Evaluation of retrieval accuracy and visual similarity in content-based image retrieval of chest CT for obstructive lung disease

The study by Choe et al. evaluated a content-based image retrieval (CBIR) system designed to identify visually similar chest CT scans for patients with obstructive lung disease, using quantitative data. The system was built on a database of 600 volumetric chest CT scans from 541 patients, with follow-up scans from 50 patients used as query cases. Quantitative CT features, such as emphysema extent and airway wall thickness, were analyzed using Aview, a fully automated segmentation software developed by Coreline Soft. The CBIR system achieved a retrieval accuracy of 68% when considering the top five similar scans. Radiologists rated 64.8% of retrieved images with high visual similarity scores. This technology shows promise for aiding clinical decision-making in chronic obstructive pulmonary disease (COPD) by facilitating the identification of patients with similar imaging phenotypes. Future work should link these findings with clinical outcomes and refine the system with larger datasets and more advanced features.