Preliminary Quantitative Evaluation of the Optimal Colour System for the Assessment of Peripheral Circulation from...

Area of Science:

• Biomedical Engineering
• Dermatology
• Computer Science

Background:

• Peripheral circulatory failure, characterized by reduced superficial capillary blood flow, is a clinical concern.
• Current prevention strategies rely on visual skin assessment during regular repositioning protocols.
• Objective, quantitative methods are needed to enhance the detection of early-stage pressure-induced skin changes.

Purpose of the Study:

• To develop and validate a machine learning approach for quantitatively analyzing skin color changes due to pressure.
• To identify optimal color space components for distinguishing skin states before and after pressure application.
• To explore the potential of this method in preventing pressure-induced skin injuries.

Main Methods:

• Continuous skin color data capture using a camera under controlled pressure application.
• Supervised machine learning model trained to classify skin states as 'before' or 'after' pressure.
• Evaluation of color space components, including JCh, CIELAB, and HSV, for discriminative performance using Area Under the Curve (AUC).

Main Results:

• The 'h' component of the JCh color space showed the highest discriminative performance (AUC = 0.88).
• The 'a*' component of CIELAB (AUC = 0.84) and 'H' component of HSV (AUC = 0.83) also demonstrated significant discriminative capabilities.
• Feasibility of quantitative skin color change evaluation for pressure-induced alterations was established.

Conclusions:

• Quantitative evaluation of pressure-induced skin color changes using machine learning is feasible.
• Specific color space components (JCh 'h', CIELAB 'a*', HSV 'H') are effective indicators.
• This approach offers a potential tool for dimensionality reduction in feature extraction and preventing pressure-induced skin injuries.