French
English
Quantitative spatiotemporal analysis of high-resolution infrared thermography for peripheral perfusion assessment
Background and research question
Circulatory shock is an acute circulatory failure resulting in inadequate tissue perfusion. Peripheral perfusion disorders are often visible at the skin level and are associated with high rates of complications and mortality. Several studies suggest that skin temperature changes, measured non-invasively using infrared thermography, could help predict clinical deterioration. Current methods mainly rely on the central-to-peripheral temperature gradient, calculated from the difference between core temperature and the temperature of a peripheral site, such as the foot. However, this global measurement cannot capture spatial and temporal variations within the imaged region. In conditions such as septic shock, heterogeneous microvascular perfusion may generate heterogeneous skin temperature patterns that are overlooked when temperature is reduced to a single average value.
The research question is therefore whether ultra-high-resolution infrared thermography can reveal this spatial heterogeneity and whether radiomic features extracted from thermal images can provide relevant markers of peripheral perfusion impairment and potential early indicators of clinical deterioration.
Objectives
The primary objective of this Master’s internship is to analyze preliminary high-resolution infrared thermography data acquired from pigs with septic shock and healthy control pigs, and to compare thermal imaging markers between these two experimental conditions. The specific objectives will be to:
extract radiomic features and spatiotemporal descriptors of skin temperature heterogeneity from thermal recordings;
compare these quantitative imaging features between septic shock and control conditions;
investigate the associations between thermal imaging features and physiological variables, particularly hemodynamic parameters.
Methodology and Work plan
The work will be organized into four main stages :
Data preparation and quality control: thermal recordings will be reviewed, organized and synchronized with the available physiological measurements.
Image preprocessing: anatomically relevant regions of interest will be defined. Appropriate preprocessing methods will be applied to ensure that thermal data are comparable across animals and recording periods.
Feature extraction: radiomic features and spatiotemporal descriptors will be computed from the selected regions.
Statistical analysis: extracted features will be compared between septic shock and control conditions. Their relationships with available physiological and hemodynamic variables will then be investigated.
Candidate profile
The candidate should have:
good programming skills, preferably in Python;
experience in image processing, quantitative data analysis or statistical analysis;
the ability to work with complex experimental datasets;
a strong interest in biomedical research and physiological applications;
scientific rigor, autonomy and good communication skills.
Research environment and supervision
The internship will be conducted jointly by the APCSe research unit at VetAgro Sup and the CREATIS medical imaging research laboratory at INSA Lyon. This collaboration will provide the student with an interdisciplinary environment combining cardiovascular physiology, critical care research, medical image processing and quantitative data analysis.
The project will be co-supervised by:
Carole Frindel, Associate Professor at INSA Lyon and researcher at CREATIS. Her research focuses on medical image processing, computer vision and machine-learning methods for clinical applications.
Email: carole.frindel@creatis.insa-lyon.fr
Mathieu Magnin, Associate Professor at VetAgro Sup and researcher in the APCSe unit. His research focuses on cardiovascular physiology, circulatory shock and the assessment of peripheral perfusion using non-invasive monitoring methods.
Email: mathieu.magnin@vetagro-sup.fr