Dankook University, Beckman Laser Institute Korea, School of Medicine, Cheonan, Republic of Korea.
Dankook University, Department of Biomedical Engineering, Cheonan, Republic of Korea.
J Biomed Opt. 2023 Jul;28(7):075005. doi: 10.1117/1.JBO.28.7.075005. Epub 2023 Jul 31.
Light emitting diodes (LEDs) are commonly utilized for tissue spectroscopy due to their small size, low cost, and simplicity. However, LEDs are often approximated as single-wavelength devices despite having relatively broad spectral bandwidths. When paired with photodiodes, the wavelength information of detected light cannot be resolved. This can result in errors during chromophore concentration calculations. These errors are particularly apparent when analyzing water and fat in the 900 to 1000 nm window where the spectral bandwidth of LEDs can encompass much of the analysis region, resulting in intense crosstalk.
We utilize and present a spectral correction (SC) algorithm to correct for the spectral bandwidth of LEDs. We show the efficacy using a narrowband technique of spectrally broad and overlapping LEDs.
Narrowband diffuse reflectance spectroscopy (nb-DRS), a technique capable of quantifying the hydration ratio () of turbid media, was utilized. nb-DRS typically requires a broadband light source and spectrometer. We reduce the hardware to just five LEDs and a photodiode detector, relying on SC to compensate for spectral crosstalk. The effectiveness of our SC approach was tested in simulations as well as in an emulsion phantom and limited selection of human tissue.
In simulations, we show that calculated errors increased with the spectral bandwidth of LEDs but could be corrected using SC. Likewise, in emulsions, we found an average error of 8.7% (maximum error 14%) if SC was not used. By contrast, applying SC reduced the average error to 2.2% (maximum error of 6.4%). We show that despite utilizing multiple, spectrally broad, and overlapping LEDs, SC was still able to restore the performance of our narrowband method, making it comparable to a much larger full broadband system.
发光二极管（LED）由于其体积小、成本低、结构简单等特点，常用于组织光谱学。然而，尽管 LED 的光谱带宽相对较宽，但它们通常被近似为单波长器件。当与光电二极管结合使用时，无法解析检测光的波长信息。这可能导致在计算发色团浓度时出现误差。当分析 900 到 1000nm 窗口中的水和脂肪时，这些误差尤其明显，因为 LED 的光谱带宽可以涵盖分析区域的大部分，从而导致强烈的串扰。
我们利用和提出了一种光谱校正（SC）算法来校正 LED 的光谱带宽。我们使用光谱较宽且重叠的窄带 LED 展示了该算法的效果。
窄带漫反射光谱（nb-DRS）是一种能够量化混浊介质水合比（）的技术，我们使用了这种技术。nb-DRS 通常需要宽带光源和光谱仪。我们将硬件简化为仅使用五个 LED 和一个光电二极管探测器，并依靠 SC 来补偿光谱串扰。我们在模拟中以及在乳液体模和有限的人体组织选择中测试了我们的 SC 方法的有效性。
在模拟中，我们表明，计算出的错误随着 LED 的光谱带宽增加而增加，但可以使用 SC 进行校正。同样，在乳液中，如果不使用 SC，我们发现平均误差为 8.7%（最大误差为 14%）。相比之下，应用 SC 将平均误差降低到 2.2%（最大误差为 6.4%）。我们表明，尽管使用了多个光谱较宽且重叠的 LED，但 SC 仍然能够恢复我们的窄带方法的性能，使其与更大的全宽带系统相当。
