Primary interpretation of thoracic MDCT images using coronal reformations.

OBJECTIVE

The objective of this study was to evaluate the accuracy and efficiency of primary interpretation of thoracic MDCT using coronal reformations as compared with transverse images.

SUBJECTS AND METHODS

Fifty patients (18 females, 32 males; age range, 15-93 years; mean age, 63.6 years) underwent 4-MDCT of the chest (detector width, 1 mm; beam pitch, 1.5). Contrast material was administered in 20 of the 50 patients. Coronal and transverse sections were reformatted into 5-mm-thick sections at 3.5-mm intervals. All available image and clinical data consensually reviewed by two thoracic radiologists served as the reference standard. Subsequently, three other thoracic radiologists independently evaluated reformatted coronal and transverse images at two separate review sessions. Each image set was assessed in 58 categories for abnormalities of the lungs, mediastinum, pleura, chest wall, diaphragm, abdomen, and skeleton. Interpretation times and number of images assessed were recorded. Sensitivity, specificity, and interobserver concordance were calculated. Differences in mean sensitivities and specificities were evaluated with Wilcoxon's signed rank test.

RESULTS

The most common findings identified were pulmonary nodules (n = 73, transverse images; n = 72, coronal images) and emphysema (n = 45, transverse; n = 40, coronal). The mean detection sensitivity of all lesions was significantly (p = 0.001) lower on coronal (44% +/- 26% [SD]) than on transverse (51% +/- 22%) images, whereas the mean detection specificity was significantly (p = 0.005) higher (96% +/- 5% vs 95% +/- 6%, respectively). Reporting findings for significantly (p < 0.001) fewer coronal images (mean, 63.0 +/- 4.6 images) than transverse images (mean, 91.9 +/- 8.8 images) took significantly (p = 0.025) longer (mean, 263 +/- 56 sec vs 238 +/- 45 sec, respectively).

CONCLUSION

Primary interpretation of thoracic MDCT is less sensitive and more time-consuming using 5-mm-thick coronal reformations as compared with transverse images.