Pérez-Alfonso Clara, Galipienso Nieves, Garrigues Salvador, de la Guardia Miguel
Laboratorio de la Inspección Farmacéutica y Control de Drogas de la Delegación de Gobierno en la Comunidad Valenciana, Área de Sanidad, Inspección de Farmacia, Muelle de la Aduana S/N, 46024 Valencia, Spain.
Department of Analytical Chemistry, University of Valencia, Research Building "Jeroni Muñoz", Dr. Moliner Street, 46100 Burjassot (Valencia), Spain.
Forensic Sci Int. 2014 Apr;237:70-7. doi: 10.1016/j.forsciint.2014.01.015. Epub 2014 Feb 7.
Direct determination of cocaine in untreated seized samples has been made based on diffuse reflectance measurements of the near infrared (NIR) radiation through samples contained inside standard glass vials. The method used a series of previously analyzed samples, by the reference gas chromatography method, to build a partial least squares calibration model which was validated using an independent set of samples. The use of a general model for samples containing from 11.38% till 86.44% (w/w) cocaine was based on the use of spectral ranges from 12500.7 to 10128.6, 9339.8 to 6967.7 and 5388.3 to 4597.6cm(-1) with previous first derivative and vector normalization data pre-processing and provided a root mean square error of prediction (RMSEP) of 4.0% (w/w) with a residual prediction deviation (RPD) of 3.9% (w/w), based on the use of 8 latent variables, 34 samples for calibration and an independent set of 44 samples for validation. The aforementioned results could be improved on considering two separate models, one for high concentrated bulk samples and another for samples diluted with cutting agents. Additionally a new set of batch samples with cocaine concentrations from 60% till 84% was evaluated by using the developed method.
基于对标准玻璃瓶内样品的近红外（NIR）辐射漫反射测量，对未经处理的查获样品中的可卡因进行了直接测定。该方法使用一系列先前通过参考气相色谱法分析的样品，建立了偏最小二乘校准模型，并使用一组独立样品进行了验证。对于可卡因含量从11.38%到86.44%（w/w）的样品，使用通用模型是基于使用12500.7至10128.6、9339.8至6967.7以及5388.3至4597.6cm⁻¹的光谱范围，并进行了一阶导数和向量归一化数据预处理，基于8个潜变量、34个校准样品和一组44个独立验证样品，预测均方根误差（RMSEP）为4.0%（w/w），残差预测偏差（RPD）为3.9%（w/w）。考虑两个单独的模型，一个用于高浓度大宗样品，另一个用于用稀释剂稀释的样品，上述结果可以得到改善。此外，使用所开发的方法对一组新的可卡因浓度从60%到84%的批量样品进行了评估。
