Sensors, Signal and Image Processing in Biomedicine and Assisted Living

• Published in: Sensors (Basel, Switzerland) Vol. 20; no. 18; p. 5071
• Main Author: Iakovidis, Dimitris K.
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 07.09.2020; MDPI
• Subjects: Algorithms; Biosensing Techniques; Brain cancer; Classification; Endoscopy; Heart rate; Image Processing, Computer-Assisted; Machine learning; Medical imaging; Methods; Monitoring, Physiologic; n/a; Neural networks; Sensors; Signal Processing, Computer-Assisted; Sleep; Strain gauges; Wavelet transforms; Wheelchairs
• ISSN: 1424-8220; 1424-8220
• DOI: 10.3390/s20185071

In this context digital signal and image processing methods play an important role in feature or quality enhancement, and compression of the acquired, transmitted, or received biomedical signals and images. Novel contributions of that work include the development of a robust hierarchical multiple-target pose tracking method in uncontrolled environments in the presence of multiple human actors; the introduction of an explicit body movement representation, called pose evolution, that can be used to complement appearance and motion cues for action recognition; and, a target-specific action classification architecture applied on video recordings of patients with PD. The experimental evaluation of the proposed method indicates that it can provide accurate target-specific classification of activities in the presence of other human actors, robust to changing environments. [...]EEG can provide information not only about motor activities that actually happen, but also about motor activities that are imagined by a human subject. The proposed system goes beyond the state-of-the-art by following a novel uncertainty-aware approach to obstacle detection, incorporating salient regions generated using a Generative Adversarial Network (GAN) trained to estimate saliency maps based on human eye-fixations.