Random forest-based similarity measures for multi-modal classification of Alzheimer's disease

Abstract

Neurodegenerative disorders, such as Alzheimer’s disease, are associated with changes in multiple neuroimaging and biological measures. These may provide complementary information for diagnosis and prognosis. We present a multi-modality classification framework in which manifolds are constructed based on pairwise similarity measures derived from random forest classifiers. Similarities from multiple modalities are combined to generate an embedding that simultaneously encodes information about all the available features. Multimodality classification is then performed using coordinates from this joint embedding. We evaluate the proposed framework by application to neuroimaging and biological data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). Features include regional MRI volumes, voxel-based FDG-PET signal intensities, CSF biomarker measures, and categorical genetic information. Classification based on the joint embedding constructed using information from all four modalities out-performs classification based on any individual modality for comparisons between Alzheimer’s disease patients and healthy controls, as well as between mild cognitive impairment patients and healthy controls. Based on the joint embedding, we achieve classification accuracies of 89% between Alzheimer’s disease patients and healthy controls, and 75% between mild cognitive impairment patients and healthy controls. These results are comparable with those reported in other recent studies using multi-kernel learning. Random forests provide consistent pairwise similarity measures for multiple modalities, thus facilitating the combination of different types of feature data. We demonstrate this by application to data in which the number of features differ by several orders of magnitude between modalities. Random forest classifiers extend naturally to multi-class problems, and the framework described here could be applied to distinguish between multiple patient groups in the future.

Learning and Combining Image Similarities for Neonatal Brain Population Studies

“…In [6], a feature is considered as important, if it is selected in the first three levels of the trees. A more sophisticated approach was used in [5], where the decrease in the Gini impurity criterion was measured for the individual features in each node. In this work, we adopt the former and simpler approach.…”

“…In [5], they were applied to manifold learning by deriving the pairwise similarity measures from random forest classifiers for different modalities. Additionally, the most important features for the classification problem could be extracted.…”

Using Data Mining Techniques to Support Breast Cancer Diagnosis

“…This is an ensemble classifier consisting of many decision trees, where the final predicted class for a test example is obtained by combining the predictions of all individual trees, as described further in this work (chapter 4.6) (46). This method can provide measures of the similarity between pairs of examples in the dataset and are often applied to high-dimensional data sets (47).…”

Assessing biofilm development in drinking water distribution systems by Machine Learning methods

“…An RF algorithm uses a random feature selection, a random subset of input features or predictive variables in the division of every node, instead of using the best variables, which reduces the generalization error. Additionally, to increase the diversity of the trees, an RF uses bootstrap aggregation (bagging) to make the trees grow from different training data subsets [212]. of each tree [212].…”

“…Additionally, to increase the diversity of the trees, an RF uses bootstrap aggregation (bagging) to make the trees grow from different training data subsets [212]. of each tree [212]. These are referred to as the 'out-of-bag' data (OOB) of the tree, for which internal test predictions can be made.…”