{"doi":"10.1101/2022.01.18.22269482","title":"3D Capsule Networks for Brain Image Segmentation","abstract":"Abstract Background and Purpose Current auto-segmentation models of brain structures, UNets and nnUNets, have limitations, including the inability to segment images that are not represented during training and lack of computational efficiency. 3D capsule networks (CapsNets) have the potential to address these limitations. Methods We used 3430 brain MRIs, acquired in a multi-institutional study, to train and validate our models. We compared our CapsNet with standard alternatives, UNets and nnUNets, based on segmentation efficacy (Dice scores), segmentation performance when the image is not well-represented in the training data, performance when the training data are limited, and computational efficiency including required memory and computational speed. Results The CapsNet segmented the third ventricle, thalamus, and hippocampus with Dice scores of 95%, 94%, and 92%, respectively, which were within 1% of the Dice scores of UNets and nnUNets. The CapsNet significantly outperformed UNets in segmenting images that are not well-represented in the training data, with Dice scores 30% higher. The computational memory required for the CapsNet is less than a tenth of the memory required for UNets or nnUNets. The CapsNet is also more than 25% faster to train compared with UNet and nnUNet. Conclusion We developed and validated a CapsNet that is effective in segmenting brain images, can segment images that are not well-represented in the training data, and are computationally efficient compared with alternatives.","journal":"medRxiv","year":2022,"id":299797,"datarank":0.20794415416798362,"base_score":1.3862943611198906,"endowment":1.3862943611198906,"self_citation_contribution":0.20794415416798362,"citation_network_contribution":0.0,"self_endowment_contribution":0.20794415416798362,"citer_contribution":0.0,"corpus_percentile":null,"corpus_rank":null,"citation_count":3,"citer_count":0,"citers_with_citation_signal":0,"citers_with_endowment":0,"datacite_reuse_total":0,"is_dataset":false,"is_dataset_confidence":0.9343,"is_data_producer":false,"deposit_databanks":null,"is_oa":true,"file_count":0,"downloads":0,"has_version_chain":false,"published_date":"2022-01-01","fair_score":null,"fair_percentile":null,"algorithm_id":"datarank_citation_only_1hop_v6","ranking_scope":"data_only","authors":[{"id":990092,"name":"Yongfeng Hui","orcid":"0000-0003-2668-5151","position":1,"is_corresponding":false},{"id":289882,"name":"Mariam Aboian","orcid":"0000-0002-4877-8271","position":2,"is_corresponding":false},{"id":407669,"name":"James S. Duncan","orcid":"0000-0002-1276-2745","position":3,"is_corresponding":false},{"id":1077,"name":"Harlan M. Krumholz","orcid":"0000-0003-2046-127X","position":4,"is_corresponding":false},{"id":520421,"name":"Sanjay Aneja","orcid":"0000-0001-5681-7528","position":5,"is_corresponding":false},{"id":985694,"name":"Arman Avesta","orcid":"0000-0002-6847-9818","position":0,"is_corresponding":true}],"reference_count":36,"raw_metadata":{"citation_network_status":"fetched"},"created_at":"2026-07-19T00:31:49.412501Z","pmid":null,"pmcid":null,"fwci":null,"citation_percentile":null,"influential_citations":0,"oa_status":null,"license":null,"views":0,"total_file_size_bytes":0,"version_count":0,"fair_f":null,"fair_a":null,"fair_i":null,"fair_r":null,"fair_zscore":null,"fair_rationale":null,"fair_model":null,"fair_agent_version":null,"fair_fulltext_source":null,"fair_has_llm":null,"fair_computed_at":null,"clinical_trials":[],"software_tools":[],"db_accessions":[],"linked_datasets":[],"topics":[]}