Click here to


Are you sure ?

Yes, do it No, cancel

Automatic Contrast Medium Extraction System Using Monochromatic CT Number

D Kawahara1*, S Ozawa2 , S Tanaka3 , K Yokomachi4 , A Saito2 , T Higaki5 , C Fujioka4 , M Tsuneda1 , M Mori1 , Y Ohno4 , Y Nagata2 , (1)Medical and Dental Sciences Course, Graduate School of Biomedical & Health Sciences, Hiroshima University, Hiroshima City 734-8551, Japan, (2)MDepartment of Radiation Oncology, Institute of Biomedical & Health Sciences, Hiroshima University, Hiroshima City 734-8551, Japan, (3) Department of Nuclear Engineering and Management, School of Engineering, University of Tokyo, Tokyo 113-8654, Japan, (4) Radiation Therapy Section, Division of Clinical Support, Hiroshima University Hospital, Hiroshima City 734-8551, Japan, (5) Departments of Diagnostic Radiology and Radiology, Hiroshima University, Hiroshima City 732-0057, Japan


(Wednesday, 8/1/2018) 10:00 AM - 10:30 AM

Room: Exhibit Hall | Forum 6

Purpose: To extract the contrast medium in human, we create automatic contrast medium extraction method using the monochromatic CT (mCT) number from dual-energy CT (DECT).

Methods: A CT-ED phantom (CIRS model 062M) that contains tissue-equivalent inserts and a multipurpose chest phantom with added fat slabs (N1 “LUNGMAN�; Kyoto Kagaku) which contains inserts with various concentrations (0-130 mg/mL), of iodinated contrast medium were scanned by DECT (Toshiba, Aquilion ONE CT Scanner). The mCT of the 40, 60 and 130 keV images were reconstructed from CT scans at tube voltages of 80 kV and 135 kV. The contrast medium was separated from other tissue materials using the threshold mCT numbers and the gradient the mCT number (GmCT), which was analyzed using an in-house software with Python. The accuracy of the iodine extraction was evaluated the ratio of the volume (ROV).

Results: The mCT number of the contrast medium and liver, muscle, and bone materials in the CT-ED phantom was obviously greater than 0 HU and mCT number of liver, muscle and bone materials in CT-ED phantom. The GmCT was over 6.0 in the contrast medium and within 4.0 in the CT-ED phantom. The accuracy of the limit concentration of the contrast medium was 1 mg/mL and the ROV was 0.97-1.00, as shown in Fig. 5.

Conclusion: We created a contrast-medium extraction method that is expected to extract only the contrast-medium region from a patient’s CT image.


Dual-energy Imaging


IM/TH- Image Analysis (Single modality or Multi-modality): Image segmentation

Contact Email