RSNA 2015

Abstract Archives of the RSNA, 2015


RC509-06

Dual-Energy CT and Virtual Monoenergetic Reconstructions: Utility of Novel and Basic Algorithms in Assessment of Intestinal Wall Enhancement and Applications for Acute Intestinal Ischemia

Wednesday, Dec. 2 9:40AM - 9:50AM Location: E350



FDA Discussions may include off-label uses.

Pedro Lourenco, MD, Vancouver, BC (Presenter) Nothing to Disclose
Ryan Rawski, BSc, MSc, Vancouver, BC (Abstract Co-Author) Nothing to Disclose
Patrick D. McLaughlin, FFRRCSI, Cork, Ireland (Abstract Co-Author) Speaker, Siemens AG
Tim O'Connell, MD, Meng, Vancouver, BC (Abstract Co-Author) President, Resolve Radiologic Ltd; Speake, Siemens AG
Savvas Nicolaou, MD, Vancouver, BC (Abstract Co-Author) Institutional research agreement, Siemens AG
PURPOSE

Acute bowel ischemia and infarction are devastating abdominal emergencies, with reported mortality rates up to 93%. CT sensitivity for detection of acute bowel ischemia is poor, in particular in the diagnosis of early bowel ischemia, and CT findings are often non-specific. Virtual monoenergetic reconstructions allow for optimized enhancement and evaluation of bowel wall for signs of ischemia and infarction. The basic algorithm has limited utility given high noise and signal to noise. Here, we evaluate the utility of a novel (VMI+) and the basic (VMI) virtual monoenergetic algorithms in acute bowel ischemia.

METHOD AND MATERIALS

18 patients with pathologically confirmed bowel ischemia or infarction presented to a quaternary hospital. Abdominal DECT (100 and 140 keV) were obtained at the time of presentation. Axial series were reconstructed with VMI+ and VMI software application (Monoenergetic Basic and Plus, Dual Energy, Siemens) and evaluated for improved noise reduction, and the reconstructions were compared with virtual 120-keV series that blended spectral information from high and low keV datasets. Images were considered to lie within the sweet spot if noise level was < 40 HU.

RESULTS

Utilizing the novel algorithm broadened the sweet spot of diagnostically acceptable monoenergetic keV levels by 416%. With VMI+, the mean diagnostic range was 57-190 keV (SD 9.3 and 0.0, respectively), whereas using VMI, mean diagnostic range was 69 -101 keV (SD 3.9 and 13.0, respectively). SNR and CNR were also significantly improved utilizing the VMI+ technique, by 107 and 76%, respectively. CNR utilizing the VMI+ algorithm at 50, 100 and 150 keV was 4.90 (SD 1.44) 3.18 (SD 1.44) 1.26 (SD 0.71), respectively, while VMI algorithm CNR was significantly inferior, at 2.39 (SD 0.92), 1.54 (SD 0.61) and 0.35 (SD 0.18). Intestinal wall enhancement was maximized at 40 keV, given the maximal CNR (5.18 SD 1.42), which allows for optimal assessment of bowel wall at this level, albeit tolerating lower SNR.

CONCLUSION

The "sweet spot" for virtual monoenergetic reconstructions was significantly increased when utilizing the VMI+ algorithm, with a diagnostic keV range increased by approximately 400%. SNR and CNR also demonstrate marked improvement by 107 and 76%, respectively, with VMI+ over VMI.

CLINICAL RELEVANCE/APPLICATION

The VMI+ reconstructions are markedly superior to the basic VMI algorithm, and are useful in assessing bowel wall enhancement.