Global Centre for Environmental Remediation (GCER), University of Newcastle, Callaghan NSW 2308, Australia; Cooperative Research Centre for Contamination Assessment and Remediation of the Environment (CRC CARE), University of Newcastle, Callaghan NSW 2308, Australia.
Global Centre for Environmental Remediation (GCER), University of Newcastle, Callaghan NSW 2308, Australia.
Water Res. 2021 Apr 15;194:116913. doi: 10.1016/j.watres.2021.116913. Epub 2021 Feb 8.
We recently developed the Raman mapping image to visualise and identify microplastics / nanoplastics (Fang et al. 2020, Sobhani et al. 2020). However, when the Raman signal is low and weak, the mapping uncertainty from the individual Raman peak intensity increases and may lead to images with false positive or negative features. For real samples, even the Raman signal is high, a low signal-noise ratio still occurs and leads to the mapping uncertainty due to the high spectrum background when: the target plastic is dispersed within another material with interfering Raman peaks; materials are present that exhibit broad Raman peaks; or, materials are present that fluoresce when exposed to the excitation laser. In this study, in order to increase the mapping certainty, we advance the algorithm to combine and merge multi-images that have been simultaneously mapped at the different characteristic peaks from the Raman spectra, akin imaging via different mapping channels simultaneously. These multi-images are merged into one image via algorithms, including colour off-setting to collect signal with a higher ratio of signal-noise, logic-OR to pick up more signal, logic-AND to eliminate noise, and logic-SUBTRACT to remove image background. Specifically, two or more Raman images can act as "parent images", to merge and generate a "daughter image" via a selected algorithm, to a "granddaughter image" via a further selected algorithm, and to an "offspring image" etc. More interestingly, to validate this algorithm approach, we analyse microplastics / nanoplastics that might be generated by a laser printer in our office or home. Depending on the toner and the printer, we might print and generate millions of microplastics and nanoplastics when we print a single A4 document.
我们最近开发了一种拉曼图谱成像技术，用于可视化和识别微塑料/纳米塑料（Fang 等人，2020 年；Sobhani 等人，2020 年）。然而，当拉曼信号较弱时，单个拉曼峰强度的图谱不确定性会增加，这可能导致图像出现假阳性或假阴性特征。对于实际样品，即使拉曼信号较强，由于目标塑料分散在具有干扰拉曼峰的另一种材料中，或者存在具有宽拉曼峰的材料，或者存在暴露于激发激光时会发生荧光的材料，仍然会出现低信噪比，这也会导致图谱不确定性。在本研究中，为了提高图谱确定性，我们改进了算法，将已经同时在拉曼光谱的不同特征峰上进行映射的多幅图像进行组合和合并，类似于通过不同的映射通道同时进行成像。这些多幅图像通过算法合并为一幅图像，包括颜色偏移以收集具有更高信噪比的信号、逻辑或以获取更多信号、逻辑与以消除噪声以及逻辑减以去除图像背景。具体来说，两个或更多拉曼图像可以作为“父图像”，通过选择的算法合并生成“子图像”，通过进一步选择的算法生成“孙图像”，依此类推，直至生成“后代图像”等。更有趣的是，为了验证这种算法方法，我们分析了可能在我们办公室或家中的激光打印机中产生的微塑料/纳米塑料。根据碳粉和打印机的不同，我们在打印单个 A4 文档时可能会生成数百万个微塑料和纳米塑料。
