RESUMO
This study presents the development and evaluation of an image processing software for computer-assisted cellular structure counting. The proposed software consists of a set of processing and analytical tools which allows its use in several applications of cell and cellular structure counting. A particular application in AgNOR quantitative analysis is presented. The knowledge obtained from experienced pathologists has been codified in a sequence of processing steps in order to allow automatic estimation of the mean number of AgNORs per cell in ameloblastomas. The performance of the presented software in such application was verified by comparing the data provided by visual analysis, by two observers previously calibrated and under supervision of two experienced specialists (Group 1) and by the computer program (Group 2). No statistical difference was observed (p<0.05) between the two groups. The use of the proposed method in AgNOR applications permitted attainment of accurate and precise data without the difficulties frequently found in the traditional visual analysis method (time, training and subjectivity). The developed software is an interesting tool as an aid in the study (estimation of the number of cells and cellular structures) of malignant and benign neoplasms.