Purpose
Improved Alzheimer’s disease (AD) diagnosis using a single structural MRI scan requires analysis of many aspects of the acquired scan.
Our combination MRI biomarker recently won the Computer-Aided Diagnosis of Dementia based on Structural MRI data(CADDementia) challenge [1] (see Fig.
1).
In an effort to further validate this marker and its clinical applicability,
we report its diagnostic performance on two recognized reference datasets.
Methods and materials
Two T1-weighted structural MRI reference datasets were considered.
ADNI:baseline scans from the "complete annual year 2 visit"1.5T standardized Alzheimer’s Disease Neuroimaging Initiativedataset [2] (169 normal controls (CTRL),
234 subjects with mild cognitive impairment (MCI),
101 AD patients).
AIBL:baseline scans from the imaging arm of the Australian Imaging,
Biomarker & Lifestyle Flagship Study of Aging [3] (88 CTRL,
29 MCI,
and 28 AD).
The challenge-winning combination MRI biomarker was applied to each scan.
First,the following individual MRI biomarkers were computed:
cortical thickness using cross-sectional FreeSurfer,
hippocampal...
Results
The method achieved the following three-classAUCs with 95 % confidence intervals: ADNI 0.779 [0.748 0.809] andAIBL 0.803 [0.752 0.857].
The Per-class AUCs for ADNI were as follows (see Fig.
2 for associated ROC-curves): CTRL 0.853,MCI 0.678,
and AD 0.819.
For AIBL,
the AUCs were CTRL 0.895m MCI0.715,
andAD0.803 (see Fig 3.
for associated ROC-curves).
Conclusion
The reported diagnostic results were comparable to the challenge-winning results achieved for the CADDementia dataset [1].
This demonstrates that the state-of-the-art performance of the combination MRI marker generalizes to recognized reference datasets,
making it a potential marker for improved diagnostic support in clinical assessment of AD.
Personal information
Lauge Sørensen
Biomediq A/S,
Copenhagen,
Denmark
[email protected]
Martin Lillholm
Biomediq A/S,
Copenhagen,
Denmark
[email protected]
Akshay Pai
Department of Computer Science,
University of Copenhagen,
Denmark
Ioana Balas
Department of Computer Science,
University of Copenhagen,
Denmark
Cecilie Anker
DTU Compute,
Technical University of Denmark,
Denmark
Christian Igel
Department of Computer Science,
University of Copenhagen,
Denmark
Mads Nielsen
Biomediq A/S,
Copenhagen,
Denmark
[email protected]
References
Bron,
E.E.,
Smits,M.,
van der Flier,W.M.,
Vrenken,
H.,
Barkhof,
F.,
Scheltens,
P.,
Papma,
J.M.,
Steketee,
R.M.,
Orellana,
C.M.,
Meijboom,
R.,
Pinto,
M.,
Meireles,
J.R.,
Garrett,
C.,
Bastos-Leite,
A.J.,
Abdulkadir,
A.,
Ronneberger,
O.,
Amoroso,
N.,
Bellotti,
R.,
C´ardenas-Pena,
D.,
Alvarez Mezao,
A.M.,
Dolph,
C.V.,
Iftekharuddin,
K.M.,
Eskildsen,
S.F.,
Coup´e,
P.,
Fonov,
V.S.,
Franke,
K.,
Gaser,
C.,
Ledig,
C.,
Guerrero,
R.,
Tong,
T.,
Gray,
K.R.,
Moradi,
E.,
Tohka,
J.,
Routier,
A.,
Durrleman,
S.,
Sarica,
A.,
Fatta,
G.D.,
Sensi,
F.,
Chincarini,
A.,
Smith,
G.M.,
Stoyanov,
Z.V.,
Sørensen,
L.,...