Automatic contrast medium extraction system using electron density data with dual-energy CT.

OBJECTIVE:: The purpose of the current study is to create a contrast medium extraction method using raw-data-based electron density (rED) and CT number from dual-energy CT (DECT) for automatic delineation of the contrast region. METHODS:: A CT-ED phantom containing tissue-equivalent inserts and an acrylic phantom with an iodinated contrast medium were scanned by DECT. The contrast medium extraction system was created using Python. The accuracy of the contrast medium extraction was evaluated by measuring the diameter in terms of the full width at half maximum (FWHM) and the ratio of the volume (ROV). RESULTS:: Mean-2SD CT numbers and the difference of the CT numbers (DCT) of the contrast medium at 0-130 mg ml-1 contrast medium concentration and the bone materials were more than -33 and -20 HU, respectively. In the correlation of rED and CT number, the gradient with the contrast medium phantom was greater than that with the CT-ED phantom. The accuracy of the contrast medium at 80 kV/135 kV and 100 kV/135 kV tube voltages. The gradient of the CT-ED and contrast medium phantoms were different. The gradient in the CT-ED phantom and the contrast medium was 0.0012 and 0.0003 at 80 kV/135 kV, and 0.0015 and 0.0005 at 100 kV/135 kV tube voltages, respectively. The ratio of the measured to the actual diameter in FWHM and ROV was 0.98-1.00 at 2-130 mg ml-1. At a tube voltage of 100 kV/135 kV. The ratio of the measured to the actual diameter in ROV was 0.66 and FWHM was 0.90 at 2 mg ml-1 contrast medium concentration. The ratio of the measured to the actual diameter in FWHM and ROV was 0.98-1.00 at 3-130 mg ml-1. CONCLUSION:: We created the contrast medium extraction method with rED and CT number images. The contrast medium extraction method could be used with DECT images at 80 kV/135 kV. The method is expected to only extract images from the region containing the contrast medium. ADVANCES IN KNOWLEDGE:: We created the contrast medium extraction method using raw-data-based electron density and CT number from DECT and it is expected to only extract information from the region containing the contrast medium.