INbreast: toward a full-field digital mammographic database.

摘要: Rationale and Objectives Computer-aided detection diagnosis (CAD) systems have been developed in the past two decades to assist radiologists of lesions seen on breast imaging exams, thus providing a second opinion. Mammographic databases play an important role development algorithms aiming at mammary lesions. However, available often do not take into consideration all requirements needed for research study purposes. This article aims present detail new mammographic database. Materials Methods Images were acquired center located university hospital (Centro Hospitalar de S. Joao [CHSJ], Breast Centre, Porto) with permission Portuguese National Committee Data Protection Hospital's Ethics Committee. MammoNovation Siemens full-field digital mammography, solid-state detector amorphous selenium was used. Results The database—INbreast—has total 115 cases (410 images) from which 90 are women both breasts affected (four images per case) 25 mastectomy patients (two case). Several types (masses, calcifications, asymmetries, distortions) included. Accurate contours made by specialists also provided XML format. Conclusion strengths actually presented database—INbreast—relies fact that it built mammograms (in opposition digitized mammograms), presents wide variability cases, is publicly together precise annotations. We believe this database can be reference future works centered or related cancer imaging.