{"id":"W4414871490","doi":"10.1109/access.2025.3617778","title":"AI-Based Multiclass Grading of Hepatic Steatosis From B-Mode Ultrasound: Generalization Across Modalities and Clinical Comparison With Radiologists","year":2025,"lang":"en","type":"article","venue":"IEEE Access","topic":"Liver Disease Diagnosis and Treatment","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Steatosis; Deep learning; Preprocessor; Pattern recognition (psychology); Test set; Fatty liver; Convolutional neural network; Artificial neural network; Grading (engineering)","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007744701,0.0008526101,0.0007540898,0.001682264,0.0003324518,0.001540227,0.0008266199,0.001212667,0.001372209],"category_scores_gemma":[0.02146938,0.0002755372,0.0007417455,0.0006657835,0.0006700314,0.001145222,0.001590519,0.001088967,0.001209831],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006184849,"about_ca_system_score_gemma":0.0005046473,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004216317,"about_ca_topic_score_gemma":0.004582831,"domain_scores_codex":[0.997626,0.001012051,0.000178545,0.0007165241,0.0003290939,0.0001378024],"domain_scores_gemma":[0.9922878,0.004451836,0.0005334869,0.001352158,0.001079965,0.000294746],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.002986856,0.0005266431,0.2571872,0.000376376,0.001197651,0.0004321419,0.0006944016,0.1128663,0.02128166,0.000933715,0.009249554,0.5922675],"study_design_scores_gemma":[0.00008791741,0.0004808237,0.1043725,0.00009173073,0.0002777617,0.0007719093,0.0003122157,0.874068,0.01341642,0.00347917,0.002533357,0.0001082205],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9120076,0.00195383,0.07621709,0.000866583,0.0002454765,0.0002577102,0.001512759,0.001933379,0.005005464],"genre_scores_gemma":[0.9807623,0.0002844047,0.01516948,0.0002065289,0.00007728292,0.00006265478,0.002059186,0.0001039797,0.001274249],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.007744701,"threshold_uncertainty_score":0.0409584,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04638956981719314,"score_gpt":0.4206552711845321,"score_spread":0.3742657013673389,"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."}}