Data underlying the publication: A novel method to assess crosstalk in single-grain luminescence detection

For more information please see the README file in the "Data-4TU" folder which contains all scripts, data, and a manual.<br># Paper abstract <br> Luminescence imaging with an EMCCD camera is an innovative approach to single-grain luminescence detection. In theory, it allows for the random placement of grains on a disc, but in practice, attributing photons to specific grains remains unachievable. Currently, single-grain discs with fixed grain placement are primarily used, though these are still susceptible to crosstalk, e.g. overlapping of luminescence signals from adjacent grains. A standardized method for identifying, correcting, or eliminating crosstalk is not yet available. In this paper, we introduce Moran’s I, a measure of spatial autocorrelation, as a novel method for assessing crosstalk in single-grain luminescence detection. Using simulations with induced crosstalk, we demonstrate that Moran’s I can effectively detect crosstalk when interpreted within the context of its pseudo-p value. These results are corroborated by those obtained on a measured luminescence dataset for which crosstalk is deliberately enhanced by increasing the region of interest (ROI) selected for luminescence signal integration. These results on simulated and measured data show that Moran’s I can be used to select optimal ROI size to enhance signals whilst limiting crosstalk. Finally, we examined the effect of modelled crosstalk on three types of equivalent-dose distributions, finding that normal distributions are unaffected by crosstalk, whereas bimodal distributions became mixed, and the shape of skewed distributions alters. These results have implications for ROI selection for well-bleached, mixed and heterogeneously bleached sediments. <br># Measurements <br>+ Luminescence measurementsAll measurements were made on a Risø TL/ OSL DA-20 reader with an automated detection and stimulation head (DASH) and an EMCCD camera (Kook et al., 2015). The system is fitted with a calibrated 90Sr/90Y beta source which delivers a dose rate of approximately 0.0954 Gy/s to grains on a single-grain disc. The sample was stimulated with IR LEDs at 850 nm and a blue package filter (BG-3 (3.0 mm) + BG-39 (4.0 mm)) was used to target the 410 nm K-feldspar emission peak. We adopted a multi-elevated-temperature (MET) pIRIR measurement protocol with IRSL read-outs at 50, 110, 170, and 230 ◦C (Table 1) (Li &amp; Li, 2011). We measured three single-grain discs per sample, which makes a total of 45 discs. We chose to only study crosstalk effects on the IRSL-50 signal for the net natural test dose (8.1 Gy) response (Table 1, step 12 manuscript), which carries information about the luminescence sensitivity of the grains. The net natural test dose response (Net_TnTx ± Net_TnTx.Error) was alculated by subtracting the background test dose response (TnTx.BG) integrated over channel 210 up to 260 (20 s) from the initial test dose response (TnTx) integrated over channel 11 up to 31 (8 s) to create a single-grain brightness distribution (Wallinga, 2002). The measured data were analyzed with Risø Viewer+ software and extracted for nine different ROI sizes: 300, 450, 600, 750, 900, 1050, 1200, 1350 and 1500 μm, i.e. ranging from the hole dimension (300 μm) up to overlapping with the adjacent grain hole (1500 μm). The spacing between grain positions on the single-grain disc is 600 μm. We assume that grains of varying sensitivity are randomly distributed over the single-grain disc grid. Subsequent analyses were performed within the R environment for statistical computing. <br>Please visit the Netherlands Centre for Luminescence dating (NCL) database for more information on the sample background at: https://www.lumid.nl/.  On the website go to the 'projects' tab and search for NCL-code 1321.<br>+ Measurement protocols and data formats The measurement protocols were written in Riso's sequence editor software (.seq files). All Riso software is downloadable at: https://www.fysik.dtu.dk/english/research/radphys/research/radiation-instruments/tl_osl_reader/software. The raw data is stored in a Riso .binx file which can be opened in Riso's analyst software environmen. We analyzed the raw data in Riso's Viewer+ to extract the luminescence signal for several Region Of Interests (ROIs). <br>+ Processing and analysis scripts in R All processing of data (modifications, calculations, etc) and analysis of data was performed in R. All  scripts are commented. Per measurement type an R script was written.  <br>+ Figures and tables All figures and tables were created through R and/ or Adobe Illustrator. Refer to the R scripts for detailed information within the script.

如需获取更多信息，请参阅"Data-4TU"文件夹中的README文件，该文件夹包含所有脚本、数据集与使用手册。
# 论文摘要
采用电子倍增电荷耦合器件（Electron Multiplying Charge-Coupled Device，EMCCD）相机的发光成像技术，是单颗粒发光检测的创新方案。理论上，该技术可实现颗粒在圆盘上的随机排布，但实际操作中，仍无法将光子归因至特定颗粒。目前主流采用颗粒排布固定的单颗粒圆盘，但此类圆盘仍易出现串扰问题，例如相邻颗粒的发光信号发生重叠。目前尚未出现可标准化识别、校正或消除串扰的方法。
本文提出将空间自相关度量莫兰指数（Moran’s I）作为一种新方法，用于评估单颗粒发光检测中的串扰问题。通过引入人为串扰的仿真实验，本文证明当结合伪p值进行解读时，莫兰指数可有效检测串扰。本研究通过刻意增大发光信号积分的感兴趣区域（Region of Interest，ROI）以强化串扰，对实测发光数据集开展分析，所得结果与仿真结论一致。针对仿真与实测数据的分析结果表明，莫兰指数可用于筛选最优ROI尺寸，在提升信号强度的同时限制串扰。
最后，本文探究了建模串扰对三类等效剂量分布的影响，发现正态分布不受串扰影响，双峰分布会发生混合，而偏态分布的形态则会发生改变。上述结论可为完全漂白、混合及非均匀漂白沉积物的ROI选择提供参考。
# 测量方案
+ 发光测量
所有测量均在搭载自动检测与激发头（DASH）及EMCCD相机的Risø TL/OSL DA-20型阅读器上完成（Kook等，2015）。该系统配备经校准的90Sr/90Yβ放射源，可向单颗粒圆盘上的颗粒提供约0.0954 Gy/s的剂量率。实验样品采用850 nm红外发光二极管（IR LEDs）激发，并搭配蓝色组合滤光片（BG-3(3.0 mm)+BG-39(4.0 mm)），以对准410 nm处的钾长石发射峰。
本研究采用多升温温度（MET）pIRIR测量方案，在50、110、170及230 ℃下开展红外释光（IRSL）读出（见表1）（Li & Li，2011）。每个样品设置3片单颗粒圆盘，总计45片。
本研究仅针对净自然测试剂量（8.1 Gy）响应下的IRSL-50信号开展串扰效应研究（见表1，论文步骤12），该信号可反映颗粒的发光灵敏度。净自然测试剂量响应（Net_TnTx ± Net_TnTx.Error）的计算方式为：从通道11至31（时长8 s）积分得到的初始测试剂量响应（TnTx）中，减去通道210至260（时长20 s）积分得到的背景测试剂量响应（TnTx.BG），以此生成单颗粒亮度分布（Wallinga，2002）。
实测数据采用Risø Viewer+软件进行分析，并提取9种不同ROI尺寸下的信号：300、450、600、750、900、1050、1200、1350及1500 μm，覆盖范围从孔穴尺寸（300 μm）直至与相邻颗粒孔穴发生重叠（1500 μm）。单颗粒圆盘上的颗粒位置间距为600 μm。本研究假设不同灵敏度的颗粒随机分布于单颗粒圆盘网格之上，后续分析均在R统计计算环境中完成。
如需了解样品背景的更多信息，请访问荷兰发光测年中心（NCL）数据库：https://www.lumid.nl/。进入网站后点击“项目”标签，搜索NCL编码1321即可。
+ 测量协议与数据格式
测量协议通过Risø序列编辑软件编写（.seq格式文件）。所有Risø软件均可从以下网址下载：https://www.fysik.dtu.dk/english/research/radphys/research/radiation-instruments/tl_osl_reader/software。
原始数据存储于Risø .binx格式文件中，可通过Risø分析软件环境打开。本研究通过Risø Viewer+软件对原始数据进行分析，提取不同感兴趣区域（ROIs）的发光信号。
+ R语言处理与分析脚本
所有数据处理（包括修改、计算等）与数据分析工作均在R语言环境中完成，所有脚本均带有注释，且针对每类测量方案编写了专属R脚本。
+ 图表与表格
所有图表均通过R语言及/或Adobe Illustrator绘制，详细信息可参阅对应R脚本。