Early bio‐maker detection through novel data equilibration methods. (December 2021)

• Record Type: Journal Article
• Authors: Pal, Prasanta; Brewer, Judson A
• Is Part Of: Alzheimer's & dementia. Volume 17(2021)Supplement 4
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-12
• Subjects: Alzheimer's disease -- Periodicals; Alzheimer Disease -- Periodicals; Dementia -- Periodicals; Démence; Maladie d'Alzheimer; Périodique électronique (Descripteur de forme); Ressource Internet (Descripteur de forme); 616.83
• DOI: 10.1002/alz.055255
• Languages: English
• ISSNs: 1552-5260

Abstract

Abstract: Background: A novel adaptive universal set of algorithms are developed to map and differentiate normal and alzheimer patient MRI images into orthogonal complementary image spaces. This adaptive method greatly enhances the visual clarity of both (normal and alzheimer) images while providing with additional metric space of outlier image for further differentiation of features in the data sets. This enables us with extra handle to compare baseline and alzheimer MRI images with unprecedented clarity and robustness Method: To observe multi‐scale features and its differences between baseline and disease condition MRI data sets (publicly available), we introduce a set of length scale kernels coupled with threshold parameters to transform the original image space to parametrically controlled equilibrium and out of equilibrium (or outlier) data sets. This forms the foundation of the new method of comparison through a novel method of outlier based orthogonal decomposition of image information space. Result: While our new comparison method confirms the findings of published results, it adds additional layers of visual clarity for further discovery of the novel never before seen features. For the first time in the literature, we introduce the concept of equilibrium (and nonequilibrium) states from raw data acquired through sensors and demonstrate the efficacy of it in the context of early detection of AD biomarkers. Conclusion: While historical methods of image difference works well for detecting crude biomarkers, there is a potential for a information loss and low contrast resolution. Our method overcomes those limitations by appropriately re‐mapping the raw data to equilibrium and nonequilibrium states. This makes the comparison between baseline and AD images much visually clear in those respective metric spaces.