{"doi":"10.1002/mp.70104","title":"Auto‐segmentation of thoraco‐abdominal organs in pediatric dynamic MRI","abstract":"PURPOSE: Dynamic magnetic resonance imaging (dMRI) is a practical imaging modality for capturing information about regional thoracic-abdominal components and their dynamics in healthy children and pediatric patients with thoracic insufficiency syndrome (TIS). We propose an auto-segmentation set-up for the lungs, kidneys, liver, spleen, and thoraco-abdominal skin outer boundary (Skn) in dMRI images. METHODS: with 6.00 mm spacing between slices. We utilized images from 89 and 10 subjects at end inspiration for training and validation, respectively. For testing, we experimented with three scenarios utilizing: (1) the images of the 90 (=189-89-10) remaining subjects at end inspiration for testing, (2) the images of the remaining 90 subjects at end expiration for testing, and (3) the images of the other 99 (=89+10) subjects at end expiration for testing. In some situations, we can take advantage of the already available ground truth (GT) segmentation for an object in a subject at a particular respiratory phase to automatically segment the same object in the same subject at a different respiratory phase, and then refine the segmentation to create the final GT for all respiratory phases in the image of a subject. We anticipate that this process of creating GT would require minimal post hoc correction. In this spirit, we conducted separate experiments where we assumed to have GT of test subjects at the end expiration for scenario (1), end inspiration for (2), and end inspiration for (3). A major contribution in this paper is the different scenarios of training and testing that we have extensively evaluated with respect to respiratory phases and the subjects to which the images in the training and testing sets belong. RESULTS: Among these three scenarios of testing, for DL-R, we achieve the best average location error (LE) of about 1 voxel for the lungs, kidneys, and spleen, and 1.5 voxels for the liver and Skn. The standard deviation (SD) of LE is about 1 or 2 voxels. For DL-D, we achieve an average Dice coefficient (DC) of about 0.92 to 0.94 for the lungs, 0.82 for the kidneys, 0.90 for the liver, 0.81 for the spleen, and 0.93 for Skn. The SD of DC is lower (0.02 to 0.07) for the lungs, liver, and Skn and slightly higher (0.06 to 0.12) for the spleen and kidneys. CONCLUSIONS: Motivated by applications in surgical planning for disorders such as TIS, adolescent idiopathic scoliosis, and early onset scoliosis, we have created an auto-segmentation system for thoraco-abdominal organs in dMRI acquisitions. This proposed setup copes with the challenges posed by low resolution, motion blur, inadequate contrast, and image intensity non-standardness in dMRI images quite well.","journal":"Medical Physics","year":2025,"id":535384,"datarank":0.0,"base_score":0.0,"endowment":0.0,"self_citation_contribution":0.0,"citation_network_contribution":0.0,"self_endowment_contribution":0.0,"citer_contribution":0.0,"corpus_percentile":null,"corpus_rank":null,"citation_count":2,"citer_count":0,"citers_with_citation_signal":0,"citers_with_endowment":0,"datacite_reuse_total":0,"is_dataset":false,"is_dataset_confidence":0.7987,"is_data_producer":false,"deposit_databanks":null,"is_oa":true,"file_count":0,"downloads":0,"has_version_chain":false,"published_date":"2025-01-01","fair_score":null,"fair_percentile":null,"algorithm_id":"datarank_citation_only_1hop_v6","ranking_scope":"data_only","authors":[{"id":387447,"name":"Jayaram K. Udupa","orcid":"0000-0002-7361-2927","position":1,"is_corresponding":false},{"id":387448,"name":"Yubing Tong","orcid":"0000-0002-2133-4910","position":2,"is_corresponding":false},{"id":466188,"name":"Tiange Liu","orcid":"0000-0003-2486-8691","position":3,"is_corresponding":false},{"id":1163614,"name":"Caiyun Wu","orcid":"0000-0002-4351-7697","position":4,"is_corresponding":false},{"id":736730,"name":"Rachel M. Kogan","orcid":null,"position":5,"is_corresponding":false},{"id":1268794,"name":"Mostafa Alnoury","orcid":"0000-0002-2155-3451","position":6,"is_corresponding":false},{"id":1353343,"name":"Mahdie Hosseini","orcid":"0000-0001-6979-3780","position":7,"is_corresponding":false},{"id":1353796,"name":"Leihui Tong","orcid":null,"position":8,"is_corresponding":false},{"id":1419490,"name":"Samarth Mannikeri","orcid":null,"position":9,"is_corresponding":false},{"id":799275,"name":"Dewey Odhner","orcid":"0000-0001-6124-8796","position":10,"is_corresponding":false},{"id":384582,"name":"Joseph M. McDonough","orcid":null,"position":11,"is_corresponding":false},{"id":511202,"name":"Carina Lott","orcid":null,"position":12,"is_corresponding":false},{"id":943160,"name":"Abigail Clark","orcid":"0000-0002-2214-2160","position":13,"is_corresponding":false},{"id":510658,"name":"Patrick J. Cahill","orcid":"0000-0002-7129-1007","position":14,"is_corresponding":false},{"id":510657,"name":"Jason B. Anari","orcid":"0000-0002-2190-6333","position":15,"is_corresponding":false},{"id":426494,"name":"Drew A. Torigian","orcid":"0000-0001-8999-9735","position":16,"is_corresponding":false},{"id":1163613,"name":"Yusuf Akhtar","orcid":"0000-0002-1838-7325","position":0,"is_corresponding":true}],"reference_count":32,"raw_metadata":null,"created_at":"2026-07-19T02:51:56.297114Z","pmid":"41206361","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":[]}