Adopting transfer learning for neuroimaging: a comparative analysis with a custom 3D convolution neural network model

Autor: Soliman, Amira, Chang, Jose R, Etminani, Kobra, Byttner, Stefan, Davidsson, Anette, Martínez-Sanchis, Begoña, Camacho, Valle, Bauckneht, Matteo, Stegeran, Roxana, Ressner, Marcus, Agudelo-Cifuentes, Marc, Chincarini, Andrea, Brendel, Matthias, Rominger, Axel, Bruffaerts, Rose, Vandenberghe, Rik, Kramberger, Milica G, Trost, Maja, Nicastro, Nicolas, Frisoni, Giovanni B, Lemstra, Afina W, Berckel, Bart N M van, Pilotto, Andrea, Padovani, Alessandro, Morbelli, Silvia, Aarsland, Dag, Nobili, Flavio, Garibotto, Valentina, Ochoa-Figueroa, Miguel
Přispěvatelé: Neurology, Amsterdam Neuroscience - Neurodegeneration, Alzheimer’s Disease Neuroimaging Initiative
Jazyk: angličtina
Rok vydání: 2022
Předmět:
Zdroj: BMC Medical Informatics and Decision Making, 22:318. BioMed Central
Soliman, Amira; Chang, Jose R; Etminani, Kobra; Byttner, Stefan; Davidsson, Anette; Martínez-Sanchis, Begoña; Camacho, Valle; Bauckneht, Matteo; Stegeran, Roxana; Ressner, Marcus; Agudelo-Cifuentes, Marc; Chincarini, Andrea; Brendel, Matthias; Rominger, Axel; Bruffaerts, Rose; Vandenberghe, Rik; Kramberger, Milica G; Trost, Maja; Nicastro, Nicolas; Frisoni, Giovanni B; ... (2022). Adopting transfer learning for neuroimaging: a comparative analysis with a custom 3D convolution neural network model. BMC medical informatics and decision making, 22(Suppl 6), p. 318. BioMed Central 10.1186/s12911-022-02054-7
BMC medical informatics and decision making
the Alzheimer’s Disease Neuroimaging Initiative 2022, ' Adopting transfer learning for neuroimaging : a comparative analysis with a custom 3D convolution neural network model ', BMC Medical Informatics and Decision Making, vol. 22, 318 . https://doi.org/10.1186/s12911-022-02054-7
ISSN: 1472-6947
DOI: 10.1186/s12911-022-02054-7
Popis: 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. Funding Agencies|Center for Applied Intelligent System Research (CAISR) at Halmstad University, Sweden; Department of Clinical Physiology, Department of Radiology; Center for Medical Imaging Visualization (CMIV) at Linkoeping University Hospital, Sweden; European DLB consortium; Analytic Imaging Diagnostics Arena (AIDA) initiative; VINNOVA; Formas [2017-02447]; Swedish Energy Agency; Swiss National Science Foundation; Velux Foundation [320030_169876, 320030_185028]; Flanders Research Foundation [1123]; [FWO 12I2121N]
Databáze: OpenAIRE
Externí odkaz: