Stroke causes behavioral deficits in multiple cognitive domains and there is a growing interest in predicting patient performance from neuroimaging data using machine learning techniques. Here, we investigated a deep learning approach based on convolutional neural networks (CNNs) for predicting the severity of language disorder from 3D lesion images from magnetic resonance imaging (MRI) in a heterogeneous sample of stroke patients. CNN performance was compared to that of conventional (shallow) machine learning methods, including ridge regression (RR) on the images’ principal components and support vector regression. We also devised a hybrid method based on re-using CNN’s high-level features as additional input to the RR model. Predictive accuracy of the four different methods was further investigated in relation to the size of the training set and the level of redundancy across lesion images in the dataset, which was evaluated in terms of location and topological properties of the lesions. The Hybrid model achieved the best performance in most cases, thereby suggesting that the high-level features extracted by CNNs are complementary to principal component analysis features and improve the model’s predictive accuracy. Moreover, our analyses indicate that both the size of training data and image redundancy are critical factors in determining the accuracy of a computational model in predicting behavioral outcome from the structural brain imaging data of stroke patients.
Background: The hippocampus and hippocampal subfields have been found to be diversely affected in Alzheimer’s Disease (AD) and early stages of Alzheimer’s disease by neuroimaging studies. However, our knowledge is still lacking about the trajectories of the hippocampus and hippocampal subfields atrophy with the progression of Alzheimer’s disease.
Objective: To identify which subfields of the hippocampus differ in the trajectories of Alzheimer’s disease by magnetic resonance imaging (MRI) and to determine whether individual differences on memory could be explained by structural volumes of hippocampal subfields.
Methods: Four groups of participants including 41 AD patients, 43 amnestic mild cognitive impairment (aMCI) patients, 35 subjective cognitive decline (SCD) patients and 42 normal controls (NC) received their structural MRI brain scans. Structural MR images were processed by the FreeSurfer 6.0 image analysis suite to extract the hippocampus and its subfields. Furthermore, we investigated relationships between hippocampal subfield volumes and memory test variables (AVLT-immediate recall, AVLT-delayed recall, AVLT-recognition) and the regression model analyses were controlled for age, gender, education and eTIV.
Results: CA1, subiculum, presubiculum, molecular layer and fimbria showed the trend toward significant volume reduction among four groups with the progression of Alzheimer’s disease. Volume of left subiculum was most strongly and actively correlated with performance across AVLT measures.
Conclusion: The trend changes in the hippocampus subfields and further illustrates that SCD is the preclinical stage of AD earlier than aMCI. Future studies should aim to associate the atrophy of the hippocampal subfields in SCD with possible conversion to aMCI or AD with longitudinal design.