Multidimensional Visualization of Medical Images with Transformer: Application of the Neutrosophic Biplot Method in the Representation of Indeterminate Correlations.

Abstract

This study addresses the challenge of representing and analyzing complex medical information, where the interpretation of multidimensional images is constrained by the presence of uncertain correlations among variables. Such difficulty in understanding these relationships poses a significant obstacle to diagnostic accuracy and the development of reliable clinical support tools. Currently, biomedical data visualization is gaining increasing importance due to the exponential growth of information generated by advanced imaging devices, where managing indeterminacy remains a pressing methodological challenge. Despite progress in artificial intelligence and deep learning, the literature reveals gaps regarding approaches that integrate multidimensional visualization techniques with mathematical frameworks capable of simultaneously handling uncertainty and the inherent variability of medical data. To overcome this limitation, the combination of Transformer architectures with the Neutrosophic Biplot method is proposed, enabling a more precise and robust graphical representation of latent patterns and indeterminate correlations. The results demonstrate improvements in the interpretation of complex structures and in the identification of hidden dependencies among clinical variables. This contribution not only expands the theoretical foundation concerning the integration of neutrosophic models in bioinformatics but also provides a practical framework applicable in medical practice to strengthen diagnostic processes and clinical decision-making.