Reduced Dose Scans Not Ready to Detect Liver Lesions
Wednesday, Dec. 02, 2015
Sinogram-affirmed iterative reconstruction (SAFIRE) and other image reconstruction techniques aren't quite ready to perform reduced-dose abdominal CT scans that are diagnostically equivalent to standard-dose scans, according to research presented Tuesday.
A head-to-head comparison of standard CT scans and reduced dose scans with SAFIRE showed that while the reduced-dose (RD) scans were adequate for detecting some types of lesions, they could miss others, including a liver mass, according to a prospective study of 20 patients at Massachusetts General Hospital.
However, RD scans may be equivalent, or in some cases superior, to standard-dose (SD) scans in detecting kidney stones, gallstones and diverticulosis.
"Further innovations in CT data acquisition or image reconstruction are needed to reduce abdominal CT doses to sub-millisievert levels," the research team concluded in the study, "Assessment of Sinogram-affirmed Iterative Reconstruction Techniques for Reduced Dose Abdomen CT."
The study compared RD images using radiation doses of less than 2 mGy, as compared to 9 mGy with SD CT. Increasing the dose slightly could yield significant improvement in the images, said presenter Atul Padole, M.D., a radiology research fellow at MGH. He plans a follow-up study with a dose of 2.6 to 3 mGy. "We need to find the dose level where we won't miss these lesions," he said.
The patients, 11 men and nine women with a mean age of 68 and a mean body mass index of 25, received both SD and RD 128-slice MDCT scans in succession. The RD images were reconstructed with SAFIRE at settings S1, S3 and S5, yielding a total of 80 studies (20 at standard dose and 20 at each of the three SAFIRE settings). The settings for the two scans were identical except for the tube current. The standard scans had a mean dose of 9 mGy (±3) and the RD scans had a mean dose of 1.4 mGy (±0.1). The mean effective dose for the standard scans was 6 mSv (±1.6), and the mean effective dose for the reduced scans was 0.9 mSv (±0.02).
Radiologists performed independent, random, and blinded comparison for lesion detection, lesion conspicuity, and visibility of abdominal structures, first for all patients on RD images and subsequently on SD images.
Out of 72 true lesions detected with standard scans, five were missed on the reduced-dose scans regardless of the SAFIRE settings and BMI of the patients. All the missed lesions were in the liver. The RD scans also picked up one false positive liver lesion on all settings. Some lesion types, such as renal calculi, gallstones, and diverticulosis, could be accurately assessed on the RD images.
Dr. Padole showed one kidney stone image that was deemed to be optimal on the RD scans at all SAFIRE settings, compared with the SD scan. Visibility of abdominal structures was also adequate with the RD images at all SAFIRE settings. Setting S3 yielded the best RD image quality for lesion conspicuity, liver parenchyma, and renal parenchyma.
SAFIRE, which is proprietary to Siemens, is not the only image reconstruction technique to show these limitations, Dr. Padole said. He has worked with systems from other manufacturers—Phillips, GE, and Toshiba—and has seen missed or suboptimal abdominal findings with all of them.