a framework for alzheimer’s diagnosis using dempster-shafer theory and multimodal mri fusion of white and gray matter

Alzheimer’s disease (ad) is a common neurodegenerative disorder that requires early diagnosis for effective treatment. mri data provides valuable insights into brain structure, which can assist in diagnosing the disease. however, traditional diagnostic methods often face errors due to expert limitations and data uncertainty. to address this issue, we propose a computer-aided diagnosis (cad) system that can automatically identify the disease. moreover, since mri images inherently contain uncertainty, the proposed method offers a solution to minimize this uncertainty. in the proposed method, after performing preprocessing, feature extraction, and selection steps, the information obtained from brain white matter (wm) and gray matter (gm) is combined. this combination is achieved using the evidence theory and dempster-shafer theory (dst). in this theory, mass functions are employed instead of probability functions. subsequently, three different classifiers are applied separately in the final stage to the combined data. experimental results demonstrate that combining gm and wm data using dst achieves higher accuracy compared to using either data type alone. this fusion-based method presents a reliable and effective approach for improving alzheimer’s diagnosis. our proposed method achieved 91% accuracy in three different binary classification cases using the lda classifier when distinguishing between ad and normal control (nc) groups. this result, obtained by combining wm and gm data, demonstrated a significant improvement compared to using each data type independently.