Many modern inspection and quality problems in pharma manufacture and supply can be solved through emerging vision technologies. Don Braggins, visions systems consultant and member of the UK Industrial Vision Association, sheds some light on their application
Vision systems have been used in the pharmaceutical industry for many years, initially as inspection systems to identify defective components. Vision inspection systems offer specific key benefits over their human counterparts, namely speed, accuracy and repeatability, to meet the industry's demands to make products at high speed and with minimal (or ideally zero) defects. However, in recent years another facet of vision technology has come to the fore - part marking and code reading - which has allowed highly efficient traceability of product.
The importance of accurate, secure and durable identification of biological or medical samples cannot be underestimated. Also of great interest has been the increasing use of vision systems utilising wavelength regions such as X-ray and infrared instead of conventional visible light.
Most people are familiar with the concept of a vision inspection system comprising a camera or sensor and a processing unit (e.g. a computer equipped with a frame grabber and analysis software). However there are a number of other crucial components, including lenses and lighting which contribute significantly to the overall system.
A detailed understanding of how all the components interact can be the difference between a successful and unsuccessful implementation of a vision system. For example, the selection of suitable illumination is crucial for the quality of image capture and can vastly improve subsequent evaluation of the image. Reproducible illumination conditions must be in place in order to permit a constant image of identical objects or identical conditions for investigating different objects.
Fluctuations in the illumination conditions must therefore be avoided if strict quality criteria are to be applied to the inspection of objects. Only if it is possible to view the desired inspection features or faults with sufficient contrast is it possible to subsequently evaluate them using image processing software.
An example of the importance of illumination in an inspection application was highlighted recently for a leading manufacturer of pharmaceutical and medical devices, which include valves and actuators, dose counting and accessories, nasal devices and inflation check valves. One particular device contained an internal plastic gear, smaller than a thumbnail. The inspection procedure, traditionally carried out mechanically, required a check for both the presence of each gear tooth, as well as the gear itself. The objective of switching to a vision system was to speed up the process and reduce cost while further improving quality.
The major problem, however, was that the gear itself was white and was mounted against a white background. The solution was to use a smart camera (a smart camera has on-board image processing capabilities instead of using PC-based image processing) equipped with a polarising filter on the lens, and an additional polarising filter on the light source. By tuning the two filters to the right position, it was possible to generate an image in which the gear stood out from the background (Figure 1), which was also helped by the different types of plastic involved in the gear and the background.
Complete identification and traceability is now an industry prerequisite for the medical and pharmaceutical equipment sectors. Applications include checking the presence or absence of labels, character recognition (for example date code recognition or product type) and print verification. A huge array of products are tagged either by a stick-on label or by information printed directly onto the packaging, with information such as bar codes, lot details and best-before codes being the most common ones that need to be checked with total accuracy. But challenges can arise with the size of containers that may need to be labelled or marked and the speed at which they must be read.
These problems were highlighted recently for a company manufacturing plastic medical devices who needed to guarantee the individual traceability of biological samples stored in particular containers. Each container holds 48 separate wells destined to hold samples of blood or serum. Prior to this each container of samples was simply identified as a whole using a bar code, and one unsatisfactory sample led to the entire container being rejected.
Making it possible to identify such tiny individual wells was a demanding requirement for DPM (Direct Part Marking) and the need for complete traceability demanded an automated marking process capable of integrating the necessary data handling. This was tailored to meet the specific needs and technical requirements of the company. As a consequence, it was necessary to be able to mark two different codes that would be both machine-legible and visible to the human eye even after long periods of time.
In addition to being completely automated, the marking process had to comply with Class 10,000 cleanroom conditions, while maintaining zero risk of error and ensure that any non-conforming pieces are rejected during the process, and meeting the inspection requirements of one piece per minute. The chosen solution was laser marking of datamatrix codes due to the sheer amount of information that needed to be securely marked onto such a small surface.
The first step in the process involves the container holding 48 wells being manually loaded and identified using bar codes. Each individual unit is then marked using a laser. Each well is marked with the container code as well as its own unique identification code in human readable text. In addition a datamatrix machine-readable code is added containing the same information (Figure 2/main pic).
At this point each of the codes has to be inspected and verified to ensure that they can be decoded. A smart camera is mounted in such a way so as to be able to individually read each of the codes on each sample. The positioning of the vials is verified by the vision system. Codes are decoded during an automated process that compares the actual information with the expected information. If an error occurs the process is interrupted and the bad piece is rejected. The entire process takes just one minute per set of 48 wells.
There is growing interest in systems that make use of radiation outside the visible range, such as infrared and X-rays. The power of these alternative techniques is that they reveal information that could not otherwise be obtained. X-ray imaging can reveal detail inside an object, but it is only the recent development of a new generation of X-ray imaging systems which has allowed the technique to be applied to industrial in-line inspection applications.
X-ray imaging can now be applied to on-line blister pack inspection. The consequences of a blister compartment without its capsule, tablet or leaflet may be a health risk to the patient and is certainly a liability issue for the drug supplier.
A particular advantage of X-rays is their ability to see through aluminium, which is widely used in blister packs because it is highly inert to medicines and drugs. This gives X-rays an advantage over many technologies, such as optical, microwave or RF, where the foil appears completely opaque. X-ray cameras, specifically designed for inspection of smaller items in production lines, can be used to produce two-dimensional images of the blister pack (Figure 3).
Details can be resolved to an accuracy around that of the width of a human hair and the system is sensitive enough to detect the tablets (and low density pharmaceuticals) even when they are moving past at some speed on a conveyor belt. When used to inspect blister packs in their final packaging, the system is even sensitive enough to check for the presence of the product instruction leaflet in the box. A variety of advanced software algorithms are employed to identify rapidly and reliably any empty blisters giving a fully automatic pass/fail decision capability. Inspection rates of 400 blisters per second are readily achieved.
Thermography is the use of an infrared imaging and measurement camera to 'see' and 'measure' thermal energy emitted from an object. Any object with a temperature above absolute zero emits heat and the higher the temperature, the greater the infrared radiation emitted. Near-Infrared, or NIR radiation lies between visible and thermal IR radiation in terms of wavelength and is generally considered to cover the 0.75-2.5 mm wavelength range. NIR imaging primarily uses reflected-light imaging unlike thermal infrared imaging. It can see properties of materials invisible to the eye (molecular absorption and emission) and also has the benefit of being able to use relatively low cost optics as used for visible wavelengths. Near infrared (NIR) imaging can be used in the pharmaceutical industry for the evaluation of pharmaceutical formulations in tablets.
Any pharmaceutical formulation must be consistent and any particular tablet must have the same characteristics wherever it is produced. Identical ingredients can have radically different performance depending upon how the ingredients are blended together. It is therefore essential that the distribution of the drugs throughout a tablet is accurately known.
A particular strength of NIR imaging is that a series of images can be acquired at different NIR wavelengths to produce a three dimensional dataset which contains both spatial information and spectral information. Appropriate processing allows chemical composition information from the NIR spectra to be to be displayed spatially to show the chemical distribution of ingredients throughout the tablet.¹ Figure 4 (a) shows an image of a cross-section of a pharmaceutical time-release tablet taken in visible light. Only particle granularity can be seen. Figure 4 (b) shows the NIR chemical image of the same tablet, showing several distinct layers and boundaries. These are consistent with the known physical structure and composition of the particular formulation.