Taking in account the fact that krokodil contains

Taking in account the fact that “krokodil” contains a mixture of morphinans, several mobile phases described in the literature for thin-layer chromatography (TLC) analysis of morphinans were tested [9,10]. For the TLC analysis a mixture of hexane/ethyl acetate/diethylamine was selected as mobile phase (6:4:0.5). The chromatographic profile of “krokodil extract” with this mobile phase is shown in Fig. 3. The FTIR spectrum of “krokodil” showed an OTX-015 cost compatible with a phenolic hydroxyl (3382cm−1, m, PhO–H stretch), a group present in acetaminophen, desomorphine and morphinan-4,5-epoxy-3-ol (Fig. 4). The presence of vinylic carbons, which was already recognized in 1H NMR spectrum, is also suggested (3028cm−1, m, =C–H stretch). In accordance with the 1H NMR spectrum, FTIR spectrum also shows several bands compatible with the aromatic C=C stretching (1559cm−1, m; 1542cm−1, m; 1509cm−1, m; 1456cm−1, m).
Conflict of interest statement
Acknowledgments Emanuele Alves and Annibal D. Pereira Netto acknowledge Brazilian agency Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) (process 245844/2012-0) for research grants and scholarship. José Soares acknowledges Fundação para a Ciência e a Tecnologia (FCT) for PhD (SFRH/BD/98105/2013). Ricardo Dinis-Oliveira acknowledges Fundação para a Ciência e a Tecnologia (FCT) for his Investigator Grant (IF/01147/2013). Sara Cravo and Carlos Afonso acknowledges FCT through the strategic project CEQUIMED-UP (Pest-OE/SAU/UI4040/2014). Artur Silva thanks FCT/MEC for the financial support of the QOPNA research Unit (FCT UID/QUI/00062/2013), through national founds and, where applicable, co-financed by the FEDER, within the PT2020 Partnership Agreement, and to the Portuguese NMR Network.
Data Dataset names and attributes are briefly presented in Table 1. Fig. 1 shows two samples of the entire dataset including 2D SEM micrographs and 3D reconstructed surfaces. A Hitachi S-4800 Field Emission Scanning Electron Microscope is used to capture the micrographs. This microscope is equipped with a computer controlled 5 axis motorized stage capable of 360° of rotation with a tilt range of −5–70°. Sample manipulation, such as Z-position, tilt, and rotation of the stage, as well as image processing and capture functions are operated through the Hitachi PC-SEM software. The working distance that would give the required depth of focus is specified at the maximum tilt for every specimen at the magnification chosen for image capture. As the sample is tilted in successive 1° increments through the software, the image is centered manually by moving the stage in the x- and y-axes with the stage positioning trackball. The working distance and magnification are kept consistent in every captured image of the tilt series by changing the Z-axis position as required. Brightness and contrast are manually adjusted for consistency between micrographs, using the same structure in every image. The micrographs are acquired with an accelerating voltage of 3kV, employing the signals from both the upper and lower secondary electron (SE) detectors. Readers interested in SEM imaging are referred to [4,5] for further information. The 3D surface models and their construction strategy are fully detailed in the paper [2]. At present, the 3DSEM dataset includes three different samples illustrated in Table 1. This dataset is an ongoing project in which further samples will be added to the dataset by near future. As we mentioned earlier, the 3DSEM dataset is freely available at [1] for any educational, research, and academic purposes.
Experimental design, materials and methods 3D surface reconstruction from a set of 2D images employs several computational technologies, including multi-view geometry, computer vision, machine learning, and optimization strategies to tackle the inverse problem going form 2D images to 3D surface models [6,7]. The complete pipeline of our proposed optimized multi-view framework for 3D SEM surface reconstruction has six stages. At the first stage, a set of 2D SEM micrographs are taken by tilting the specimen across variant angles. The step requires SEM imaging styles, such as changing magnifications, tilting the specimen, employing SE or/and BSE detectors. We then detect the feature points in each 2D image in the set and estimate image motion based on a set of corresponding points. Once we are done with estimating the relative position of the images, the 3D position of all corresponding points will be reconstructed by linear triangulation [6–8]. The final step is doing a refinement process by defining a cost function for any set of parameters (e.g., SEM extrinsic parameters and 3D positions) as to whether this is a good or bad set and find the best fitness model in the set.