Predictive selection rule of favourable image processing methods for X-ray micro-computed tomography images of tablets
Oct. 29, 2021, midnight
Remove from favorites
Add to favorites
Processed X-ray micro-computed tomography (micro-CT) images of actual pharmaceutical formulations offer the potential to validate in silico tools for predicting disintegration and dissolution. However, determining suitable image processing pathways can be a time-consuming task. This study aimed to establish the transferability of image processing methods and develop an approach for selecting potentially favorable image processing methods for datasets with similar characteristics to expedite the evaluation process. Data from a prior assessment of image processing methods and parameters were utilized to assess the robustness of image processing through statistical resampling and to formulate a predictive rule set. This rule set was validated with both a new ratio of Active Pharmaceutical Ingredient (API) and excipient, both within and outside the ratios used in rule development. The rule was then applied to images of a binary mixture containing new compounds with similar identified image properties to demonstrate the transferability of the rule set. Robust image processing pathways with narrow ranges of input parameters were successfully identified. Predictions from the image processing pathways yielded high desirabilities, which were visually confirmed for ratios within the calibrated range. The successful transfer of the rule set to the new binary mixture further affirmed its effectiveness.