Pilot study assessing differentiation of steatosis hepatis, hepatic iron overload, and combined disease using two-point dixon MRI at 3 T: in vitro and in vivo results of a 2D decomposition technique.

OBJECTIVE: The purpose of our study was to evaluate whether two-point Dixon MRI using a 2D decomposition technique facilitates metabolite differentiation between lipids and iron in standardized in vitro liver phantoms with in vivo patient validation and allows semiquantitative in vitro assessment of metabolites associated with steatosis, iron overload, and combined disease. MATERIALS AND METHODS: The acrylamide-based phantoms were made to reproduce the T1- and T2-weighted MRI appearances of physiologic hepatic parenchyma and hepatic steatosis-iron overload by the admixture of triglycerides and ferumoxides. Combined disease was simulated using joint admixtures of triglycerides and ferumoxides at various concentrations. For phantom validation, 30 patients were included, of whom 10 had steatosis, 10 had iron overload, and 10 had no liver disease. For MRI an in-phase/opposed-phase T1-weighted sequence with TR/TE(opposed-phase)/TE(in-phase) of 4.19/1.25/2.46 was used. Fat/water series were obtained by Dixon-based algorithms. In-phase and opposed-phase and fat/water ratios were calculated. Statistical cluster analysis assessed ratio pairs of physiologic liver, steatosis, iron overload, and combined disease in 2D metabolite discrimination plots. RESULTS: Statistical assessment proved that metabolite decomposition in phantoms simulating steatosis (1.77|0.22; in-phase/opposed-phase|fat/water ratios), iron overload (0.75|0.21), and healthy control subjects (1.09|0.05) formed three clusters with distinct ratio pairs. Patient validation for hepatic steatosis (3.29|0.51), iron overload (0.56|0.41), and normal control subjects (0.99|0.05) confirmed this clustering (p < 0.001). One-dimensional analysis assessing in vitro combined disease only with in-phase/opposed-phase ratios would have failed to characterize metabolites. The 2D analysis plotting in-phase/opposed-phase and fat/water ratios (2.16|0.59) provided accurate semiquantitative metabolite decomposition (p < 0.001). CONCLUSION: MR Dixon imaging facilitates metabolite decomposition of intrahepatic lipids and iron using in vitro phantoms with in vivo patient validation. The proposed decomposition technique identified distinct in-phase/opposed-phase and fat/water ratios for in vitro steatosis, iron overload, and combined disease.

Pancreaticoduodenectomy: frequency and outcome of post-operative imaging-guided percutaneous drainage.

BACKGROUND: To study the frequency and outcomes of percutaneous imaging-guided drainage following pancreaticoduodenectomy and to assess if fluid collection location correlates with pancreatic duct leak. METHODS: IRB approval was obtained. Three hundred and seventy-three subjects (age 21-84 years) who underwent pancreaticoduodenectomy were included in this retrospective study. Eighty-three of these subjects underwent post-operative imaging-guided drainage (CT 77; US 6). Medical and imaging records were reviewed. Procedural details including collection location, size, catheter size, drain duration, fluid type, fluid chemistry, and fluid culture were recorded. Collection location was correlated with fluid amylase. RESULTS: The frequency of imaging-guided percutaneous drainage following Whipple was 22.2%. The immediate technical and overall success rates for fluid collection drainage were 97.6% and 79.6%, respectively. Rate of complication was 4.8% (4/83). 74.7% (62/83) of fluid collections were proven abscesses, and 61.4% (51/83) were complicated by pancreatic fistula. Collections near the pancreatic resection site were more likely to have elevated fluid amylase. CONCLUSION: Approximately one-fifth of subjects requires percutaneous drainage following pancreaticoduodenectomy. Percutaneous imaging-guided drainage is an effective means of managing post-pancreaticoduodenectomy fluid collections. Collections near the pancreas resection site often have a pancreatic duct leak.

Noninvasive evaluation of active lower gastrointestinal bleeding: comparison between contrast-enhanced MDCT and 99mTc-labeled RBC scintigraphy.

OBJECTIVE: The purpose of our study was to compare contrast-enhanced MDCT and (99m)Tc-labeled RBC scanning for the evaluation of active lower gastrointestinal bleeding. SUBJECTS AND METHODS: Over 17 months, 55 patients (32 men, 23 women; age range, 21-92 years) were evaluated prospectively with contrast-enhanced MDCT using 100 mL of iopromide 300 mg I/mL. Technetium-99m-labeled RBC scans were obtained on 41 of 55 patients and select patients underwent angiography for attempted embolization. Each imaging technique was reviewed in a blinded fashion for sensitivity for detection of active bleeding as well as the active lower gastrointestinal bleeding location. RESULTS: Findings were positive on both examinations in eight patients and negative on both examinations in 20 patients. Findings were positive on contrast-enhanced MDCT and negative on (99m)Tc-labeled RBC in two patients; findings were negative on contrast-enhanced MDCT and positive on (99m)Tc-labeled RBC in 11 patients. Statistics showed significant disagreement, with simple agreement = 68.3%, kappa = 0.341, and p = 0.014. Sixteen of 60 (26.7%) contrast-enhanced MDCT scans were positive prospectively, with all accurately localizing the site of bleeding and identification of the underlying lesion in eight of 16 (50%). Nineteen of 41 (46.3%) (99m)Tc-labeled RBC scans were positive. Eighteen of 41 matched patients went on to angiography. In four of these 18 (22.2%) patients, the site of bleeding was confirmed by angiography, but in 14 of 18 (77.8%), the findings were negative. CONCLUSION: Contrast-enhanced MDCT and (99m)Tc-labeled RBC scanning show significant disagreement for evaluation of active lower gastrointestinal bleeding. Contrast-enhanced MDCT appears effective for detection and localization in cases of active lower gastrointestinal bleeding in which hemorrhage is active at the time of CT.