Automating the Koedam Parietal Atrophy Scale for Alzheimer's Using MRI Features and Clustering Techniques

Early detection of Alzheimer’s disease (AD) is crucial for effective intervention, and imaging biomarkers are pivotal in this process. The search for imaging biomarkers is important in diagnosing AD, offering a non-invasive and potentially early method to identify brain changes associated with the disease. These biomarkers can provide valuable insights into the progression of AD and aid in differential diagnosis, enabling the application of more effective treatment strategies. In this context, the Koedam visual scale for parietal atrophy is a valuable tool for assessing structural changes in the parietal lobe associated with AD. This study proposes an automated approach for the Koedam scale using attributes extracted from T1-weighted magnetic resonance imaging (MRI) combined with clustering techniques. Initially, a preprocessing pipeline is applied to the images to skull stripping, to mitigate noise and bias field effects and to define the ROI (parietal region). Subsequently, a finite mixture model is applied to segment the images into gray matter, white matter, and cerebrospinal fluid. The volume of each tissue is then utilized as a feature for clustering, effectively simulating the visual categorization of the Koedam scale. Our method, tested on 103 MRI images, demonstrates potential for automating the assessment of parietal atrophy, providing a more objective and efficient evaluation tool.