{"ID":5937961,"CreatedAt":"2026-07-07T03:14:33.014478982Z","UpdatedAt":"2026-07-07T12:19:32.242771905Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.03853","arxiv_id":"2607.03853","title":"CogRad: A Cognitively-Inspired Multi-Agent Framework for Radiology Report Generation","abstract":"Automated radiology report generation (RRG) can ease radiologist workload, yet most existing systems produce a report in a single forward pass, with no mechanism to check a claim against the image or revisit a finding once stated. We present CogRad, a cognitively inspired multi-agent framework that structures generation around four stages of a radiologist's reading process. A Scout agent discovers anatomical regions directly from image patches via slot attention and assigns region and disease-level triage scores; an Investigator agent concentrates representational capacity on the regions Scout flags as suspicious; a Writer agent compiles these signals into a disease gated visual prefix for a large language model; and a Verifier agent supervises training with a visual entailment loss and, at inference, re-examines its own draft sentence by sentence, regenerating any report it judges insufficiently grounded. On CheXpert Plus, CogRad attains a BLEU-4 of 0.316 and a CIDEr of 0.322, the best scores among the methods we compare against. On IU X-Ray, it attains a BLEU-4 of 0.201 and a CIDEr of 0.724, leading every baseline on every standard NLG metric. We further evaluate CogRad with RadGraph F1, CheXbert F1, and a hallucination analysis to assess clinical accuracy beyond standard text-overlap metrics, complemented by ablation studies and Grad-CAM-based visualizations that characterize each agent's contribution and the model's visual grounding.","short_abstract":"Automated radiology report generation (RRG) can ease radiologist workload, yet most existing systems produce a report in a single forward pass, with no mechanism to check a claim against the image or revisit a finding once stated. We present CogRad, a cognitively inspired multi-agent framework that structures generatio...","url_abs":"https://arxiv.org/abs/2607.03853","url_pdf":"https://arxiv.org/pdf/2607.03853v1","authors":"[\"Saif Ur Rehman Khan\",\"Hasaan Maqsood\",\"Sebastian Vollmer\",\"Andreas Dengel\",\"Muhammad Nabeel Asim\"]","published":"2026-07-04T12:43:04Z","proceeding":"cs.CV","tasks":"[\"cs.CV\"]","methods":"[\"Language Model\"]","has_code":false}