Semantic and content-based medical image retrieval for lung cancer diagnosis with the inclusion of expert knowledge and proven pathology

This paper involves the analysis and experimentation of chest CT scan data for the detection and diagnosis of lung cancer. In lung cancer computer-aided diagnosis (CAD) systems, having an accurate ground truth is critical and time consuming. The contribution of this work include the development of lung nodule database with proven pathology using content based image retrieval (CBIR) and algorithms for detection and classification of nodules. A study and analysis of 246 patients have been carried out for the detection of benign, malignant as well as metastasis nodules. The whole research work has been carried out using Lung Image Database Consortium (LIDC) database by National Cancer Institute (NCI), USA and achieved an average precision of 92.8% and mean average precision of 82% at recall 0.1. Finally, the validations have been carried out with the PGIMER, Chandigarh test cases and achieved an average precision of 88%. Experimental studies show that the proposed parameters and analysis improves the semantic performance while reducing the computational complexity, reading and analysing all slices by physicians and retrieval time.