{"id":"W2991523350","doi":"10.1109/tmi.2020.3006437","title":"Confidence Calibration and Predictive Uncertainty Estimation for Deep Medical Image Segmentation","year":2020,"lang":"en","type":"article","venue":"IEEE Transactions on Medical Imaging","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":341,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of British Columbia","funders":"National Institute of Biomedical Imaging and Bioengineering; Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; National Institutes of Health","keywords":"Dice; Segmentation; Normalization (sociology); Image segmentation; Calibration; Convolutional neural network; Pattern recognition (psychology); Medical imaging","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005256086,0.0001635431,0.0001799781,0.00008189462,0.0003121311,0.0001441371,0.0003892737,0.00009963079,0.000133681],"category_scores_gemma":[0.0006353469,0.000158871,0.00005708048,0.0002678584,0.0001921815,0.001087331,0.000009622826,0.0004800004,0.000006921889],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000654065,"about_ca_system_score_gemma":0.0001859701,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002949383,"about_ca_topic_score_gemma":0.000007974839,"domain_scores_codex":[0.9978356,0.0001598467,0.0003347735,0.0004838254,0.00094711,0.0002388428],"domain_scores_gemma":[0.9985315,0.0006884234,0.00009949214,0.0001566048,0.00009104401,0.0004329053],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001393985,0.00009956065,0.00004031784,0.0001266268,0.0000523985,0.00005407016,0.004424831,0.3593577,0.001000397,0.003442092,0.0003036627,0.630959],"study_design_scores_gemma":[0.0009357228,0.00008166437,0.00001958131,0.00006931243,0.00002499325,0.00002735509,0.0002105829,0.9953997,0.001872099,0.001179466,0.00003215694,0.0001474066],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0003178106,0.00002498326,0.9743273,0.02425236,0.0004389846,0.0003470863,0.000007038051,0.0002462982,0.00003816092],"genre_scores_gemma":[0.884614,0.00002561598,0.1114827,0.003618028,0.0001367228,0.00008575025,0.0000137576,0.00001703899,0.000006471494],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8842962,"threshold_uncertainty_score":0.647857,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01186382904086464,"score_gpt":0.2884710260404177,"score_spread":0.2766071969995531,"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."}}