Model-based iterative reconstruction (MBIR, GE Healthcare), approved by the FDA in September 2011, delivered a 74 percent mean dose reduction relative to standard-dose abdominal CT imaging and yielded improved image quality scores compared with other dose reduction techniques, according to a study published in this month's American Journal of Roentgenology.
Novel iterative reconstruction techniques have emerged as a promising alternative beyond the traditional reconstruction method of filtered back projection. These include adaptive statistical iterative reconstruction (GE), iDose (Philips Healthcare), iterative reconstruction in image space (IRIS, Siemens Healthcare) and adaptive iterative dose reconstruction (AIDR, Toshiba America Medical Systems). However, MBIR may provide improvements beyond these methods.
Perry J. Pickhardt, MD, from the department of radiology at University of Wisconsin School of Medicine and Public Health in Madison, and colleagues designed a prospective study to compare image quality and diagnostic utility of abdominal CT with MBIR with CT performed with adaptive statistical iterative reconstruction (ASIR) and filtered back projection reconstruction.
Preliminary results, comprising 21 standard-dose contrast-enhanced and 24 unenhanced abdominal CT exams and 45 matched ultralow-dose exams with targeted dose reduction of 70 to 90 percent, were reported. Ultralow-dose exams were reconstructed with filtered back projection, ASIR and MBIR.
Two abdominal radiologists completed focal organ-based lesion detection to assess diagnostic accuracy. Two blinded readers reviewed all ultralow-dose series. Image quality was graded on a 5-point scale, ranging from 0, nondiagnostic, to 4, well-depicted, without artifacts.
The mean and median dose reductions for the ultralow-dose series were 74 percent and 78 percent, respectively. The mean and median effective doses for ultralow-dose studies were 1.90 mSv and 1.11 mSv.
The standard-dose CT exams used ASIR-driven protocols. Consequently, the dose reductions were less than they would have been if filtered back projection-driven protocols had been employed as the standard, according to Pickhardt et al.
Image noise was significantly lower for MBIR than standard-dose filtered back projection and ASIR. Low-dose MBIR also delivered significantly higher mean subjective image quality scores than low-dose filtered back projection and ASIR.
“Our preliminary findings show that MBIR is a substantial improvement over ASIR and filtered back projection in terms of image noise, subjective image quality and diagnostic performance,” wrote Pickhardt and colleagues.
The diagnostic performance of ASIR, which is not intended for dose reduction levels used in the trial, was slightly worse than low-dose filtered back projection in terms of lesion detection and much poorer than MBIR. Other, vendor-specific, non-model-based algorithms may face similar limitations. “Further investigation for these algorithms is needed to determine the radiation dose levels at which diagnostic adequacy can be maintained relative to the current clinical reference standard in terms of lesion detection capability,” wrote Pickhardt and colleagues.
The researchers also detailed potential limitations of MBIR, particularly its lengthy reconstruction time and vendor-specific approach. Nevertheless, Pickhardt and colleagues referred to the "great promise" of MBIR and called for additional research to determine optimal indication-specific dose levels to maintain diagnostic performance.