T. Podsiadły-Marczykowska, A. Przelaskowski, A. Wróblewska, and P. Boninski (Poland)
Ontology, graphical user interface, computer-aided diagnosis, reference image database, integrated diagnostic workstation.
The paper presents a framework for improvement of diagnostic effectiveness in diagnostic mammography. Main human-oriented causes lowering the diagnostic value of mammography have been identified as: variability in mammogram description and assessment, radiologist’s perception and interpretation errors. Means and tools addressing those problems are: ontology to standardize description and reduce the variability of interpretation because of increased objectivity, CAD and reference database as aiding tools in perception and diagnosis improvement. Ontology has been used as a partial set of design assumptions for development of graphical editor dedicated to mammograms description. The use of domain ontology to design and control data entry improves its semantics and completeness, reducing description variability and hierarchical knowledge dependencies. Iconic data presentation model minimizes the burden of data capture and improves their visualization. Editor has been implemented in Java language. CAD tool and reference image database supported by formal knowledge platform are integrated into editor, and accessible during mammogram description, serving as aiding indicators, descriptors and automatic classifiers. The proposed system is still a work in- progress. Future works include: the objectification of lesions features and closer semantic integration of all applications by means of mammographic image ontology, as well as the technological optimization of the integrated diagnostic workstation.
Important Links:
Go Back
