UV imaging as a novel, ultrafast chemical mapping tool for pharmaceutical tablet analysis

Abstract

This paper presents a proof of concept for a novel pharmaceutical analytical tool, which utilizes a UV imaging-based machine vision system in order to rapidly acquire maps showing the components of the inspected uniformly white tablets. Two types of tablets were produced, one containing acetylsalicylic acid and one with caffeine, both consisting of two more excipients. Maps were produced by k-means clustering from the pixel information of UV images taken of the tablets, which is a new approach in this field. The results were then compared to maps acquired using Raman chemical imaging. It was found that the particles recognized on the UV maps are similar to the ones on Raman maps. Furthermore, the developed UV imaging-based machine vision system can investigate entire tablet surfaces in under 4 milliseconds, whereas Raman chemical imaging may take over 19 h to measure just a small fraction of the tablet surface. Therefore, the obtained results confirm that UV imaging could be used to rapidly gather highly informative component distribution maps of tablets, which could provide a novel approach of pharmaceutical tablet analysis in a solid state.

Introduction

Vibrational spectroscopic chemical imaging is a widely used tool in the field of pharmaceutical analysis [7]. The combination of vibrational spectroscopy and microscopy has introduced numerous innovations in solid-state analysis and local characterization [4]. Raman and near-infrared (NIR) chemical images can contain a wide range of information taken from the surface of a solid dosage form, e.g. a tablet [13]. With Raman chemical imaging, various critical quality attributes of tablets, such as active pharmaceutical ingredient (API) content [8,22], particle size distribution of components [1] and homogeneity assessment [9] can be determined. However, in order to extract the desired information from chemical images, the gathered hyperspectral datacubes first have to be converted into concentration maps for easier handling, which can be a challenging task [25]. The produced maps can then be further processed with classical image analysis techniques [23,24] or, thanks to their recent advances, convolutional neural networks [12]. Another limiting factor of Raman mapping, in addition to their high investment cost, is the extended data collection time. Even with newer fast Raman techniques, the time it takes to inspect a 3000 μm × 3000 μm area is around 3.5 h with a 0.1 s data acquisition time and a step size of 10 μm. This area could be also investigated utilizing NIR chemical mapping as well, achieving a total measurement time of 13 min with a step size of 25 μm and an acquisition time of 0.05 s [2]. Using laser direct infrared (LDIR) imaging, which is a new approach for chemical image acquisition, it still takes around 17 s to gather the spectral data required for the construction of a same sized chemical map with a step size of 10 μm [3].

UV imaging is a novel, non-destructive analytical method for tablet inspection. Multispectral UV imaging was used first to visualize API in the tablets and to predict the physical properties of tablets [14,21,27]. However, the measurement time was 18 s per multispectral image, because of the consecutive manner of data acquisition at different wavelengths, limiting the number of inspected samples.

By exposing the inspected tablets to only a single wavelength of UV light, the APIs can display different colours thanks to a slight difference in their fluorescence and absorption. This nature APIs may provide a suitable differentiation between components, and by taking images of the illuminated surface with an RGB camera, several critical tablet properties, such as API content, particle size distribution and dissolution profile could be determined with a measurement time as low as a few milliseconds [19]. Mészáros et al. utilized single wavelength UV/VIS imaging to determine the drug content, crushing strength and dissolution profile of tablets containing a yellow API, meloxicam [20]. They have used an acquisition time of 1.2 ms for the determination of API content which is four magnitudes faster than multispectral UV imaging. They were able to predict the meloxicam content of the tablets with a relative error of 4.9 % with algorithms utilizing RGB and CIELAB color space. Ficzere et al. [10] were the first to apply single-wavelength UV imaging to predict the content of colourless APIs in their tablets. They have utilized artificial neural networks for the prediction, and after optimization, have achieved a relative error of 4.41 % and 3.98 % in the case of amlodipine and valsartan containing tablets respectively. However, no cases have been reported where single wavelength UV imaging was used as a rapid chemical mapping tool for the analysis of the distribution of all the tablet components, including excipients on the surface.

Recently, UV imaging has also gained attention in materials science and optoelectronics, where fast and sensitive photodetectors are being developed using advanced semiconducting materials. For instance, quasi-2D Ruddlesden–Popper perovskite films have demonstrated significantly enhanced UV photoresponse through tailored defect passivation strategies, leading to high-resolution, self-powered photodetectors capable of rapid imaging [28]. In another study, vacuum-evaporated lead-free metal halide films enabled wafer-scale UV photodetector arrays with high uniformity and stability [6], showing promise for monolithic UV imaging systems. In addition, recent advances in the large-area fabrication of 2D perovskite oxide nanosheets have enabled ultra-flexible, high-resolution photodetector arrays with exceptional performance in spatial imaging and motion trajectory recognition, further expanding the scope of UV imaging applications [5]. These works demonstrate the potential of UV imaging beyond conventional spectroscopy, especially in applications requiring high-speed and high-resolution spatial information [15]. Despite their differing scopes, these applications collectively highlight the growing importance of UV imaging as a versatile and sensitive tool for spatially resolved analysis.

The aim of this study is to showcase the ability of UV imaging as a part of a high throughput evaluation system for pharmaceutical tablet analysis. The authors intend to use API and excipient combinations to present the method’s wide range of applicability across different material characteristics. We want to compare the resulted UV imaging- and the Raman chemical imaging-based maps of the prepared tablets in order to evaluate the methods performance focusing on both spatial and spectral identification of the API and the excipients. UV imaging can be used to get fast insight into tablet composition, making it a valuable tool for the pharmaceutical industry.

Download the full article as PDF here UV imaging as a novel, ultrafast chemical mapping tool for pharmaceutical tablet analysis

or read it here

Materials

Acetylsalicylic acid (ASA) was purchased from Molar Chemicals (Halásztelek, Hungary) and was used as a non-fluorescent API. Microcrystalline cellulose (MCC) and calcium hydrogen phosphate (CaHPO4) were obtained from JRS Pharma (Rosenberg, Germany), while maize starch (Starch 1500) was acquired from Colorcon Inc. (Dartford, UK). Caffeine was purchased from BASF (Ludwigshafen, Germany) and applied as a fluorescent API. Dextrose was obtained from Hungrana (Szabadegyháza, Hungary) and magnesium stearate was purchased from Sigma-Aldrich (Budapest, Hungary).

2.2.1. Preparation of tablets

A Dott Bonapace CPR-6 (Limbiate, Italy) single punch tablet press was used to manufacture biconvex tablets with the diameter of 14 mm, using concave punches. The compression force was set to 15 kN. Two sets of tablets with different compositions and total weights were produced, as detailed in Table 1. Both formulations were designed to model commercially available products. For each formulation, powders were manually pre-blended for 5 min in quantities sufficient for the production of 10 tablets. One tablet from each formulation, denoted as Tablet A and Tablet B, was randomly selected for UV imaging-based mapping.

Máté Ficzere, Lajos Madarász, Szabina Kádár, Attila Farkas, Lilla Alexandra Mészáros, Zsombor Kristóf Nagy, UV imaging as a novel, ultrafast chemical mapping tool for pharmaceutical tablet analysis, Microchemical Journal, Volume 217, 2025, 115032, ISSN 0026-265X, https://doi.org/10.1016/j.microc.2025.115032.