Background The diagnosis of inconspicuous osteoblastic metastases (OBMs) is a challenge in computed tomography (CT) images. The virtual monochromatic spectral (VMS) image of spectral CT is useful for the detection of the low-contrast lesions. Purpose To select the optimal monochromatic level for VMS images of spectral CT for detecting and diagnosing inconspicuous OBMs of the vertebra from lung cancer. Material and Methods Thirty-five patients underwent spectral CT for chest or abdomen. The CT number and standard deviation (SD) of lesions and adjacent normal bone and the SD value of subcutaneous fat were measured on the conventional polychromatic image (140 kVp) and 40-140 keV VMS images. The contrast-to-noise ratio (CNR) was compared among the 11 VMS images and 140 kVp images. The scores of two observers for different images and the inter-observer agreement were evaluated. The diameter and CNR of the detected and missed lesions were assessed. Results The lowest image noise was distributed in 70 and 140 keV images and the highest CNR was noted in 70 keV images. Good and moderate inter-observer agreement were identified for the evaluation of diagnostic ability, and the subjective scores of two observers for 60 and 70 keV images were increased compared with 140 kVp images ( P < 0.05). The diameter had no significant difference between the detected and missed lesions. The CNR of the missed lesions was reduced compared with detected lesions. Conclusion VMS images at 70 keV may be optimal for detecting and diagnosing inconspicuous OBMs from lung cancer.
Keywords: Spectral computed tomography (CT); contrast-to-noise ratio; image noise; lung cancer; monochromatic imaging; osteoblastic metastases.