Figure 1. This reticle meets USP 788 standards for particle inspection. It has four symmetric quadrants, with transparent and black reference circles and a linear scale graduated in 10µm increments. This figure is reprinted with permission, copyright 2009 United States Pharmacopeial Convention, all rights reserved.

The manual method of counting particulate matter has been used for decades. According to the government guidelines, a manual microscope particle counting system must include a circular diameter reticle (also known as a graticule) designed specifically for this purpose and calibrated using a certified stage micrometer. The reticle is symmetrically divided into four quadrants by crosshairs; each quadrant shows transparent and black reference circles 10µm and 25µm in diameter at 100x magnification.  The reticle also must contain a linear scale graduated in 10µm increments (Figure 1). 

The operator scans each entire filter or other standard-size substrate using the reticle, carefully comparing any particles seen under the microscope to the reference circles and keeping track of the number of particles of various specific sizes (more than 10µm and 25µm in diameter, respectively, for instance). Operators are then required to enter results into workbooks, create reports and file pertinent data into either electronic or paper folders.

Needless to say, detecting, measuring and keeping an accurate count of particles of various sizes in this manner can be tedious, time consuming, subjective and prone to human error.  For this reason, many pharmaceutical companies have moved to semi-automated or fully automated systems.

Semi-Automated Systems

Figure 2. Semi-automatic particle detection software categorizes particles according to size and morphology. All particles can be counted, measured and classified for further evaluation. Figures 2-7 courtesy of Olympus America Inc., Center Valley, PA, USA.

With of the addition of modern image-analysis software and high-resolution microscope digital cameras, semi-automated particle counting is now available for almost any microscope that has a standard camera mount (Figure 2). In a typical semi-automated configuration, images of foreign particles are acquired at various fields of view over the entire area of the filter. Image analysis software is then used to analyze each individual image.  First, each image is segmented by the image-analysis software into particles vs. non-particles, based on user-definable gray-scale threshold values. 

Next, using advanced image processing algorithms, morphological filters are applied to each image to automatically separate touching particles that would otherwise falsely be counted as one (Figure 3). Last, the operator selects the size parameter (area, diameter, equivalent circle diameter, feret, perimeter, etc.) or morphological parameter (aspect ratio, convexity, elongation, shape factor, circularity, etc.) of interest to be calculated on each particle (Figure 4). The user has the ability to set a size or morphological range (i.e. exclude any particles <10µm) and also can classify particle results into bins (i.e. Class 1: ≥ 10µm; Class 2: ≥ 25µm). 

Semi-automatic particle detection offers numerous advantages over counting and measuring particles manually. Perhaps the most significant benefit is the increased accuracy and repeatability gained by minimizing the variation that can creep in through manual examination of each filter.

Further, images and resultant data can be archived automatically into an integrated database. Operators can query this database at any time for both raw data and trends information that may affect quality assurance procedures or process control. Users or managers can generate electronic reports with the click of a mouse, automatically inserting particle images, spreadsheets and histograms. 

Fully Automated Microscope-Based Particle Counting

While it may take as long as an hour to perform a particle count on a 47mm filter with a manual microscope-based particle counting system, semi-automated image-analysis techniques can do it in a fraction of the time.

Even more impressive are fully automated microscope-based particle counting systems, which can complete such a task in as little as three minutes (Figure 5). New developments in hardware and software have brought such automated systems to the attention of pharmaceutical manufacturers, and they are now in place in many production facilities. While semi-automated counting relies on the operator manually positioning the XY stage underneath the microscope and then manually triggering image capture at various fields of view, the new systems automate the entire workflow, resulting in enormous savings in time and great gains in repeatability and precision. 

Figure 4.  A wide variety of measurement parameters can selected from software menus.

With a motorized XY scanning stage and specialized image-analysis software, a fully automated microscope-based particle counting system scans the entire area of the media to be examined (usually the area of a 25mm or 47mm filter), with the image-analysis software triggering image acquisition automatically at each frame. Since the image-analysis software uses application-specific algorithms to determine how many images are to be acquired over the area of the media (as well as the spatial XY position of each frame), the operator no longer needs to figure out how many, or from where, images are to be acquired.

Should a particle lie on the border between image frames, the software automatically reconstructs the particle among the image frames by carefully “stitching” it together from various segments, without inaccurately truncating the particle and/or counting as a duplicate. In a fully automated microscope-based particle counting system, focus is achieved at each frame via a high-performance, advanced laser-based autofocus module, which can perform accurate autofocus on high contrast samples within milliseconds.

As a cost-effective alternative to the highly sophisticated laser system, some fully automated image analysis software offers “Predictive Focus” algorithms, which are quite successful at compensating for any potential linear or non-linear variances in sample topography (Figure 6). Predictive focus works based on a series of in-focus points that the operator defines prior to the scan. Once that step is completed, the image-analysis software is able to calculate an extrapolated Z-axis position at every frame, which is then communicated back to the microscope, allowing the system to accurately predict points of focus across the filter media.

Figure 5. Fully automated systems like the Olympus BX51 particle inspection system automate the positioning of the XY stage along with entire measurement workflow, for enhanced productivity and greatly improved repeatability and precision.

Fully automated systems offer additional benefits even beyond these performance gains.  All relevant data including spreadsheets, histograms and reports can be generated in compliance with any company-defined or other predetermined standard (such as USP 788) (Figure 7). Data can be archived into an integrated database automatically, virtually eliminating the potential for data loss that can occur when operators forget to input data.

Administrators can customize and define database fields according to their specific experiment requirements and workflow. Later, users can search by any of these fields (such as experiment number, operator name, customer name, date range, etc.) to find important information or to look for patterns. In addition to such data management tools, some of the more advanced automated microscope-based particle-counting systems also are designed to comply with 21 CFR Part 11 guidelines concerning electronic records and electronic signatures. 

Though their capabilities are immense, operating the newest fully automated systems is quite simple. In a typical pharmaceutical manufacturing environment, an administrator or more advanced user sets up the system, pre-defining such experimental criteria as segmentation thresholds, scan areas, report templates and database settings. Non-qualified users can be locked out from adjusting these critical settings.

Figure 6. Predictive focus algorithms calculate where focus should be; they provide highly cost-efficient alternatives to more expensive laser focus systems. By mapping where a series of known points are on the X, Y and Z axes, these systems extrapolate Z-axis positions for other points across the sample.

Users find that their interface is streamlined, intuitive and extremely user friendly. They simply put the filter media in a fixture under the microscope, enter a few key data points with the touch of a button, and press “Start.” In just moments, the automatic particle detection is complete, all the resultant data and reports are automatically generated, and all the data has been archived into the database.  

Optical microscopes are commonly used for particle counting to quantify the level of foreign particulate matter in orally inhaled and nasal drug products and injection liquids. With the addition of a modular, high-resolution digital camera, along with modern image-analysis software and a motorized microscope XY scanning stage, this process can be semi-automated or even fully automated, significantly increasing throughput, accuracy and repeatability.

Figure 7. Sample USP 788 histogram.

Further, relevant data and reports can be generated automatically and archived electronically.

Data is easy to analyze for enhanced decision-making during the production process, and is produced in compliance with either company-defined or 21 CFR Part 11 standards. With these developments, pharmaceutical companies are increasingly looking at fully automated systems to help streamline the analysis, production and verification of manufacturing procedures and standards for these types of pharmaceutical products.

About the Author

Mario Gislao is Applications Specialist, Scientific Equipment Group, Industrial Microscopes, for Olympus America Inc. He can be contacted at [email protected].

References

1.      Thomas A. Barber. Control of Particulate Matter Contamination in Healthcare Manufacturing.  Interpharm Press, 2000, pp 115-129. http://books.google.com/books?id=1F1Rs_KnH3AC&pg=PA116&lpg=PA116&dq=circular+diameter+graticule&source=bl&ots=vzsF_v2RxV&sig=reqKsqp_cz0yXTI0xsXoAZ8mCJs&hl=en&ei=3kbQS_HTLoH58Aa2w8maDw&sa=X&oi=book_result&ct=result&resnum=2&ved=0CA8Q6AEwAQ#v=onepage&q=circular%20diameter%20graticule&f=false

2.      James Blanchard, et al. “Foreign Particles Testing in Orally Inhaled and Nasal Drug Products,” Pharmaceutical Research, Vol. 21, No. 12, December 2004 (© 2004)

3.      United States Pharmacopeial Convention.  <788> Particulate Matter in Injections http://www.usp.org/pdf/EN/USPNF/788ParticulateMatter.pdf.

4.      Lee M. Nagao, et al. “Aspects of Particle Science and Regulation in Pharmaceutical Inhalation Drug Products,” Crystal Growth & Design, 2005, 5 (6), pp 2261–2267.