Single-pixel infrared imaging thermometry maps human inner canthi temperature

Abstract

Abstract Efficiently and accurately mapping the temperature of human inner canthi is crucial for disease diagnostics and monitoring. The specific anatomical location of the inner canthi precludes temperature screening methods that are invasive, require tissue contact, and/or demand active illumination. Camera-based thermography, although capable of passive and non-contact temperature mapping, suffers from low efficiency in pixel allocation to the inner canthi as well as from measurement inaccuracies due to background blending and moderate pixel sensitivity. In response to these challenges, we develop single-pixel infrared imaging thermometry (SPIRIT). We design diagonally aggregated two-dimensional transmissive encoding masks using a cyclic S-matrix, which supports compressed data acquisition in a single scan and high image quality through non-iterative reconstruction. SPIRIT maps the temperature distribution of human inner canthi with a resolution of 0.3 °C, which enables human temperature mapping via single-pixel imaging. Using SPIRIT, we reveal sub-degree temperature differences induced by daily physical activities and the glasses-wearing habit. These findings shed light on SPIRIT’s contribution to improving evaluation criteria for public health, including COVID-19 febrile screening.