RESUMEN
Background: Description of the external shape of the particles for pharmaceutical use has usually been subjective, cumbersome or slow, even when the techniques have used digital image processing. This study sought to evaluate the crystal habit of raw material of phenytoin by two digital techniques, one of them based on the edges definition of particles images, using conventional geometric descriptors, and the other one describing the surface texture. These estimates were optimized using multivariate statistical analysis. Objective: The purpose of this study was to evaluate an appropriate way to characterize the crystalline habit of raw materials used in the pharmaceutical industry, from information obtained when it used some digital measuring techniques. Methods: Classical and mathematical descriptors were used, in addition to a modified method of texture analysis technique based on the gray levels co-occurrence matrix (GLCM). Phenytoin was used as the raw material model, since this showed different crystal habits when some conditions were changed in the recrystallization process. The evaluation of measurements was made previously calibrating the meaning of the descriptors with different geometric shapes, and then solving it with the help of multivariate analysis techniques such as principal component analysis (PCA) and hierarchical cluster analysis (ACJ). Results: Calibration figures were grouped into two descriptors based on the degree of homogenization depicted in images of the particles. The crystal habits of phenytoin were defined by compaction and by elongation and irregularity, which was also found with the technique of Matrix Co-occurrence of gray levels of texture measurement. Conclusion: It decreased the subjectivity of the definition of crystalline habits of phenytoin when using a combination of less than four classical and mathematical descriptors, obtained through multivariate analysis.