supported by VINNOVA (Grant 2017-02447), Formas and the Swedish Energy Agency. VG was supported by the Swiss National Science Foundation (projects 320030_169876, 320030_185028) and the Velux Foundation (project 1123). RB is a senior postdoctoral fellow of the Flanders Research Foundation (FWO 12I2121N). Besides financial support, AIDA organized multiple meetings to share knowledge among participating institutions. Publisher Copyright: © 2022, The Author(s).
PY - 2022/11/1
Y1 - 2022/11/1
N2 - Background: In recent years, neuroimaging with deep learning (DL) algorithms have made remarkable advances in the diagnosis of neurodegenerative disorders. However, applying DL in different medical domains is usually challenged by lack of labeled data. To address this challenge, transfer learning (TL) has been applied to use state-of-the-art convolution neural networks pre-trained on natural images. Yet, there are differences in characteristics between medical and natural images, also image classification and targeted medical diagnosis tasks. The purpose of this study is to investigate the performance of specialized and TL in the classification of neurodegenerative disorders using 3D volumes of 18F-FDG-PET brain scans. Results: Results show that TL models are suboptimal for classification of neurodegenerative disorders, especially when the objective is to separate more than two disorders. Additionally, specialized CNN model provides better interpretations of predicted diagnosis. Conclusions: TL can indeed lead to superior performance on binary classification in timely and data efficient manner, yet for detecting more than a single disorder, TL models do not perform well. Additionally, custom 3D model performs comparably to TL models for binary classification, and interestingly perform better for diagnosis of multiple disorders. The results confirm the superiority of the custom 3D-CNN in providing better explainable model compared to TL adopted ones.
AB - Background: In recent years, neuroimaging with deep learning (DL) algorithms have made remarkable advances in the diagnosis of neurodegenerative disorders. However, applying DL in different medical domains is usually challenged by lack of labeled data. To address this challenge, transfer learning (TL) has been applied to use state-of-the-art convolution neural networks pre-trained on natural images. Yet, there are differences in characteristics between medical and natural images, also image classification and targeted medical diagnosis tasks. The purpose of this study is to investigate the performance of specialized and TL in the classification of neurodegenerative disorders using 3D volumes of 18F-FDG-PET brain scans. Results: Results show that TL models are suboptimal for classification of neurodegenerative disorders, especially when the objective is to separate more than two disorders. Additionally, specialized CNN model provides better interpretations of predicted diagnosis. Conclusions: TL can indeed lead to superior performance on binary classification in timely and data efficient manner, yet for detecting more than a single disorder, TL models do not perform well. Additionally, custom 3D model performs comparably to TL models for binary classification, and interestingly perform better for diagnosis of multiple disorders. The results confirm the superiority of the custom 3D-CNN in providing better explainable model compared to TL adopted ones.
KW - Brain Neurodegenerative Disorders
KW - Convolution Neural Networks
KW - Medical Image Classification
KW - Transfer Learning
UR - http://www.scopus.com/inward/record.url?scp=85143570393&partnerID=8YFLogxK
U2 - https://doi.org/10.1186/s12911-022-02054-7
DO - https://doi.org/10.1186/s12911-022-02054-7
M3 - Article
C2 - 36476613
SN - 1472-6947
VL - 22
JO - BMC medical informatics and decision making
JF - BMC medical informatics and decision making
M1 - 318
ER -