MR venography of the fetal brain using susceptibility weighted imaging

Abstract

Purpose: To evaluate the feasibility of performing fetal brain magnetic resonance venography using susceptibility weighted imaging (SWI).

Materials and methods: After obtaining informed consent, pregnant women in the second and third trimester were imaged using a modified SWI sequence. Fetal SWI acquisition was repeated when fetal or maternal motion was encountered. The median and maximum number of times an SWI sequence was repeated was four and six respectively. All SWI image data were systematically evaluated by a pediatric neuroradiologist for image quality using an ordinal scoring scheme: 1. diagnostic; 2. diagnostic with artifacts; and 3. nondiagnostic. The best score in an individual fetus was used for further statistical analysis. Visibility of venous vasculature was also scored using a dichotomous variable. A subset of SWI data was re-evaluated by the first and independently by a second pediatric neuroradiologist. Kappa coefficients were computed to assess intra-rater and inter-rater reliability.

Results: SWI image data from a total of 22 fetuses were analyzed. Median gestational age and interquartile range of the fetuses imaged were 32 (29.9-34.9) weeks. In 68.2% of the cases (n = 15), there was no artifact; 22.7% (n = 5) had minor artifacts and 9.1% (n = 2) of the data was of nondiagnostic quality. Cerebral venous vasculature was visible in 86.4% (n = 19) of the cases. Substantial agreement (Kappa = 0.73; 95% confidence interval 0.44-1.00)) was observed for intra-rater reliability and moderate agreement (Kappa = 0.48; 95% confidence interval 0.19-0.77) was observed for inter-rater reliability.

Conclusion: It is feasible to perform fetal brain venography in humans using SWI.

Keywords: cerebral; fetal imaging; gradient echo; phase; pregnancy; veins.

Figure 1

Susceptibility weighted images of the fetal brain. Top Row: Images were acquired in axial orientation relative to the fetus using a 3D SWI sequence at 3.0T field strength from a fetus at 36 3/7 weeks of gestation. Bottom Row: Images were acquired in coronal orientation relative to the fetus using a 2D SWI sequence at 1.5T field strength from a fetus at 35 3/7 weeks of gestation. a,d: Original magnitude images. b,e: Corresponding filtered phase images. c,f: Corresponding susceptibility weighted magnitude image (SWI image). Veins are already seen to some extent in the original magnitude images (a,d) due to T2* signal loss; white arrow head in (a) indicating the thalamostriate vein. Phase images in (b,e) also show the veins with their characteristic phase signature. After applying the mask generated using this phase information to the corresponding original magnitude images, processed SWI images in (c,f) clearly improve the vessel contrast (white arrow in (a) and black arrow head in (f)). Imaging resolution corresponding to the scan in the top row (3.0T - 2D SWI) was 0.78 × 1.5 × 3 mm³ (reconstructed to 0.78 × 0.78 mm² in-plane) and 0.7 × 0.1.4 × 5 mm³ (reconstructed to 0.7 × 0.7 mm² in-plane) for the bottom row (1.5T - 2D SWI).

Figure 2

Unprocessed SWI magnitude images illustrating typical image quality and artifacts seen and their respective scores. a: Image containing fetal motion artifact and an artifact due to small FOV; score 3. b: Image with varying signal to noise ratio across the fetal brain region. This is due to insufficient reference lines required for image reconstruction with parallel imaging; score ×2. c: Another image with score 2, containing what appears as a ripple artifact in one part of the fetal brain. This artifact is due to signal contamination from tissue outside the small field of view; score 2. d: Fetal image with visibly no artifacts in the fetal brain region; score 1.

Figure 3

Proportion of cases with visible cerebral venous vasculature by gestational age at image assessment.

Figure 4

Fetal brain venograms generated from SWI images (minimum intensity projections). Venograms from different fetuses at different gestational ages are shown to illustrate the variation of visualized cerebral venous vasculature in the human fetus with gestational age.

Figure 5

Venous vasculature in a 37 1/7 week fetus, identifiable based on adult venous anatomy. a: Septal vein (red), Thalamostriate vein (blue), Internal cerebral veins (yellow). b: medial atrial vein (green), superior sagittal sinus (orange). c: Basal vein of Rosenthal (light blue), deep middle cerebral vein (pink). Images shown here are minimum intensity projection SWI images, projected over 12 mm thickness of tissue to show contiguity of venous vasculature within. These fetal data were acquired at 3T field strength using a 3D SWI sequence. SWI images from an adult brain are shown in (d) for visual comparison of the venous anatomy