This poster is published under an
open license. Please read the
disclaimer for further details.
Keywords:
Computer applications, Neuroradiology brain, MR, Computer Applications-Detection, diagnosis, Comparative studies, CAD, Dementia
Authors:
E. Bron1, M. Smits1, F. Barkhof2, A. J. Bastos-Leite3, J. C. van Swieten1, W. J. Niessen1, S. Klein1; 1Rotterdam/NL, 2Amsterdam/NL, 3Porto/PT
DOI:
10.1594/ecr2015/B-0244
Methods and materials
A total of 384 clinically representative T1-weighted MRI scans were acquired at the VU Medical Center (Amsterdam,
Netherlands),
Erasmus MC (Rotterdam,
Netherlands),
and University of Porto (Portugal).
Details on the data can be found in Fig. 2.
Fig. 2: Data characteristics
Fig. 3 shows an example scan for an AD patient.
The diagnostic label (AD,
MCI,
CN) was based on clinical criteria as reference standard.
Fig. 3: Example of a T1-weighted MRI scan of the training set.
The algorithms were trained on a small training set (n=30) and optionally on other data (e.g.,
Alzheimer’s Disease Neuroimaging Initiative (ADNI; Jack et al.,
2008),
Australian Imaging Biomarkers and Lifestyle flagship study of aging (AIBL; Ellis et al.,
2009).
The test set consisted of 354 scans with the diagnoses blinded.
Via our web-based framework,
15 research teams uploaded a total of 29 algorithms.
More information on the algorithms can be found in Bron et al.
(2014).
We analyzed area-under-the-receiver-operating-characteristic-curve (AUC) and accuracy of the algorithms.
Confidence intervals (CI) were estimated with bootstrapping.
Differences between classifiers were assessed using McNemar’s test.