Detection sensitivity of a commercial lung nodule CAD system in a series of pathologically proven lung cancers.

PURPOSE: To evaluate the performance of a commercially available computer-aided detection (CAD) system in a series of pathologically proven lung cancers.
MATERIALS AND METHODS: Sixty-nine chest computed tomography (CT) scans obtained in 12 subjects (8 females, 4 males, age 51 to 75 y, mean 63 y) with 15 pathologically proven lung cancers were retrospectively selected from 2156 entry and follow-up CT scans from a lung cancer screening program. CT scans were retrospectively analyzed using a commercially available CAD system for detecting lung nodules.
RESULTS: When first detectable proven lung cancer nodules ranged in maximum diameter from 3 to 38 mm (10.4+/-9.2 mm) with CAD detection sensitivity stratified by size: 0/2 (0%) < or =3 mm, 5/8 (62.5%) 4 to 10 mm, 2/3 (66.7%) 11 to 15 mm, 0/0 16 to 20 mm, 2/2 (100%) >20 mm, and overall sensitivity 9/15 (60%). The sensitivity for all CT scans (first detectable and follow-up), stratified by nodule size as above, was, respectively, 0/2, 18/25, 24/28, 6/9, 5/5, and overall 53/69 (76.8%). Excluding nodules <4 mm and pure ground-glass nodules, the sensitivity for all CT scans by size was 18/24 (75%) 4 to 10 mm, 21/22 (95.4%) 11 to 15 mm, 6/6 (100%) 16 to 20 mm, 5/5 (100%) >20 mm, and overall 50/57 (87.7%). At resection (13) or biopsy (2) nodules were: adenocarcinoma (10), squamous cell carcinoma (3), and small cell carcinoma (2).
CONCLUSIONS: The CAD system showed good sensitivity for solid and semisolid cancers > or =4 mm (sensitivity 87.7%) and excellent for those > or =11 mm (sensitivity >95.4%).