Multi-Level Auto-Annotation and Semantic Retrieval for Medical Images

C.-Y. Lin, L.-H. Ma, and J.-Y. Chen (PRC)


Semantic modeling, medical image, similarity measurement, multi-level fuzzy bayesian networks


This study proposes a medical image semantic modeling approach and similarity measurement based on multi-level fuzzy Bayesian networks, to enable medical images retrieval at different semantic levels. In the method, given an uncertain nature of medical image features, Gaussian mixture models (GMM) are adopted to extract the middle level semantics of the pathological objects. Finally, as an intelligent tool dealing with problems of uncertainty, a Bayesian network is utilized to combine these middle level semantics to build a multi-level semantic model. Based on the probability model, a similarity measurement with a hierarchical strategy is proposed. To validate our method, we apply the approach to a small set of astrocytona MRI (Magnetic Resonance Imaging) samples. The experiment results show that this approach is very effective to enable the auto-annotation and semantic retrieval for medical images. The model outperforms the Bayesian network based crisp quantification model using k-nearest neighbor classifiers (K-NN). This study provides a novel way to retrieve medical images at various semantic levels assisting radiologists in diagnosis.

Important Links:

Go Back