Rapid, Non-destructive Inspection and Classification of Inhalation Blisters Using Low-Energy X-ray Imaging

Abstract

Dry powders packaged in aluminum foil blisters are an increasingly common dosage form found in inhalation drug products. Filling of inhalation blisters often involves compressing bulk aerated powder into dense compacts. Sealed blisters are conditioned, e.g., by ultrasonic vibration, to loosen the consolidated powder within, to enable ready dispersion to a respirable aerosol when actuated by an inhaler device. Currently, the presence of residual powder consolidation within the blister is monitored manually by cutting open blisters for visual inspection. X-ray imaging has gained increased acceptance as a non-destructive analytical technique for pharmaceutical capsules and tablets, typically with masses on the order of ~ 100 mg. Here, an X-ray inspection approach was investigated for inhalation blisters having a significantly smaller powder fill mass of 2 mg. The challenge of sensing a small powder mass (2 mg) packaged within a significantly heavier blister (~ 75 mg) was met using a low-energy X-ray imaging system. The measurement principle relies on denser, consolidated powder appearing as darker regions in the recorded image. Proof-of-concept experiments were performed using empty blister strips, and blister strips filled with 2 mg of placebo powder, half of which were subjected to ultrasonic conditioning. The tests demonstrated that a supervised machine learning approach based on digitally processed X-ray images reliably distinguished between the three types of blisters tested, i.e., empty blisters and conditioned and un-conditioned blisters of 2-mg fill mass. Using independent training and validation sets of 948 images each, an automated classification accuracy ≥ 99.8% was demonstrated.