Sensor fusion and artifact reduction for non-contact physiological measurements for medical applications

Leuven - PhD
More than two weeks ago

Help improve patient comfort and safety by making reliable unobtrusive physiological monitoring possible through novel algorithmic solutions



Traditionally, physiological measurements require a sensor of some kind to be attached to the subject/patient. These sensors can be connected to large hospital equipment or in more convenient wearable form factors such as patches, watches etc. depending on the use case and how critical the reliability and accuracy of the measured parameters is.

In either case, this creates an additional burden for the subject which implies that systematic monitoring of physiological parameters is typically only done in cases where there is a clear need, for example in high-care units in hospitals, or for follow-up of high-risk patients. Nevertheless, there are many potential cases where such systematic monitoring would be beneficial (assuming proper privacy protections and data security are in place), such as monitoring in lower-care hospital units, screening and prevention of diseases or accidents, etc.

Truly unobtrusive measurements, where once sensors are installed the subject does not have to take any specific actions (besides consenting once to the monitoring), would potentially enable this category of applications, if an acceptable level of reliability and accuracy can be obtained.

Recently significant progress has been made in non-contact sensing of physiological parameters, including capacitive measurements[1],[2], radar based measurements [3],[4], camera based measurements[5] and even based on wifi-like signals[6]. Nevertheless a main obstacle towards adoption, especially for medical applications, remains the reliability and accuracy of the obtained physiological parameters. The non-contact sensors remain significantly more susceptible to artefacts, mainly due to motion of the subject, than their traditional contact-based counterparts.

Main scientific hypothesis to be researched in the PhD

Unique about imec's approach to non-contact sensing is that it has capacitive sensing and radar sensing under the same roof. The main scientific hypothesis is that through smart sensor fusion of both technologies the reliability and accuracy of the physiological measurements under real-world conditions can be brought to a significantly higher level, enabling applications that so far have been seen as out of reach.

One such application could be monitoring of hospital patients that are discharged from high- or medium-care wards towards lower-care wards where suddenly no monitoring at all is done anymore. This lack of monitoring is one of the reasons patients are sometimes kept longer in high- or medium-care wards, especially in case of concerns there remains a small but non-negligible risk of respiratory arrest or cardiac arrest. In the hypothesis that through sensor fusion the non-contact monitoring technology  can be brought to a medically acceptable level of reliability, even though inferior to the reliability of the traditional monitoring devices used in the higher-care units, this could enable savings in human lives (more emergency situations such as respiratory or cardiac arrest can be detected in time in low care wards and thus improve the outcome) as well as in costs (patients can be moved more quickly to significantly cheaper low-care wards).

Draft 4-year research plan

  • Year 1: Literature/state-of-the-art study. Applying the existing imec systems. Propose a measurement framework to enable medical applications. Initial pre-study data collection on a limited number of healthy volunteers.
  • Year 2: Develop sensor fusion algorithms and pre-validate them on existing data from literature, previous imec data and from pre-study.
  • Year 3: Draft protocol for clinical data collection, perform clinical data collection, analyze results. Validate the developed algorithms. Publish results.
  • Year 4: Analyze learnings from clinical data collection, finetune algorithms and measurement approach, recommendations for further improvements, conclusions, thesis writing.

[1] Castro, I.D.; Varon, C.; Torfs, T.; Van Huffel, S.; Puers, R.; Van Hoof, C., "Evaluation of a Multichannel Non-Contact ECG System and Signal Quality Algorithms for Sleep Apnea Detection and Monitoring.", Sensors 2018, 18, 577.

[2] Wartzek, T.; Eilebrecht, B.; Lem, J.; Lindner, H.; Leonhardt, S.; Walter, M., "ECG on the Road: Robust and Unobtrusive Estimation of Heart Rate.", IEEE Trans. Biomed. Eng. 2011, 58, 3112–3120.

[3] M. Mercuri et al., "A Direct Phase-Tracking Doppler Radar Using Wavelet Independent Component Analysis for Non-Contact Respiratory and Heart Rate Monitoring," in IEEE Transactions on Biomedical Circuits and Systems, vol. 12, no. 3, pp. 632-643, June 2018.

[4] C. Li, V. M. Lubecke, O. Boric-Lubecke and J. Lin, "A Review on Recent Advances in Doppler Radar Sensors for Noncontact Healthcare Monitoring," in IEEE Transactions on Microwave Theory and Techniques, vol. 61, no. 5, pp. 2046-2060, May 2013.

[5] W. Wang, A. C. Den Brinker, S. Stuijk, and G. De Haan, "Algorithmic Principles of Remote PPG,"IEEE Trans. Biomed. Eng., vol. 64, no. 7, pp. 1479–1491, 2017.

[6] Shichao Yue, Hao He, Hao Wang, Hariharan Rahul, and Dina Katabi. 2018, "Extracting Multi-Person Respiration from Entangled RF Signals", Proc. ACM Interact. Mob. Wearable Ubiquitous Technol. 2, 2, Article 86 (July 2018)

Required background: Engineering Science, Computer Science

Type of work: 15% literature study, 50% design of algorithms, 35% experimental validation

Supervisor: Chris Van Hoof

Daily advisor: Tom Torfs

The reference code for this position is 1812-63. Mention this reference code on your application form.


Share this on


This website uses cookies for analytics purposes only without any commercial intent. Find out more here. Our privacy statement can be found here. Some content (videos, iframes, forms,...) on this website will only appear when you have accepted the cookies.

Accept cookies