New Dimensions in Tablet Imaging

Like most of the world, the scientific community is now more than ever reliant on images to gain valuable information. When pixels replace pens, and the image of a new drug candidate can make or break a project, it becomes clear that a picture can be worth much more than 1000 words.

(BACKGROUND: PHOTOS.COM, IMAGING PHOTOS: CHEMIMAGE)

Uniting single-point spectroscopy and digital imaging, chemical imaging is slowly being adapted for the analysis of tablets and capsules. Advanced from traditional optical microscopy, which uses a material's refractive index among other material properties as the basis for generating image contrast, chemical imaging instead uses the underlying spectroscopy associated with the materials being analyzed at each spatial location as a means to produce image contrast. Multivariate analysis and chemometric software then translate this chemical image information into useful quantitative data.

The motivation for adapting chemical imaging lies primarily in the changing nature of today's dosage forms. "Definitely the business is moving in the direction of more complicated formulations, and the standard tools don't provide the information needed to characterize these products," says Linda Kidder, product manager, Chemical Imaging Systems, at Malvern Instruments, Inc. (Columbia, MD). "The single-point spectroscopies just can't do it, and HPLC [high-performance liquid chromatography] just can't do it. This is where the need is growing, and most of the companies working on advanced formulation, really understand this need."

Applications

The advantage of chemical imaging in solid-dosage form analysis is the ability determine the distribution and size of the active pharmaceutical ingredients (APIs) and excipients. The information is then correlated to optimize process operations and understand how ingredients interact within the tablet. "This information aids in quality control, trying to help understand how things dissolve, and their lifetimes with respect to shelf lives," says Richard Bormett, PhD, business manager at Renishaw (Hoffman Estates, IL). "Tableting processing can change the crystal form of the API, and being able to map where the API is and discriminate between its polymorphic forms is an important tool."

Previous methods of viewing the distribution didn't offer the quality that analysts needed. The time required to collect the data was long, making it not feasible, and high spatial-resolution mapping could take days. As Bormett observes, "If you looked at a whole tablet at a time, the spatial resolution of the probe became poorer and poorer so you had to look at larger and larger sections. If an API reached 5 to 1 microns, it wasn't always easy to see or determine its distribution."

Variations. The primary distinction between chemical imaging systems is the type of spectroscopic technique. Within the pharmaceutical industry, numerous spectroscopies have advanced, including Raman, mid-infrared (mid-IR), near-infrared (NIR) absorption–reflectance, UV–vis absorption–reflectance and luminescence. Each have been demonstrated useful for many applications.

Vibrational techniques such as Raman spectroscopy can pick up the unique chemical signatures for polymorphs of an API, which help analysts determine the cause of poor efficacy. Imaging with Raman spectroscopy is also useful in contamination identification. "In the world of tablet or capsule analysis, we often have the luxury of knowing what is present in the tablet based on a list of ingredients," says Matthew Nelson, PhD, director of application science at ChemImage (Pittsburgh, PA). "Often, this information allows us to develop a spectral library of the compounds that are present and to understand the chemical signatures for those components. We can collect data on small parts of a tablet, the whole tablet, or multiple tablets using chemical imaging. With data processing software, we can provide qualitative and quantitative information" (see Figure 1).

Figure 1: Data processing software allows analysts to obtain quantitative information about a whole tablet or portions of a tablet.

Chemical imaging using NIR spectroscopy can obtain data from rounded surfaces, powders, granules, beads, and intact capsules. "With NIR spectroscopy, we can look at beads in a capsule without having to take a capsule apart, it would be very difficult to do that with the mid-IR or Raman," explains Linda Kidder.

"NIR, mid-IR, and Raman can tell formulators whether an API is co-localizing with a particular excipient," adds Kidder. "[Formulators can determine] whether the API has a tendency to associate or aggregate with a particular excipient, which may be affecting the tablet's dissolution properties. You can't get that [information] with HPLC and you can't get it with single-point spectroscopy. That is absolutely unique to imaging."

Another common application of NIR, mid-IR, and Raman imaging is mapping thicknesses of coatings and (for Raman) of pigments (see Figure 2). This analysis can be performed two ways. First, imaging can be a primary technique; that is, no correlation with other data is necessary. The tablet is cut in half and the cross section is imaged. The coating is seen as a series of pixels and pixel size variation relates to coating thickness. "We can generate an image through the use of color and show where the thin spots on the tablet are. Because it's imaging, you can actually tell that the coating is 100 μm. And by looking at hundreds of locations, you can obtain a statistical representation of the coating distribution—the mean value—and a standard deviation in that value," explains Kidder.

Figure 2: An RGB (red/green/blue) composite generated from a Raman Chemical Image of a controlled release system. The image was taken on a "Falcon II Raman chemical imaging system (ChemImage). The green and red areas of the image indicate chemically different layers (or coatings) on a bead (indicated by in blue). The goal of the project was to study the uniformity of thickness of multiple controlled-release coatings. This image clearly indicates that the coatings are not uniform in thickness across the bead's surface.

It is also possible to collect data from the surface of a tablet without first taking a cross section. A tablet is placed on its bottom or its top, and the analyst looks down on the surface. The variation in intensity of the vibrational band associated with the coating across the surface of the tablet provides a statistical variation of the coating thickness. This is a secondary analytical technique, and therefore obtaining absolute (rather than relative) values of the coating thickness requires a correlation with real distance measurements.

Tools

For NIR techniques, there has been some confusion between the terms "chemical imaging" and "mapping" because both provide an image and generate the same data. The difference is in how an image is produced and data are collected. "Chemical imaging uses a two-dimensional detector, so you are taking pictures, whereas mapping or 'fast mapping' uses either linear arrays or collects a line of spectra at one time to build up an image while the sample is moving on a stage," explains Kidder. In chemical imaging, the sample does not move—a picture is taken over a series of wavelengths—whereas mapping uses an interferometer, which is based on a moving mirror. This differentiation is also found in both Raman and mid-IR, where both mapping and imaging are used.

Because NIR mapping is based on an interferometer, sample surfaces must be very flat to collect good data. In contrast, "the depth of focus for NIR chemical imaging is very forgiving," says Kidder. "We can look at rounded sample and get reasonably good data from the whole sample" (see Figure 3). In addition, Malvern has developed an NIR-based imaging system that has no moving parts. As a stable system, it can move from an R&D setting into QA/QC setting or manufacturing setting and collect data. Mapping systems based on an interferometer have moving sample stages and the instrument stability and therefore spectral quality may be compromised by the vibrations found in a manufacturing or QA/QC environment.

Figure 3: NIR chemical image reveals granule distribution in tablet (wavelength = 2060 nm) IMAGE COURTESY OF MALVERN INSTRUMENTS INC.

Tablet manufacturers seeking vibrational spectroscopy-based chemical imaging use three factor when choosing between Raman, mid-IR, NIR—and in some cases fluorescence—methods: specificity, sensitivity, and spatial resolution.

Patrick Treado, PhD, chief technology officer at ChemImage explains that in general, fluorescence is the most sensitive approach but the least specific, and spatial resolving power can be on the order of a couple hundred nanometers. NIR is reasonably sensitive but also relatively nonspecific, and resolution is on the order of tens of microns. NIR is the most inexpensive, however, and provides fast data collection. Raman is not as inexpensive, though cost effective, but it's very specific although a little less sensitive. Moreover, explains Bormett,"IR is limited in its lateral spatial resolution. It's hard to perform IR imaging with lateral spatial resolutions better than about 30 microns."

NIR. "NIR spectroscopy has been seen to be a workhorse in industrial technology, whereas the Raman and the mid IR are seen more as specialist techniques," says Kidder. In some ways this parallels on the imaging side as well. "NIR can probably solve 85% of the problems we see for the pharmaceutical industry. Some of our customers who have multiple technologies tend to go to NIR first and determine whether they can solve it with NIR because of its fast data collection time time (just over a minute for a standard tablet), no sample preparation required, and ability to accommodate different types of samples. For samples that require higher spatial resolution or higher spectral selectivity though, other techniques may be needed." As with the single-point technique, chemical imaging with NIR and mid-IR spectroscopies do not work well with inorganic compounds.

Raman. Raman spectroscopy, however, can accommodate inorganic compounds and recent advances are extending its application. ChemImage, for example, has innovated a dynamic chemical imaging capability. "Historically if you are going to use Raman to evaluate an entire tablet or a process that is occurring in real time, it took a long time to gather the information," says Treado. "Our approach has been to capture images dynamically in real time or near real time, in the time scale it is relevant to the process you are trying to monitor. Another way of thinking of this is Raman imaging 'movies' —it is drilling down and extracting information content from the monitoring of these dynamic processes." Capabilities include monitoring processes that can't be slowed down (through temperature control for example) and the kinetics of the process is not on the time scale of seconds or a few minutes. "Crystallization, dissolution, and recrystallization are specific application where dynamic chemical imaging is proving itself," says Treado.

Dynamic chemical imaging with Raman spectroscopy uses wide-field imaging. "Typically this means the use of wide-field illumination in combination with liquid crystal imaging spectrometers, electron multiplying CCD detectors, and a high degree of automation in the software to allow the instrument to monitor a process in real time or near real time in an unbiased fashion," says Treado.

Raman mapping techniques also have advanced. Renishaw, for example, has developed a Raman spectrometer that is designed for high-speed imaging of small parts or a whole tablet. For Raman-based instruments measure individual spectra in a "raster mode" on the tablet. Rows of data are collected to create a two-dimensional map. "Ten to 15 years ago when the pharmaceutical industry started doing this kind of mapping on a tablet, it might take from 8 to 12 hours to obtain," says Bormett. "Providing a tool that can create images for these same tablets in tens of minutes, at an hour at the longest, is a phenomenal tool for the industry."

Innovation drivers

Helping to push the industry into adopting chemical imaging into its analytical labs are developments in both hardware and software.

Hardware. New types of electronic packages have been developed for spectrometers' data acquisition software. "Historically Raman spectrometers were put together with piece components that were just bought off the shelf," explain Bormett. "You would buy a spectrometer, gratings, and a detector. Once you decided exactly how these pieces fit and function together, you could customize the electronic packages in each of these components to allow high-speed data acquisition." Renishaw customized its electronic packages that go into the CCD to control stage and sample movement, then synchronized it with spectrometer data acquisition. "The individual pieces have been always there, but they have never been able to be packaged together until the last couple of years," says Bormett.

Data transfer also has evolved. "In our initial products, the camera spoke to a frame grabber which then dumped the data," says Kidder. "Now our cameras connect through gigabyte Ethernet cables, enabling data to transfer more quickly." Kidder also explains that the speed and duty cycle of the cameras have increased. Whereas earlier systems would collect 80,000 spectra in 4 minutes, for example, data can now be collected in about a minute with the same signal-to-noise properties. "You may wonder if [data collection speed] matters," says Kidder. "One of the interesting cases is when you are doing QbD [quality by design] and your DOE [design of experiments] includes 100 different tablets, for example. If it takes you 40 minutes to collect the data, you have to multiply that by100, compared with 10 minutes or even 1 minute for data collection. It really starts to make a difference if the technique is used as a routine tool."

Treado attributes the innovations in spectrometers, specifically the liquid crystal imaging spectrometer. He also credits the use of imaging cameras or focal-point arrays. In IR, it's typically indium gallium arsenide, indium antimonide, or mercury cadmium telluride cameras. For Raman- and fluorescent-based systems, it's CCD or CMOS detectors. "If you couple those with multivariate software, you have powerful combination," he says.

Software. As the tools to analyze data advance, so too should the tools to increase the understanding of the information that is generated. "Early on, people would pull out a single wavelength image without having properly preprocessed their data with a baseline correction or normalization, for example," says Kidder. "These steps were standard for single-point spectroscopy, but somehow analysts lost track of the fact that imaging was really just parallel spectroscopy, and the basics still had to be applied. There have also been strides in the quantitative , versus qualitative, assessments that can be made. People used to look at a picture and compare it with another picture and say 'this looks more homogeneous.' Obviously that doesn't work, and it doesn't help you."

Data analysis tools are now much more sophisticated, and researchers are developing metrics to see an image from a statistical point of view. "The 'standard deviation' of a image can actually give you a quantitative measure of the heterogeneity of a tablet, and you can then start putting the image to good quantitative use," says Kidder. "In our early days, most of our reports were filled with pictures. Now, they're mostly filled with tables of numbers comparing the statistics of our results.The ability to generate quantitative, objective, and reproducible results allows imaging to become an analytical workhorse, rather than just a fun tool."

The big picture

Chemical imaging companies see real promise in applying their technologies to the tableting market and are eyeing validation. "The industry is struggling with the best way to validate these techniques, but we have started working with customers in QA/QC departments to get them accepted," says Kidder. "Large pharmaceutical companies want to roll chemical imaging out to replace or complement wet chemistry techniques." Although single-point spectroscopies methods have been validated for some time, the respective imaging technique will still take considerable revalidation work.

But the promises that imaging may bring can make that the revalidation work worthwhile. One future application, for example, includes correlating an image with a dissolution curve. "If they can prove this correlation—which has been done for a variety of products already—and because it is a fast technique, we could look at a lot of tablets in a batch to make sure they have the correct 'image parameter,'" says Kidder. The parameter could indicate whether the tablets were good dissolution candidates and wet chemistry would only be used to sample a small subsection of that. Regular dissolution testing would still be done on a subset of samples, but more if not all samples could be tested using vibrational imaging. "It's definitely cutting edge ... that is the big picture people are looking at."