Effect of reconstruction algorithm on image quality and identification of ground-glass opacities and partly solid nodules on low-dose thin-section CT: experimental study using chest phantom.

PURPOSE

The purpose of this study was to assess the influence of reconstruction algorithm on identification and image quality of ground-glass opacities (GGOs) and partly solid nodules on low-dose thin-section CT.

MATERIALS AND METHODS

A chest CT phantom including simulated GGOs and partly solid nodules was scanned with five different tube currents and reconstructed by using standard (A) and newly developed (B) high-resolution reconstruction algorithms, followed by visually assessment of identification and image quality of GGOs and partly solid nodules by two chest radiologists. Inter-observer agreement, ROC analysis and ANOVA were performed to compare identification and image quality of each data set with those of the standard reference. The standard reference used 120 mA s in conjunction with reconstruction algorithm A.

RESULTS

Kappa values (kappa) of overall identification and image qualities were substantial or almost perfect (0.60<kappa). Assessment of identification showed that area under the curve of 25 mA reconstructed with reconstruction algorithm A was significantly lower than that of standard reference (p<0.05), while assessment of image quality indicated that 50 mA s reconstructed with reconstruction algorithm A and 25 mA s reconstructed with both reconstruction algorithms were significantly lower than standard reference (p<0.05).

CONCLUSION

Reconstruction algorithm may be an important factor for identification and image quality of ground-glass opacities and partly solid nodules on low-dose CT examination.