The incidence of breast cancer varies greatly among countries, but statistics show that every year 720,000 new cases will
be diagnosed world-wide. However, a high percentage of these cases can be 100% healed if they are detected in early stages.
Because symptoms are not visible as far as advanced stages, it makes the treatments more aggressive and also less efficient.
Therefore, it is necessary to develop new strategies to detect the formation in early stages.
We have developed a tool based on a Case-Based Reasoning kernel for retrieving mammographic images by content analysis. One
of the main difficulties is the introduction of knowledge and abstract concepts from domain into the retrieval process. For
this reason, the article proposes integrate the human experts perceptions into it by means of an interaction between human
and system using a Relevance Feedback strategy. Furthermore, the strategy uses a Self-Organization Map to cluster the memory
and improve the time interaction.
Keywords Breast Cancer - Bioinformatics Tools - Relevance Feedback - Knowledge Discovery & Retrieval Data - Case-Based Reasoning - Self-Organization Map