A minimalistic approach to classifying Alzheimer’s disease using simple and extremely small convolutional neural networks
The results indicate that using the minimal 3D convolutional neural network SFCN with a minimal preprocessing pipeline can achieve competitive performance in AD classification, challenging the necessity of employing more complex architectures with a larger number of parameters. This finding supports the efficiency of simpler deep learning models for neuroimaging-based AD diagnosis, potentially aiding in better understanding and diagnosing Alzheimer's disease.
Authors
- Anders M. Fjell (ORCID: https://orcid.org/0000-0003-2502-8774)
- Bradley J. MacIntosh (ORCID: https://orcid.org/0000-0001-7300-2355)
- Till Schellhorn (ORCID: https://orcid.org/0000-0002-5336-0515)
- Inge K. Amlien (ORCID: https://orcid.org/0000-0002-8508-9088)
- Øystein Sørensen (ORCID: https://orcid.org/0000-0003-0724-3542)
- Atle Bjørnerud (ORCID: https://orcid.org/0000-0002-3486-3141)
- Esten H. Leonardsen (ORCID: https://orcid.org/0000-0002-5099-116X)
- Edvard Grødem (ORCID: https://orcid.org/0000-0003-1915-1872)
Institutions
- Oslo University Hospital (NO)
- University of Oslo (NO)
- Sunnybrook Hospital (CA)
Publication Details
- Journal
- Journal of Neuroscience Methods
- Published
- 2024-08-22
- DOI
- https://doi.org/10.1016/j.jneumeth.2024.110253
- Citations
- 8
- Primary Topic
- Brain Tumor Detection and Classification
- Type
- article
- Field-Weighted Citation Impact
- 1.98
Funders
- Helse Sør-Øst RHF