Assessing the patient’s metabolism with infra-red video acquisition
Critically ill patients are continuously and closely monitored in order to quickly intervene in cases where their state of health decreases abruptly. The Pediatric Intensive Care Unit (PICU) at the Centre Hospitalier Universitaire (CHU) Ste-Justine receives patients in critical condition ranging from newborns to 18-year-olds.
The heat distribution across the patient’s body surface provides information regarding modifications of the vascularization inside the body. More specifically, cold extremities (hands for instance) indicate the existence of a vascularization centralization that corresponds to a hyper-vascularization of the deep organs, which is detrimental to the superficial organs such as the skin in particular. The phenomenon of peripheral hypo-vascularization happens in situations of severe hemodynamic stress, in particular in the cases of heart failures, severe infections threatening the patient’s survival prognosis, or massive bleedings. Presently, physicians evaluate the peripheral hypo-vascularization by touching and pressing the patient’s extremities in order to assess their relative temperature.
Our goal is to support physician evaluations in the ICU by developing methods based on video analysis and machine learning to assess patient distress, thus addressing these observational limitations by providing objective, quantitative, and continuous monitoring.
Infra-Red (IR) imaging measures the temperature on the skin surface. This is a functional imaging technique since temperature distribution reflects metabolic activity that is caused by a change of vascularization. To measure and analyze a patient’s spatial and temporal body temperature profile, we plan to develop methods based on IR imaging by using an IR camera which will be positioned above the patient to acquire a thermal video of a patient’s skin. Our objective is to analyze the temperature differential between the patient’s extremities and his/her internal organs. The body surface will be segmented. The use of cloud points and skeleton joints provided by an RGB-D camera will be studied and mapped to the IR video. The temperature gradient will be measured and analyzed; the temperature profile will be correlated with a quantitative evaluation of cardiac failure derived from other clinical parameters available in the patient’s electronic medical record.
The candidate must demonstrate:
- Knowledge or interest in artificial intelligence algorithms
- Excellent motivation
- Good research record and/or good academic curriculum
- Ability to work as a team and independently
- Ability to communicate well in writing
- Excellent programming knowledge