Skip to content

view

Prof. Irini Reljin
presents
New Approaches in Biomedical Signal Analysis
 
On 2015-06-18 13:00 at KN:G-205
 
Biomedical signals are very important for physicians because they contain
relevant information describing humans’ vital signs. Very often signals are
followed by significant amount of noise and/or artifacts. In these cases
standard signal processing methods are not so efficient. The main activities of
the group for Image Processing, Telemedicine, and Multimedia (IPTM) at the
School of Electrical Engineering, University of Belgrade, are addressed to
development of signal processing methods targeted to extraction relevant
information even from biomedical signals corrupted by noise. Our main results
are in the field of fractal geometry (FG) and multifractal analysis (MFA). The
FG is efficient in describing complex shapes and phenomena, while the MFA,
particularly the inverse MFA, was applied for detecting and extracting
microcalcifications from mammograms. By combining the MFA and joint
time-frequency approach (such as the Wigner-Ville distribution and wavelets) the
accurate analysis of phonocardiograms is realized and the detection of subtle
clicks characterizing prolapse mitral valve was enabled. Blanket fractal
dimension was used for phonocardiongam processing as well, while the MFA and
Katz fractal dimension are applied to the differentiation of malignant from
benign cases from UV/VIS spectra of body liquids (ascites and pleural
effusions).
Back to the list