A Novel Semantic-based CAD System for Mammography

A semantic-based CAD system for mammography is proposed. Semantic layer is added to the computer aided diagnosis procedure by comprehend the lesion symptom of mammogram and clinical information. This provides doctor a complete and standardization semantic interpretation of the case. Final result is obtained through the synthetic analysis of the acquired semantic information by a Bayesian semantic network which constructed under direction of the experts. Besides, the result is graded according to the BI-RADS rating system. Preliminary experiments demonstrate that the system is conducive to the interaction with doctor. It's capable to provide assistant diagnosis for a whole mammary disease case other than a single X-ray image.