Comparing MRI Techniques for Alzheimer's Diagnosis
Author Information
Author(s): Westman Eric, Cavallin Lena, Muehlboeck J-Sebastian, Zhang Yi, Mecocci Patrizia, Vellas Bruno, Tsolaki Magda, Kłoszewska Iwona, Soininen Hilkka, Spenger Christian, Lovestone Simon, Simmons Andrew, Wahlund Lars-Olof
Primary Institution: Karolinska Institutet, Stockholm, Sweden
Hypothesis
Can multivariate MRI classification methods outperform traditional visual assessments in diagnosing Alzheimer's disease?
Conclusion
Visual rating assessment of the medial temporal lobe gave similar prediction accuracy to multivariate classification and manual hippocampal volumes.
Supporting Evidence
- The study included 252 subjects: 75 AD patients, 101 MCI patients, and 81 healthy controls.
- Prediction accuracies were 83% for multivariate classification, 81% for visual ratings, and 89% for manual hippocampal volume measurements.
- The OPLS model showed better specificity compared to visual assessments.
Takeaway
This study looked at different ways to check for Alzheimer's disease using brain scans. It found that new computer methods can be just as good as doctors at spotting the disease.
Methodology
The study used high-resolution MRI scans from patients with Alzheimer's disease, mild cognitive impairment, and healthy controls to compare visual assessments and multivariate classification methods.
Potential Biases
Potential biases in manual measurements and the reliance on visual assessments by neuroradiologists.
Limitations
The data is not neuropathologically confirmed and the follow-up of MCI subjects was only for one year.
Participant Demographics
{"AD":{"number":75,"female":50,"male":25,"age":"74.2±6.0","education":"8.3±4.2"},"MCI":{"number":101,"female":52,"male":49,"age":"74.0±5.8","education":"8.7±4.3"},"CTL":{"number":81,"female":45,"male":36,"age":"73.6±6.3","education":"11.0±4.8"}}
Statistical Information
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website