Search for a command to run...
AIM: To prepare a dataset head and neck CT scans with lymph nodes for further use in the development of artificial intelligence tools. METHODS: We selected patients who underwent CT scans with intravenous contrast enhancement (CE) from 2020 to 2023 were selected for this retrospective study. The selected patients were over 18 years old. Inclusion criteria: presence of a histologically confirmed malignancy, presence of CT with CE. Exclusion criteria: artefacts from dental implants at the level of target lymph nodes (categories 1 and 5 according to Node-RADS), presence of motion artifacts. Two samples were formed: studies with signs of metastatic lymph nodes and without. The studies were performed on Toshiba Aquilion units. The dataset includes images obtained in the venous phase of contrast enhancement: start at 70 seconds when peak density values in the the aorta was 130 HU. All studies were anonymized. The dataset was labeled by three radiologists with more than 3 years of experience. A labeling table was created containing the following data: patient number, sex, age, presence or absence of secondary changes in lymph nodes, bilateral or unilateral lesion, side of lesion, size of the largest lymph node along the short axis, size of the largest lymph node along the long axis, lymph node group, image series and number. RESULTS: 84 CT studies with intravenous contrast enhancement were selected in accordance with the inclusion and exclusion criteria. Among the selected patients, 61 had signs of metastatic lymph node lesions, while 23 had no signs of secondary lesions. 75 lymph nodes of category Node-RADS 5 were labeled. The dataset contains 84 DICOM files with a total volume of 19.0 GB. CONCLUSION: A dataset of 84 patients has been prepared and made available in the public domain, containing labels of lymph nodes in the neck according to the Node-RADS classification (categories 1 and 5) in CT scans with intravenous contrast enhancement, as well as clinical data. The collected data can be used to create and test artificial intelligence for detecting and evaluating lymph nodes.