A novel stress sensor enables accurate estimation of micro-scale tissue mechanics in quantitative micro-elastography
Kai L Metzner,Qi Fang,Rowan W Sanderson,Yen L Yeow,Celia Green,Farah Abdul-Aziz,Juliana Hamzah,Alireza Mowla,Brendan F Kennedy
DOI: https://doi.org/10.1063/5.0220309
2024-09-23
Abstract:Quantitative micro-elastography (QME) is a compression-based optical coherence elastography technique enabling the estimation of tissue mechanical properties on the micro-scale. QME utilizes a compliant layer as an optical stress sensor, placed between an imaging window and tissue, providing quantitative estimation of elasticity. However, the implementation of the layer is challenging and introduces unpredictable friction conditions at the contact boundaries, deteriorating the accuracy and reliability of elasticity estimation. This has largely limited the use of QME to ex vivo studies and is a barrier to clinical translation. In this work, we present a novel implementation by affixing the stress sensing layer to the imaging window and optimizing the layer thickness, enhancing the practical use of QME for in vivo applications by eliminating the requirement for manual placement of the layer, and significantly reducing variations in the friction conditions, leading to substantial improvement in the accuracy and repeatability of elasticity estimation. We performed a systematic validation of the integrated layer, demonstrating >30% improvement in sensitivity and the ability to provide mechanical contrast in a mechanically heterogeneous phantom. In addition, we demonstrate the ability to obtain accurate estimation of elasticity (<6% error compared to <14% achieved using existing QME) in homogeneous phantoms with mechanical properties ranging from 40 to 130 kPa. Furthermore, we show the integrated layer to be more robust, exhibiting increased temporal stability, as well as improved conformity to variations in sample surface topography, allowing for accurate estimation of elasticity over acquisition times 3× longer than current methods. Finally, when applied to ex vivo human breast tissue, we demonstrate the ability to distinguish between healthy and diseased tissue features, such as stroma and cancer, confirmed by co-registered histology, showcasing the potential for routine use in biomedical applications.