Skip to main content
Fig. 3 | Alzheimer's Research & Therapy

Fig. 3

From: Optimal transport- and kernel-based early detection of mild cognitive impairment patients based on magnetic resonance and positron emission tomography images

Fig. 3

Our novel OT TL framework and pipeline were used to train the patient stratification model. Using the pre-processing workflow, we extract VBM and FDG features from the manually labeled regions-of-interest (ROIs) in MRI and FDG images respectively. Then, we use the rMLTFL framework as well as one-way ANOVA to select features from two modalities (FDG and VBM) both separately and simultaneously. We then separate the target dataset into training (80%) and testing (20%) sets. After that, we sample subsets of training samples, use the regularized OT to mapping selected samples to the AD vs NC data domain, and train classifiers using labeled AD, NC, and transformed samples. Finally, we aggregate these models to form a robust BAg model and make predictions on OT transformed testing samples

Back to article page