Study Design
A consecutive sample of 24 fetuses underwent brain MR imaging (MRI) at 36 ± 0.8 weeks of gestational age in 1.5 T Siemens scanners (Siemens Healthcare, Erlangen, Germany) from different hospitals in Spain: Vall d’Hebron University Hospital, Barcelona; Hospital Clínic, Barcelona; and Hospital La Paz, Madrid. This sample included 14 fetuses previously diagnosed with congenital heart diseases, and 10 control fetuses. The protocol (PR-AMI-317/2012) was locally approved by each institutional committee: 2012/225, 2015/84, and 2014/3, respectively. After anatomical T2-weighted MRI in the 3 orthogonal planes, single voxel point resolved spectroscopy proton MRS data were acquired from the frontal lobe region (15 × 22 × 38 mm voxel) ( Figure , A) with 1600-millisecond repetition time and 136-millisecond echo time, partial chemical shift selective water suppression, and in 17 free-induction-decay blocks, 8 transients each. The total acquisition time was 3.6 minutes. A reference water spectrum was also obtained (chemical shift selective module off), followed by an axial T2-weighted MRI, to confirm the fetal head position. The protocol was repeated if gross movements and/or spectral artifacts were detected. MRS data were processed in 2 ways using MATLAB R2010a (MathWorks Inc, Natick, MA): method A, standard averaging of 128 transients (first free-induction-decay block discarded); and method B, spectral sorting (discarding data with gross artifacts and lipid contaminations) and alignment (based on the residual water peak), before averaging. Spectra were then analyzed for quality assessment and selection: both with jMRUI v4.0 (MRUI Consortium, EU), spectral pattern well resolved for major peaks, such as choline compounds, total creatine, and N-acetylaspartate, and amplitude ratio of lipid (1.3-0.9 ppm) to choline peak (3.21 ppm) <10%; and with LCModel v6.3 (Stephen Provencher Inc, Oakville, Canada), peak full-width at half-maximum (FWHM) <0.1 ppm and signal-to-noise ratio (SNR) ≥3. Statistical analyses comparing methods A and B were carried out with a 2-tailed t test: paired and unpaired, for full dataset and quality-selected spectra, respectively (SPSS v19.0; IBM, Armonk, NY).