作者: Darren Boone , Susan Mallett , Justine McQuillan , Stuart A. Taylor , Douglas G. Altman
DOI: 10.1371/JOURNAL.PONE.0136624
关键词: Population 、 Radiology 、 CAD 、 Reference standards 、 Colonoscopy 、 Medicine 、 Computer aided detection
摘要: OBJECTIVES To quantify the incremental benefit of computer-assisted-detection (CAD) for polyps, inexperienced readers versus experienced CT colonography. METHODS 10 and 16 radiologists interpreted 102 colonography studies unassisted with CAD utilised in a concurrent paradigm. They indicated any polyps detected on study sheet. Readers' interpretations were compared against ground-truth reference standard: 46 normal 56 had at least one polyp (132 total). The primary outcome was difference net (a combination change sensitivity specificity CAD, weighted towards sensitivity) detection patients polyps. RESULTS Inexperienced readers' per-patient rose from 39.1% to 53.2% fell 94.1% 88.0%, both statistically significant. Experienced 57.5% 62.1% 91.0% 88.3%, non-significant. Net assistance significant but not readers: 11.2% (95%CI 3.1% 18.9%) 3.2% -1.9% 8.3%) respectively. CONCLUSIONS Concurrent resulted when used by identify nearly four times magnitude that observed readers. did significantly CAD.