Content-based image retrieval system to support the diagnosis of human papillomavirus

This article describes the development of a Content- Based Image Retrieval system of digital images of cervix biopsy infected with Human Papillomavirus. Through a query image, the system performs a search in the database of digital images of cervix biopsies. The search makes possible to recover the diagnoses of the images stored in the database that might represent lesions similar to the query image. By using feature extraction techniques of low-level, a representation of the visual content of the images is calculated. This representation is used to compare the images by means of similarity measures. The system has been built as a helping tool for medical diagnoses; the purpose is to reuse the knowledge gained by carrying out diagnoses and treatments in order to diagnose new images. The evaluation of the system produced an average precision of 80 % for the retrieval of similar images stored in the database.