Content Based Image Retrieval by Metric Learning from Radiology Reports Application to Interstitial Lung Diseases - 2015
Content Based mostly Image Retrieval (CBIR) is a search technology that might aid medical diagnosis by retrieving and presenting earlier reported cases that are related to the one being diagnosed. To retrieve relevant cases, CBIR systems rely on supervised learning to map low level image contents to high level diagnostic ideas. However, the annotation by medical doctors for training and analysis functions could be a tough and timeconsuming task, which restricts the supervised learning phase to specific CBIR problems of well outlined clinical applications. This paper proposes a replacement technique that automatically learns the similarity between the several exams from textual distances extracted from radiology reports, thereby successfully reducing the quantity of annotations required. Our method firstly infers the relation between patients by using information retrieval techniques to work out the textual distances between patient radiology reports. These distances are subsequently used to supervise a metric learning algorithm, that transforms the image space accordingly to textual distances. CBIR systems with different image descriptions and totally different levels of medical annotations were evaluated, with and while not supervision from textual distances, employing a database of pc tomography scans of patients with interstitial lung diseases. The proposed methodology consistently improves CBIR mean average precision, with improvements that can reach 38percent, and more marked gains for little annotation sets. Given the general availability of radiology reports in Image Archiving and Communication Systems, the proposed approach will be broadly applied to CBIR systems in numerous medical problems, and might facilitate the introduction of CBIR in clinical observe.
Did you like this research project?
To get this research project Guidelines, Training and Code... Click Here