A decision support system for type 1 diabetes mellitus diagnostics based on dual channel analysis of red blood cell membrane fluidity. (August 2018)

• Record Type: Journal Article
• Authors: Cordelli, Ermanno; Maulucci, Giuseppe; De Spirito, Marco; Rizzi, Alessandro; Pitocco, Dario; Soda, Paolo
• Is Part Of: Computer methods and programs in biomedicine. Volume 162(2018)
• Page Start: 263
• Page End: 271
• Publication Date: 2018-08
• Subjects: Type 1 Diabetes -- Two-photon microscopy -- Image processing -- Feature extraction -- Machine learning; Medicine -- Computer programs -- Periodicals; Biology -- Computer programs -- Periodicals; Computers -- Periodicals; Medicine -- Periodicals; Médecine -- Logiciels -- Périodiques; Biologie -- Logiciels -- Périodiques; Biology -- Computer programs; Medicine -- Computer programs; Periodicals; Electronic journals; 610.28
• DOI: 10.1016/j.cmpb.2018.05.025
• Languages: English
• ISSNs: 0169-2607

Abstract

Highlights: Investigate the use of human cells membrane fluidity for type 1 diabetes monitoring. We present a decision support system that classifies type 1 diabetes mellitus patients. The experiments were carried out on a wide dataset of images from the subjects. The results outperform the glycosylated hemoglobin test used in the state-of-the-art. Abstract: Background and objective: Investigation of membrane fluidity by metabolic functional imaging opens up a new and important area of translational research in type 1 diabetes mellitus, being a useful and sensitive biomarker for disease monitoring and treatment. We investigate here how data on membrane fluidity can be used for diabetes monitoring. Methods: We present a decision support system that distinguishes between healthy subjects, type 1 diabetes mellitus patients, and type 1 diabetes mellitus patients with complications. It leverages on dual channel data computed from the physical state of human red blood cells membranes by means of features based on first- and second-order statistical measures as well as on rotation invariant co-occurrence local binary patterns. The experiments were carried out on a dataset of more than 1000 images belonging to 27 subjects. Results: Our method shows a global accuracy of 100%, outperforming also the state-of-the-art approach based on the glycosylated hemoglobin. Conclusions: The proposed recognition approach permits to achieve promising results.