Addressing variability in dried methotrexate droplet patterns with deep learning of spatially resolved features

This repository was developed to support research on the automated classification of methotrexate (MTX) dry-droplet patterns using convolutional neural networks (CNNs). The dataset presents a collection of images of dried MTX droplets at four concentration levels (100%, 80%, 60%, and 40%), where the lower concentrations correspond to samples adulterated with 20%, 40%, and 60% water. Each image includes regions of interest (ROIs) extracted from both the central and peripheral areas of each droplet at multiple rotational angles, capturing morphological variations to enhance model generalization.