{"id":"W4308580198","doi":"10.1016/j.neuroimage.2022.119741","title":"A benchmark for hypothalamus segmentation on T1-weighted MR images","year":2022,"lang":"en","type":"article","venue":"NeuroImage","topic":"Sleep and Wakefulness Research","field":"Neuroscience","cited_by":11,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Conselho Nacional de Desenvolvimento Científico e Tecnológico; Coordenação de Aperfeiçoamento de Pessoal de Nível Superior; Fundação de Amparo à Pesquisa do Estado de São Paulo","keywords":"Benchmark (surveying); Segmentation; Artificial intelligence; Computer science; Pattern recognition (psychology); Psychology; Cartography; Geography","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003160961,0.003420387,0.001413904,0.003114179,0.0008373316,0.002113799,0.002871042,0.003155936,0.003316714],"category_scores_gemma":[0.006451995,0.0004800782,0.001682832,0.002255335,0.0008207862,0.0009161396,0.001620997,0.001189896,0.00268721],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001658884,"about_ca_system_score_gemma":0.001451902,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01554546,"about_ca_topic_score_gemma":0.01799574,"domain_scores_codex":[0.9981546,0.0003864974,0.0001728145,0.0006697343,0.0003950854,0.0002213104],"domain_scores_gemma":[0.9978193,0.0008964286,0.0002176544,0.0003255032,0.0005509984,0.000190132],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.005336214,0.001473262,0.01938495,0.005040733,0.001544914,0.002570755,0.000401825,0.1848874,0.03064633,0.002187751,0.1644162,0.5821097],"study_design_scores_gemma":[0.0008527566,0.002485001,0.04598566,0.0009922332,0.0006061063,0.004988953,0.0008495948,0.8090661,0.05129852,0.007389455,0.07527366,0.0002119661],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"methods","genre_scores_codex":[0.6113598,0.04412066,0.1534482,0.003853825,0.00259069,0.00243189,0.1083389,0.05040328,0.02345281],"genre_scores_gemma":[0.4882481,0.005456556,0.2225468,0.001164835,0.0005469691,0.0008987901,0.2676611,0.002368669,0.01110821],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.01554546,"threshold_uncertainty_score":0.03090996,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05849388644744533,"score_gpt":0.3180425434349149,"score_spread":0.2595486569874696,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}