CT-SCOPE: annotated dataset of CT SCans for the automatic semantic segmentation of the Osseous structures of the Paranasal sinusEs

This article introduces a dataset of computed tomography (CT) scans of the paranasal sinuses collected from 6 distinct hospitals, using 4 different CT devices, and ensuring diverse recording conditions. The dataset includes CT axial scans from 40 subjects, 13 of which were manually annotated with a semantic segmentation of the osseous structures of the area surrounding the paranasal sinuses, while the remaining 27 subjects contain unannotated CT scans. The data was organized into raw DICOM files and was also stored as uncompressed PNG images. The dataset includes an average of 212±105 slices per subject, while the annotated subset contains 696 masks paired with their corresponding CT slice. To further enhance the dataset, a set of automatically delineated masks (i.e., pseudo-labels) is also included for the unannotated CT scans. This dataset is highly valuable for medical image analysis, particularly to train and evaluate deep learning sematic segmentation models to identify the osseous structures surrounding the paranasal sinuses, as well as to explore domain adaptation techniques across different imaging devices. Additionally, it supports research in areas such as resolution enhancement and cross-device generalization, positioning it as an essential resource for advancing the robustness and generalizability of artificial intelligence driven medical image analysis tools.