Micrographs with Cloud Computing
similar drug loading but different levels of crystallinity were imaged (Figures 1a,1b,1c). Conventional threshold and gra- dient-based segmentation faced numerous technical chal- lenges that were difficult to overcome. Upon a five-minute interactive training, the DigiM I2S AI image segmentation engine successfully segmented four material phases, namely, crystalline drug phase, amorphous drug phase, polymer phase, and porosity phase (Figures 1d,1e) on tens of thou- sands of images during an overnight cloud computing ses- sion. Te size distribution of pores and the crystallinity of the API (Figures 1f,1g), as well as the volumes and surface areas for all four phases, were computed and validated. Tese quantifications are difficult, or impossible, to obtain with other laboratory methods. Only a limited amount of data was shown here for both clarity and confidentiality reasons. Trough the DigiM I2S cloud, however, tens of thousands of
measurements of dozens of microstructure phases from tens of 3D datasets can be AI-segmented automatically and com- pared to provide insight into the effectiveness of how an API is rendered in an amorphous state. Tus, DigiM I2S enables an analysis project to be completed in weeks, that would oth- erwise require years of effort from the same analysts. Such approaches could potentially improve drug-development efficiencies with respect to time and material costs, as well as reduce the burden of animal and/or clinical screening studies. Image-based characterization data were also used to correlate formulation parameters with drug release perfor- mance through numerical simulations of effective diffusivity coefficients, disintegration patterns, and various designed release behaviors [4]. Geoscience. Tight rock, including shale, is not only
an important hydrocarbon resource, but it is of interest in environmental research con- cerning underground stor- age of greenhouse gas and nuclear waste. Te physical properties of these rocks, with sub-micron pore sizes and sub-millidarcy perme- ability, are difficult to study using physical laboratory core analysis. For example, in order to measure how hydro- carbons might flow through a rock sample, fluid is pushed through the sample via pres- sure. When the pores are small, excessive pressure will either damage the sample or deform the sample, altering the pore
space being mea-
Figure 3: Multi-scale 3D tomographs of a proton exchange membrane (PEM) fuel cell [7]. (a) A TEM tomograph image showing a nanometer-scale view within a catalyst layer of a PEM fuel cell. (b) One of the 385 x-y cross-section slices showing the same region reconstructed from the TEM tomograph. (c) Element map by X-ray spectrometry of the same area showing the Pt/C agglomerate with Pt (red), pores (blue), and ionomer (green). (d) A micrometer-scale view of a cross section the PEM fuel cell from an FIB-SEM tomographic dataset collected from the catalyst layer corresponding to (a). (e) Image segmentation results corresponding to the FIB-SEM image stack in (d), with white areas representing solid support material and black areas representing voids. (f) 3D reconstructed porous catalyst layer with light blue representing solid support material. (g-i) Millimeter-scale views of the PEM fuel cell assembly from X-ray MicroCT tomographs taken at three magnifications. The circles in (g) and (h) correspond to the approximate field of view in (i). Numbers in (i) indicate different layers of a typical PEM fuel cell assembly: 1– anode; 2 – separator; 3 – cathode, where TEM data corresponding to (a) and FIB-SEM data corresponding to (d) were collected; 4 – delamination; 5 – micro- porous layer; 6 – gas diffusion layer with polymer-coated carbon fibers.
2019 March •
www.microscopy-today.com
sured. Microscopy imaging, in combination with AI-based image processing and com- putational physics methods provided by DigiM I2S, offer great potential in character- izing these difficult rocks at reservoir temperature and pressure. A suite of charac- terization tools was developed in DigiM I2S and validated to correlate rock images with petrophysical properties of rocks such as relative perme- ability [5]. Raw SEM images were acquired on a Helios 660 FIB-SEM (Termo Fisher Scientific, Waltham, MA), and X-ray MicroCT images were acquired with a Versa 520 (Carl Zeiss Microscopy, Pleasanton, CA). Cloud com- puting and AI image pro- cessing make it possible to
31
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56
