{"database":"biostudies-literature","file_versions":[],"scores":null,"additional":{"omics_type":["Unknown"],"volume":["11(6)"],"submitter":["Tariq A"],"pubmed_abstract":["<h4>Purpose</h4>Self-supervised pre-training can reduce the amount of labeled training data needed by pre-learning fundamental visual characteristics of the medical imaging data. We investigate several self-supervised training strategies for chest computed tomography exams and their effects on downstream applications.<h4>Approach</h4>We benchmark five well-known self-supervision strategies (masked image region prediction, next slice prediction, rotation prediction, flip prediction, and denoising) on 15 M chest computed tomography (CT) slices collected from four sites of the Mayo Clinic enterprise, United States. These models were evaluated for two downstream tasks on public datasets: pulmonary embolism (PE) detection (classification) and lung nodule segmentation. Image embeddings generated"],"journal":["Journal of medical imaging (Bellingham, Wash.)"],"pagination":["064003"],"full_dataset_link":["https://www.ebi.ac.uk/biostudies/studies/S-EPMC11550486"],"repository":["biostudies-literature"],"pubmed_title":["Self-supervised learning for chest computed tomography: training strategies and effect on downstream applications."],"pmcid":["PMC11550486"],"pubmed_authors":["Patel B","Banerjee I","Tariq A","Ramasamy G"],"additional_accession":[]},"is_claimable":false,"name":"Self-supervised learning for chest computed tomography: training strategies and effect on downstream applications.","description":"<h4>Purpose</h4>Self-supervised pre-training can reduce the amount of labeled training data needed by pre-learning fundamental visual characteristics of the medical imaging data. We investigate several self-supervised training strategies for chest computed tomography exams and their effects on downstream applications.<h4>Approach</h4>We benchmark five well-known self-supervision strategies (masked image region prediction, next slice prediction, rotation prediction, flip prediction, and denoising) on 15 M chest computed tomography (CT) slices collected from four sites of the Mayo Clinic enterprise, United States. These models were evaluated for two downstream tasks on public datasets: pulmonary embolism (PE) detection (classification) and lung nodule segmentation. Image embeddings generated","dates":{"release":"2024-01-01T00:00:00Z","publication":"2024 Nov","modification":"2026-06-12T04:58:32.792Z","creation":"2026-06-12T03:08:07.525Z"},"accession":"S-EPMC11550486","cross_references":{"pubmed":["39529658"],"doi":["10.1117/1.JMI.11.6.064003","10.1117/1.jmi.11.6.064003"]}}