Breaking with tradition

Fully automated image analysis extends particle characterisation beyond its limits of just measuring size

Figure 1: Close up of the sample dispersion unit

TECHNOLOGY developments in imaging and data analysis for particle characterisation continue to open up new areas of application for what is essentially a highly sophisticated microscope-based technique. Throughout the manufacturing industry there is a growing demand to extend particle characterisation beyond the traditional measurement of only size. Now full automation and minimal user interaction mean that image analysis based techniques deliver reliable and reproducible, statistically significant size and shape information, as well as providing microscope-quality images for every particle. Taking the Morphologi G3 as an example, this article looks at recent advances and their implications for further extending the use of this technique, including a new application for foreign particle detection.

Historically even very small differences in particle size or shape can have a significant effect on the properties and performance of materials. Automated techniques for measuring particle size and size distribution have been with us for many years, but until quite recently the only practical way to determine particle shape was using manual microscopy. This has always suffered from being both labour-intensive and subject to operator influence - measuring a statistically significant number of particles is almost impossible. The advent of image analysis based techniques has delivered a highly effective means of measuring both particle size and shape that is complementary to established sizing techniques such as laser diffraction. So, not only does it provide additional information, it also offers a means of validating other methods.

Image analysis captures a 2-dimensional image of a 3-D particle, from which it calculates various size and shape parameters. Circle equivalent diameter, whereby the captured 2-D image is converted to a circle of equivalent area, is often used to calculate particle size. For particle shape, many parameters may be used to build a complete picture. Calculating multiple shape parameters for every particle, and generating distributions for each, allows the identification and quantification of even the most subtle differences.

Analytical techniques must generally meet a number of requirements to be truly suited for routine use. Such attributes include automated procedures that minimise user involvement and eliminate operator bias, high throughput with measurement of statistically significant numbers of particles, and rapid, meaningful data analysis and reporting. For particle characterisation by image analysis, developments in automated sample preparation, measurement driven by standard operating procedures (SOP), and powerful data management and analysis software are all making a contribution.

Sample preparation
Fast, consistent sample preparation, normally to a glass slide for measurement on an x-y stage, is an absolute necessity. Sample preparation must be reproducible to ensure repeatable results, and the prepared samples have to be homogeneously distributed so that a subsection can be analysed knowing that it is representative of the entire sample. Together these add up to shorter measurement times and increased throughput, both critically important for routine use.

While particle characterisation by image analysis accommodates a variety of sample types, one of those most commonly encountered, dry powders, poses a particular preparation challenge, requiring strict control of dispersion conditions. One solution, implemented in the Morphologi G3, is the integration of a sample dispersion unit (SDU) within the particle characterisation system itself (Figures 1 and 2). The software-controlled SDU disperses sample with an instantaneous pulse of compressed air. Precise control of dispersion pressure, injection and settling times ensures highly reproducible measurements across a variety of sample types. Sample preparation times are reduced and measurement repeatability improved in comparison with conventional manual preparation methods. In addition, since measurements are made in an enclosed sample carrier, environmental exposure is minimised. As one key area for this type of analysis is the pharmaceutical industry, the SDU is aiding safer materials handling for substances such as pharmaceutical actives.

SOP-driven operation
Consistent and transferable methods that are independent of the operator and developed for specific analyses are critical to the consistent generation of reproducible and comparable results. When SOPs lock down all aspects of the measurement process, including hardware configuration, analysis settings, results, filtering, classification and parameter reporting, the measurement is run simply by selecting the desired SOP from the system menu. This removes the possibility of operator error in making measurements and contributes significantly to the reliability and repeatability of the data. Taking out the element of operator error, means that the technique can be confidently applied to even the most demanding measurements.

Data management and analysis
Automated image analysis provides a vast amount of data on different

Figure 2: Morphologi G3 showing the integrated sample dispersion unit

morphological parameters for every single particle measured. While all this makes possible the objective characterisation of changes in a product or process, sophisticated analysis, presentation and reporting software is needed to translate it into usable information.

The task of deciding which morphological parameter is most appropriate to use - for example when looking for similarities and/or differences between samples - is simplified with the use of the latest data comparison tool. This allows the rapid comparison of a large number of data sets in terms of all the morphological parameters measured, and indicates which varies most. Figure 3 shows an example of how the tool then clusters the data sets; in this case according to how similar they are, enabling the user to make informed decisions about pass/fail criteria.

A typical application might be the comparison of different batches of materials. Often it is important to identify unsatisfactory batches before they can enter a manufacturing process. In pharmaceuticals, for example, the ability to screen out batches of excipient that may fail in a tableting process can prevent wastage of expensive active ingredient. The challenge is finding a measurable parameter that identifies batches likely to fail in the process so they can be rejected at an early stage. Comparing many data sets quickly using the comparison tool allows the identification of such a parameter.

Being able to compare and cluster in order to find differences or similarities between multiple measurements is invaluable. Clear visualisation of measurement data, the facility to plot scattergrams using any size and shape measurement, to filter on any parameter and to group and classify information, are all tools that help focus on the information that is significant for a specific measurement or process.

A new application for automated image analysis systems is in characterising

Figure 3: 25mm filter holder

foreign particles, traditionally the preserve of manual light microscopy. Image analysis now enables the fully automated detection, enumeration and size classification of foreign particulate matter (FPM) collected on a filter.

In many industries there is a need to develop and quality control testing procedures that involve the enumeration of FPM. These range from the classification of particulate contaminants in fluids such as hydraulic transmission and fuel injection systems, to the automated control of particulate contamination in injectable solutions.

For the pharmaceutical industry such testing is particularly important with the US Pharmacopoeia specifying acceptable levels of FPM in injectable drugs and the FDA’s guidance documentation providing regulatory recommendations for foreign particle testing in Metered Dose Inhalers (MDIs), Dry Powder Inhalers (DPIs) and Nasal Spray and Inhalation Solution, Suspensions and Spray Drug Products. Testing for FPM in Orally Inhaled and Nasal drug products (OINDPs) is also discussed by the International Pharmaceutical Aerosol Consortium on Regulation and Science (IPAC-RS).

Here FPM are contaminant particles that can typically derive from the active or

Figure 4: The dendogram chart displayed in the comparison tab which clusters similar batches together in this case according to the solidity parameter

excipient material or from the administration device, for example from abrasion or shearing during activation. They can include many different materials, including glass, transparent synthetic fibres, stainless steel, rubber, aluminium and plastic particles.

On the Morphologi G3, several different sample carrier plates are available for mounting a prepared filter on to the system. Essential for automated analysis is their ability to hold the filter paper stretched flat. Figure 4 shows the 25 mm filter holder, which allows two filters to be presented and analysed successively with no manual intervention. The application is fully controlled through the system software. FPM are detected, down to 2 µm in size, in a short time, and are automatically classed into specified size brackets.

Particle characterisation is critically important across a diverse range of manufacturing industry, from R&D through to online analysis during manufacture. Driven by pressures for efficiency and profitability, and in the pharmaceutical industry by regulatory and safety issues, there is a need for increasingly sensitive and precise information. The growing recognition that measuring both particle size and shape distribution of many materials provides valuable and discriminating information, has fuelled the development of a new generation of automated analysers, the best of which provide size and morphological information through imaging particles and generating a variety of quantitative shape data. Perhaps most importantly, their full automation enables both high throughput sample preparation and the rapid analysis of extremely large numbers of particles. With the latest technology has come a range of new applications, such as the one described here for foreign particle characterisation. We can expect further extension of the range of applications as routine use of this type of automated analysis becomes increasingly widespread.