{"id":"W4398163930","doi":"10.1371/journal.pone.0303278","title":"Explainability of three-dimensional convolutional neural networks for functional magnetic resonance imaging of Alzheimer’s disease classification based on gradient-weighted class activation mapping","year":2024,"lang":"en","type":"article","venue":"PLoS ONE","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":17,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Canadian Institutes of Health Research; National Institutes of Health; H. Lundbeck A/S; University of California, Los Angeles; Servier; China Scholarship Council; Eisai; Northern California Institute for Research and Education; University of Southern California; Biogen; Eli Lilly and Company; Bristol-Myers Squibb; BioClinica; U.S. Department of Defense; Meso Scale Diagnostics; Alzheimer's Disease Neuroimaging Initiative; Japan Society for the Promotion of Science; Wake Forest University; Novartis Pharmaceuticals Corporation; Pfizer; National Institute on Aging; Alzheimer's Association","keywords":"Functional magnetic resonance imaging; Convolutional neural network; Magnetic resonance imaging; Nuclear magnetic resonance; Pattern recognition (psychology); Artificial intelligence; Class (philosophy); Computer science; Neuroscience; Physics; Medicine; Biology; Radiology","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"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":[0.001502409,0.001088378,0.0005401853,0.0010221,0.0003615928,0.0007524026,0.0009047812,0.0009591029,0.001180909],"category_scores_gemma":[0.003410874,0.0003818683,0.001443927,0.0003630159,0.0004812498,0.0006827608,0.0006771203,0.001145019,0.0002186476],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001350226,"about_ca_system_score_gemma":0.0007480458,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01978761,"about_ca_topic_score_gemma":0.01324694,"domain_scores_codex":[0.9997054,0.0001054433,0.00001569951,0.00008389933,0.0000424292,0.00004714407],"domain_scores_gemma":[0.9987984,0.0008246721,0.0001176014,0.00009086046,0.0001256181,0.00004282549],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0001705331,0.00005177778,0.00630828,0.00004566908,0.0001131167,0.0002007687,0.00009775143,0.9522559,0.002530227,0.004204918,0.001019464,0.0330015],"study_design_scores_gemma":[0.00000306014,0.000009134576,0.0005493952,0.000003845747,0.000007785429,0.00001247615,0.000004051573,0.9969505,0.0002721105,0.002101629,0.00008217074,0.000003814458],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5500718,0.001683766,0.4411027,0.001691914,0.0001000283,0.00009255505,0.0008563127,0.001793509,0.002607326],"genre_scores_gemma":[0.982107,0.0002371203,0.01572998,0.00009568764,0.00002585406,0.00005786009,0.0006063587,0.00008513054,0.001054977],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01978761,"threshold_uncertainty_score":0.03934485,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1045108701252096,"score_gpt":0.2544230166117626,"score_spread":0.149912146486553,"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."}}