Premium
IC‐P‐085: CHARACTERIZATION OF LENTICULOSTRIATE ARTERIES USING ARTERIAL SPIN LABELING AND HIGH‐RESOLUTION 3D BLACK‐BLOOD MRI AS AN IMAGING MARKER IN VASCULAR COGNITIVE IMPAIRMENT AND DEMENTIA
Author(s) -
Ma Samantha J.,
Jann Kay,
Barisano Giuseppe,
Shao Xingfeng,
Yan Lirong,
Casey Marlene,
D'Orazio Lina,
Ringman John M.,
Wang Danny JJ.
Publication year - 2019
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2019.06.4928
Subject(s) - medicine , magnetic resonance imaging , montreal cognitive assessment , cardiology , cerebral blood flow , magnetic resonance angiography , vascular dementia , dementia , radiology , disease
status, familial expected years to symptom onset (EYO), and demographic variables were included as feature vectors for training. For each cortical and subcortical region, a deep feed forward artificial neural network (ANN) was trained to predict changes in neurodegenerative markers (metabolism and volumetrics) over time for each region. 70% of the data was used for training with the remaining 30% used for testing. The mean squared error (MSE) was calculated for each ROI in each test case matrix and averaged across all the test cases to evaluate the overall model performance. Results: The average MSE for FDG was .001 (+/.01), indicating the model performed well over all ROIs. Figure 1 shows the actual versus predicted values over all ROIs for the test data for a 5-year prediction (R 1⁄4 .98). The average MSE for volumetrics was .002 (+/.02). Figure 2 plots the actual versus predicted values for the test data for a 5 year prediction (R 1⁄4 .91). Figure 3 depicts the error histogram for the test data for 1-5 year predictions. Most errors were clustered around zero, indicating little to no difference in the actual and predicted values. Conclusions: Machine learning algorithms can provide decision support and predictive analytics in medicine. Our model is highly accurate and capable of reliably forecasting neurodegenerative changes (structural and metabolic) due to ADAD, and illustrates the feasibility of applying machine learning algorithms for targeted patient care in people with AD. rday, July 13, 2019 P75