Vasilis Marmarelis
· Dean's Professor of Biomedical Engineering and Professor of Biomedical EngineeringUniversity of Southern California · Alfred E. Mann Department of Biomedical Engineering
Active 1999–2022
About
Vasilis Z. Marmarelis received his diploma in Electrical and Mechanical Engineering from the National Technical University of Athens in 1972. He earned his M.S. in Information Science and Ph.D. in Engineering Science with a focus on Bio-Information Systems from the California Institute of Technology in 1973 and 1976, respectively. He served as a Lecturer and Research Fellow at Caltech in BioInformation Systems from 1976 to 1978 before joining the University of Southern California in 1978, where he is currently a Professor in the Departments of Biomedical and Electrical Engineering. Dr. Marmarelis served as Chairman of Biomedical Engineering from 1990 to 1996 and is Co-Director of the Biomedical Simulations Resource, a research center dedicated to modeling and simulation of physiological systems funded by the NIH since 1985. His research interests primarily focus on biomedical systems modeling and signal analysis with applications to physiology and medical diagnosis. He specializes in nonlinear modeling and closed-loop multi-variate systems, with applications including neural information processing, cardio-vascular autoregulation, neuronal ensemble modeling, neurostimulation, physiological feedback, cerebral flow autoregulation, and control of blood glucose. He pioneered the use of Principal Dynamic Modes for modeling nonlinear dynamic systems and improving clinical diagnosis. Dr. Marmarelis invented the 'Multimodal Ultrasound Tomography' system for early breast cancer detection, which has been clinically validated in Europe. His work extends to clinical validation and potential applications of this technology in biopsy assistance, therapy monitoring, osteoporosis diagnosis, ligament and tendon imaging, cardiovascular and brain imaging, and interventional ultrasound techniques. He is a Fellow of the IEEE and AIMBE.
Research topics
- Medicine
- Internal medicine
- Cardiology
- Anesthesia
- Nuclear medicine
Selected publications
Alzheimer s & Dementia · 2022
1st authorCorresponding- Medicine
- Cardiology
- Internal medicine
Abstract Background Alzheimer’s Disease Centers in Los Angeles (USC), Dallas (UT‐SWMC) and Kansas City (KUMC) have joined together to study the regulation of cerebral perfusion through analysis of spontaneous time‐series data using a modeling methodology that previously yielded indications of reduced CO2 dynamic vasomotor reactivity (DVR) in amnestic MCI patients. We report initial results from this ongoing multi‐center study that confirm those preliminary findings and, furthermore, elucidate the effects of slow‐paced breathing. Method Five‐minute spontaneous changes in arterial blood pressure (ABP), end‐tidal CO2 (etCO2) and cerebral blood flow velocity (CBFV) in middle cerebral arteries were obtained before and after 5‐min session of slow‐paced breathing (8 breaths/minute). Using data from 25 MCI patients, 7 AD patients and 45 age‐matched cognitively normal controls (NC), we obtained predictive models of the dynamic effects of ABP and etCO2 (proxy for blood CO2) upon CBFV via our kernel‐based modeling methodology. These predictive models were used to compute indices (physio‐markers) that quantify the DVR in each participant, as time‐average of the model‐predicted CBFV response to unit‐step change of etCO2 over the first 30 seconds. Result The obtained DVR indices were significantly different for the 32 patients (MCI and AD lumped together due to small number of AD) vs. 45 age‐matched controls (p= 0.0099). Notably, the delineation between patients and controls improved (p= 0.0011) after a 5‐min session of slow‐paced breathing. Figure 1 shows the average model‐predicted CBFV response to unit‐step change of etCO2 for 32 MCI/AD patients (red line) and 45 controls (blue line) before and after the 5‐min paced‐breathing session. Note the negative average steady‐state CBFV response of the patients (red) that indicates polarity reversal of the normal CO2 vasomotor reactivity (blue). Better delineation between patients and controls is achieved ( p= 2x10 ‐5 ) when the DVR indices from the two sessions, before and after slow‐paced breathing, were averaged (see histograms in Figure 2). Conclusion Significantly lower DVR indices under resting spontaneous conditions were observed in 32 MCI/AD patients relative to 45 age‐matched controls. This reduction is detected more reliably after 5‐min of slow‐paced breathing. Averaging of the DVR indices obtained before and after slow‐paced breathing improved further this delineation ( p= 2x10 ‐5 ).
Alzheimer s & Dementia · 2021
1st authorCorresponding- Medicine
- Cardiology
- Internal medicine
Abstract Background Three AD Centers have joined together to study the regulation of cerebral perfusion at resting conditions using spontaneous time‐series data with a modeling methodology that previously yielded indications of reduced dynamic vasomotor CO2 reactivity in amnestic MCI patients [ Marmarelis et al., J. Alzh. Dis. 56:89–105, 2017 ] and significant correlation of this reduction with functional impairment of the chemoreflex in these MCI patients [ Marmarelis et al., J. Alzh. Dis. 75:1‐16, 2020 ]. The goal of this multi‐center study is to confirm these preliminary findings with larger cohorts of MCI patients, AD patients and age‐matched cognitively normal controls (NC), and to examine the time‐course of these impairments with longitudinal data over 5 years. Method Quantitative results were obtained through dynamic modeling of the effects of spontaneous changes in arterial blood pressure (ABP) and end‐tidal CO2 (etCO2) upon cerebral blood flow velocity (CBFV) in the middle cerebral arteries measured via transcranial Doppler. The obtained subject‐specific input‐output predictive models were used to compute indices (physio‐markers) that quantify the Dynamic Vasomotor Reactivity (DVR) in each subject/patient and to compare differences between NC, and MCI/AD patients. Result The obtained DVR indices were significantly different for patients (9 MCI and 4 AD taken together due to small numbers) vs. 32 controls (p=0.026). Notably, the delineation between patients and controls improved (p=0.009) after a session of slow paced‐breathing (8 breaths/minute) that led to larger average DVR for controls and to reduction of inter‐subject variability of DVR indices. This is illustrated in Figure 1, where the average model‐predicted responses (representing the CBFV response to a unit‐step change of etCO2) are shown for 13 MCI/AD patients (red line) and 32 controls (blue line) before and after the paced‐breathing session. Further improvement in delineating patients from controls is achieved (p=9x10 ‐5 ) when the DVR indices from the two sessions were averaged for each subject/patient. Conclusion Significantly lower DVR index under resting spontaneous conditions was observed in 13 MCI/AD patients relative to 32 NC before (p= 0.026) and after (p=0.009) slow paced‐breathing session. Averaging of the DVR indices obtained from data before and after paced‐breathing improved substantially this delineation (p=9x10 ‐5 ).
Is dysregulation of cerebral perfusion an early marker and cause of MCI and AD?
Alzheimer s & Dementia · 2020
1st authorCorresponding- Cardiology
- Internal medicine
- Medicine
Abstract Background Our previous work has shown that the dynamic vasomotor CO2 reactivity of cerebral vasculature is significantly reduced (p<0.01) in amnestic MCI patients relative to age‐matched controls [Marmarelis et al., J. Alzh. Dis . 56:89–105, 2017]. This quantitative result was obtained through dynamic modeling of the effects of spontaneous changes of arterial blood pressure (ABP) and end‐tidal CO2 (etCO2) upon cerebral blood flow or cortical tissue oxygenation (CTO) measured via Near‐Infrared Spectroscopy at the lateral prefrontal cortex. This study incorporated the dynamic effects of additional contemporaneous changes of heart‐rate (HR), respiratory‐rate (RR) and tidal‐volume (TV) to examine whether and how the intertwined dynamics of these complex mechanisms of physiological regulation influence differentially the cerebral/cortical perfusion in MCI patients versus age‐matched controls. Method We employed our novel multi‐variate predictive modeling methodology to quantify the input‐output dynamic relationship between contemporaneous changes of CTO (output) and five input variables: ABP, etCO2, HR, RR and TV, under spontaneous resting conditions. The obtained subject‐specific input‐output predictive models were used to compute indices that quantify relevant physiological mechanisms and compare differences between MCI patients (MP) and control subject (CS). Result The obtained model‐based indices were significantly different for MP vs. CS with regard to the dynamic relationships between CTO and etCO2 or TV changes. This is illustrated in Figure 1, where the average model‐kernels (representing the CTO response to a unit‐impulse change of etCO2 or TV) are shown for 40 MP (red line) and 15 CS (blue line), along with SD bounds (dashed lines). Improved delineation between MP and CS resulted (p<0.001) when the obtained model‐based indices of Dynamic Vasomotor CO2 Reactivity (DVCR) and Dynamic Cortical Oxygenation Reactivity (DCOR), respectively, were combined in a composite index defined by the Fisher Discriminant of the respective scatter‐plot (see Figure 2). Conclusion Dysregulation of cerebral/cortical perfusion can be detected early and quantified through dynamic predictive modeling of spontaneous hemodynamic/respiratory time‐series data, allowing delineation between MCI patients and controls (p< 0.001) for improved MCI diagnosis. This is viewed as an indication that such dysregulation of cerebral perfusion may be an early trigger/link in the pathogenic cascade towards neurodegenerative disease.
Frequent coauthors
- 9 shared
Rong Zhang
The University of Texas Southwestern Medical Center
- 6 shared
Dae C. Shin
University of Southern California
- 6 shared
Danilo Cardim
Institute for Exercise and Environmental Medicine
- 6 shared
C. Munro Cullum
- 3 shared
Sandra A. Billinger
University of Kansas Medical Center
- 3 shared
Elizabeth Joe
University of Southern California
- 3 shared
Suhaib Hashem
University of Southern California
- 3 shared
Helena C. Chui
University of Southern California
Labs
Awards & honors
- Fellow of the IEEE (2005)
- Fellow of the American Institute of Medical and Biological E…
Similar researchers at University of Southern California
- Resume-aware match score
- Save to shortlist
- AI-drafted outreach
See your match with Vasilis Marmarelis
PhdFit ranks faculty by your research interests, methods, and publications — grounded in their actual work, not templates.
- Free to start
- No credit card
- 30-second signup