Clinical application of low-dose CT combined with computer-aided detection in lung cancer screening

Abstract

Objective To investigate the clinical value of chest low-dose CT (LDCT) combined with computer-aided detection (CAD) system for lung cancer screening in high risk population. Methods Two hundred and nineteen healthy candidates underwent 64-slice LDCT scan. All images were reviewed in consensus by two radiologists with 15 years of thoracic CT diagnosis experience. Then the image data were analyzed with CAD alone. Finally images were reviewed by two radiologists with 5 years of CT diagnosis experience with and without CT Viewer software. The sensitivity, false positive rate of CAD for pulmonary nodule detection were calculated. SPSS 11.5 software and Chi-square test were used for the statistics. Results Of 219 candidates ,104(47.5% )were detected with lung nodules. There were 366 true nodules confirmed by the senior radiologists. The CAD system detected 271 (74.0%)true nodules and 424 false-positive nodules. The false-positive rate was 1.94/per case. The two junior radiologists indenfifid 292(79.8%), 286(78.1%) nodules without CAD and 336 (91.8%), 333 (91.0%) nodules with CAD respectively. There were significant differences for radiologists in indentifying nodules with or without CAD system (P<0.01). Conclusions CAD is more sensitive than radiologists for indentifying the nodules in the central area or in the hilar region of the lung. While radiologists are more sensitive for the peripheral and sub-pleural nodules,or ground glass opacity nodules, or nodules smaller than 4 mm. CAD can not be used alone. The detection rate can be improved with the combination of radiologist and CAD in LDCT screen. Key words: Coin lesion,pulmonary; Tomography,X-ray computed; Image processing,computer-assisted