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


Por: Soliman, A, Chang, JR, Etminani, K, Byttner, S, Davidsson, A, Martinez-Sanchis, B, Camacho, V, Bauckneht, M, Stegeran, R, Ressner, M, Agudelo-Cifuentes, M, Chincarini, A, Brendel, M, Rominger, A, Bruffaerts, R, Vandenberghe, R, Kramberger, MG, Trost, M, Nicastro, N, Frisoni, GB, Lemstra, AW, van Berckel, BNM, Pilotto, A, Padovani, A, Morbelli, S, Aarsland, D, Nobili, F, Garibotto, V, Ochoa-Figueroa, M

Publicada: 7 dic 2022
Resumen:
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.

Filiaciones:
Soliman, A:
 Halmstad Univ, Ctr Appl Intelligent Syst Res CAISR, Halmstad, Sweden

Chang, JR:
 Halmstad Univ, Ctr Appl Intelligent Syst Res CAISR, Halmstad, Sweden

 Natl Cheng Kung Univ Tainan, Taipei, Taiwan

Etminani, K:
 Halmstad Univ, Ctr Appl Intelligent Syst Res CAISR, Halmstad, Sweden

Byttner, S:
 Halmstad Univ, Ctr Appl Intelligent Syst Res CAISR, Halmstad, Sweden

Davidsson, A:
 Inst Med & Hlth Sci, Dept Clin Physiol, Linkoping, Sweden

Martinez-Sanchis, B:
 La Fe Univ Hosp, Dept Nucl Med, Med Imaging Area, Valencia, Spain

Camacho, V:
 Univ Autonoma Barcelona, Serv Med Nucl, Hosp Santa Creu & St Pau, Barcelona, Spain

Bauckneht, M:
 IRCCS Osped Policlin San Martino, Nucl Med Unit, Genoa, Italy

Stegeran, R:
 Linkoping Univ Hosp, Dept Diagnost Radiol, Linkoping, Sweden

Ressner, M:
 Linkoping Univ Hosp, Dept Med Phys, Linkoping, Sweden

Agudelo-Cifuentes, M:
 La Fe Univ Hosp, Dept Nucl Med, Med Imaging Area, Valencia, Spain

Chincarini, A:
 Natl Inst Nucl Phys INFN, Genoa Sect, Genoa, Italy

Brendel, M:
 Ludwig Maximilians Univ Munchen, Univ Hosp, Dept Nucl Med, Munich, Germany

Rominger, A:
 Univ Hosp Bern, Dept Nucl Med, Inselspital, Bern, Switzerland

Bruffaerts, R:
 Univ Antwerp, Dept Biomed Sci, Antwerp, Belgium

Vandenberghe, R:
 KU, Lab Cognit Neurol, Dept Neurosci, Leuven, Belgium

 Univ Hosp Leuven, Neurol Dept, Leuven, Belgium

Kramberger, MG:
 Univ Med Ctr, Dept Neurol, Ljubljana, Slovenia

Trost, M:
 Univ Med Ctr, Dept Neurol, Ljubljana, Slovenia

 Univ Ljubljana, Fac Med, Ljubljana, Slovenia

Nicastro, N:
 Geneva Univ Hosp, Dept Clin Neurosci, Geneva, Switzerland

Frisoni, GB:
 Univ Hosp, Dept Psychiat, LANVIE Lab Neuroimagerie Vieillissement, Geneva, Switzerland

Lemstra, AW:
 VU Med Ctr Alzheimer Ctr, Amsterdam, Netherlands

van Berckel, BNM:
 Vrije Univ Amsterdam, Dept Radiol & Nucl Med, Amsterdam UMC, Amsterdam Neurosci, Amsterdam, Netherlands

Pilotto, A:
 Univ Brescia, Dept Clin & Expt Sci, Neurol Unit, Brescia, Italy

Padovani, A:
 IRCCS Osped Policlin San Martino, Nucl Med Unit, Genoa, Italy

Morbelli, S:
 Stavanger Univ Hosp, Ctr Age Related Med SESAM, Stavanger, Norway

Aarsland, D:
 Univ Genoa, Dept Neurosci DINOGMI, Genoa, Italy

 Kings Coll London, Inst Psychiat Psychol & Neurosci, Dept Old Age Psychiat, London, England

Nobili, F:
 Univ Genoa, Dept Neurosci DINOGMI, Genoa, Italy

Garibotto, V:
 Univ Geneva, Univ Hosp, Div Nucl Med & Mol Imaging, Geneva, Switzerland

 Univ Geneva, NIMTLab, Geneva, Switzerland

Ochoa-Figueroa, M:
 Inst Med & Hlth Sci, Dept Clin Physiol, Linkoping, Sweden

 Linkoping Univ Hosp, Dept Diagnost Radiol, Linkoping, Sweden

 Linkoping Univ, Ctr Med Image Sci & Visualizat CMIV, Linkoping, Sweden
ISSN: 14726947
Editorial
BMC, CAMPUS, 4 CRINAN ST, LONDON N1 9XW, ENGLAND, Reino Unido
Tipo de documento: Article
Volumen: 22 Número: SUPPL 6
Páginas:
WOS Id: 000904994900001
ID de PubMed: 36476613
imagen Green Published, gold, All Open Access, Gold, Green

MÉTRICAS