Purpose
To evaluate the ability of CAD prototype to detect proven malignant findings on prior chest radiograph and to determine the factors affecting the sensitivity of the algorithm.
Methods and Materials
35 digital chest-radiographs with biopsy proven findings were retrospectively analyzed. Two independent expert readers identified on the radiographs 43 findings confirmed by CT, 35 of which were malignant (15 nodules [5-30mm] and 20 masses). The readers assigned each finding a 3-grade visibility- score (0=least visible). For each patient, the time interval between the first available chest radiograph and biopsy was recorded.All chest radiographs were evaluated using a prototype CAD-algorithm (Siemens), which marked suspicious findings on each radiograph (Fig. 1).The performance of the CAD algorithm was...
Results
The overall sensitivity of the CAD algorithm was 86% (median lesion size 39.3mm, mean visibility-score 2.3). As shown in Fig. 1, masses compared to nodules had significantly higher median size (50.2 vs. 20.4 mm, p<0.03) and mean visibility-score (2.5 vs. 2.0, p<0.001). Nevertheless, the sensitivity for masses (92.3%), as shown in Fig 2,was non-significantly higher (p=0.097) than for nodules (76.5%). The improvement in sensitivity for readily visible lesions (94.7%) compared to less and moderately visible lesions (79.2%) was also non-significant (p=0.063). Six malignant cases had...
Conclusion
Both lesion size and visibility had a non-significant impact on CAD sensitivity for thoracic lesions. This may have had an impact on the ability of CAD to detect malignant chest findings on earlier radiographs up to 59 weeks prior to biopsy.
References
1. White CS, Flukinger T, Jeudy J, Chen JJ. Use of a computer-aided detection system to detect missed lung cancer at chest radiography. Radiology 2009 ;252(1):273-81.2. Li F, Engelmann R, Metz CE, Doi K, MacMahon H. Lung cancers missed on chest radiographs: results obtained with a commercial computer-aided detection program.Radiology 2008 ;246(1):273-80.3. De Boo DW, Prokop M, Uffmann M, van Ginneken B, Schaefer-Prokop CM. Computer-aided detection (CAD) of lung nodules and small tumours on chest radiographs.Eur J Radiol. 2009 ;72(2):218-25.
Personal Information
Dorith Shaham, M.D.Head, Center for Cardiothoracic ImagingDepartment of RadiologyHadassah-Hebrew University Medical Center, Ein-KeremPOB 12000Jerusalem, Israel 91120Tel: 972-2-677-8157, 972-2-677-6901Fax: 972-2-6437531Email:
[email protected]