AMDI is a system that integrates modules to permit the addition of new cases into the mammographic database by authorized radiologists, and to assist research and education activities in breast cancer through a flexible and easy-to-use interface via the Web.
The mammographic database was developed using the PostgreSQL with Image-Handling Extension (PostgreSQL-IE), which is a eXtension Relational Database Management System (XRDBMS) developed by our research group and available for downloading at www.lcc.ufu.br/dpi/download. The mammographic database was projected to include cases with all of the available mammographic views, radiological findings, diagnosis proven by biopsy, the patient's clinical history, and information regarding the life style of the patient. Each exam of each case includes four views (two views of each breast: cranio-caudal or CC, and medio-lateral oblique or MLO). To address the teaching and research aspects, the database links each mammogram with the contour of the breast, the boundary of the pectoral muscle (MLO views only), the contours of masses (if present), the regions of clusters of calcifications and the number of calcifications (if present), and the locations and details of any other features of interest. The contours of masses and regions of clusters of calcifications may be drawn interactively by an authorized expert radiologist, when he or she is including a new case in the database. The mammographic database also supports the inclusion of several mammographic exams of the same patient performed at different instants of time. This information can be used for temporal analysis of the breast, due to natural modifications that occur during the life of the woman, or to analyze interval cancer.
The research system called SISPRIM --- Sistema de Pesquisa para Recuperação de Imagens Mamográficas, that is, Research System for Retrieval of Mammographic Images --- integrated with AMDI, allows physicians and oncologists to study possible statistical correlation between the incidence of breast cancer and the life style of the patient. For this purpose, some of the important items of data supported by AMDI relate to information regarding the patient, including basic data about the clinical history of the patient as well as life style, such as food habits, exercise, diet, and the use of antidepressive medication, tobacco, or alcohol. The SISPRIM research system also allows content-based image retrieval to assist the radiologist in breast cancer diagnosis. The SISPRIM can answer queries as ''return the five images that are more similar to the given reference image and the patient takes antidepressive medication'', or " return the identification of the patient, and the diagnosis associated with each one of the ten images more similar to the reference image, and that the density of parenchyma tissue is P2, and that the age at menarche is 10 years old''.
The e-learning system called INDIAM --- INterpretation and Diagnosis of Mammograms ---- was designed to assist a medical student or resident in the interpretation of mammograms and diagnosis of breast cancer, and is being developed using Web services. INDIAM makes available a tutorial that uses education techniques to guide the users (doctors, students, or researchers) through concepts related to the diagnosis of breast cancer. It also makes available a module to simulate the analysis and diagnosis of breast cancer using cases retrieved from a mammographic database, and another module for training the student in the interpretation of mammograms. The e-learning system is being developed utilizing an ontology for the interpretation of mammograms (OntoBreastCancer), that provides controlled and consistent vocabularies to describe concepts and relationships, thereby enabling knowledge sharing. The system makes available a user-friendly graphical Web interface that is configured according to the service being provided.
AMDI provides a tool that enables the user to download cases from the mammographic database, so as to make the information available to authorized medical and research communities interested in breast cancer diagnosis.