Reliability of Functional and Diffusion MR Imaging Near Cerebral Cavernous Malformations.

BACKGROUND AND PURPOSE: Surgical resection of cerebral cavernous malformations close to eloquent regions frequently uses fMRI and DTI for surgical planning to best preserve neurologic function. This study investigates the reliability of fMRI and DTI near cerebral cavernous malformations. MATERIALS AND METHODS: Consecutive patients with cerebral cavernous malformations undergoing presurgical fMRI and DTI mapping were identified. Each cerebral cavernous malformation was hand-contoured; 2 sequential 4-mm expansion shells (S1 and S2) were created, generating 2 ROIs and 2 contralateral controls. Fractional anisotropy and regional homogeneity measurements were then extracted from each ROI and compared with the contralateral controls. Reliability, accuracy, and precision were compared as appropriate. RESULTS: Fifty-four patients were identified and included. Errors of fractional anisotropy were significantly lower than those of regional homogeneity in S1 and S2 (P < .001), suggesting that fractional anisotropy is more reliable than regional homogeneity near cerebral cavernous malformations. Proximity to cerebral cavernous malformations worsened the reliability of regional homogeneity (S1 versus S2, P < .001), but not fractional anisotropy (P = .24). While fractional anisotropy was not significantly biased in any ROI (P > .05), regional homogeneity was biased toward lower signals in S1 and S2 (P < .05), an effect that was attenuated with distance from cerebral cavernous malformations (P < .05). Fractional anisotropy measurements were also more precise than regional homogeneity in S1 and S2 (P < .001 for both). CONCLUSIONS: Our findings suggest that hemosiderin-rich lesions such as cerebral cavernous malformations may lead to artifactual depression of fMRI signals and that clinicians and surgeons should interpret fMRI studies near cerebral cavernous malformations with caution. While fMRI is considerably affected by cerebral cavernous malformation-related artifacts, DTI appears to be relatively unaffected and remains a reliable imaging technique near cerebral cavernous malformations.

Spinal cord microstructure integrating phase-sensitive inversion recovery and diffusional kurtosis imaging.

PURPOSE: The aim of this prospective study was to determine the feasibility in terms of repeatability and reproducibility of diffusional kurtosis imaging (DKI) for microstructural assessment of the normal cervical spinal cord (cSC) using a phase-sensitive inversion recovery (PSIR) sequence as the anatomical reference for accurately defining white-matter (WM) and gray-matter (GM) regions of interests (ROIs). METHODS: Thirteen young healthy subjects were enrolled to undergo DKI and PSIR sequences in the cSC. The repeatability and reproducibility of kurtosis metrics and fractional anisotropy (FA) were calculated in GM, WM, and cerebral-spinal-fluid (CSF) ROIs drawn by two independent readers on PSIR images of three different levels (C1-C4). The presence of statistically significant differences in DKI metrics for levels, ROIs (GM, WM, and CSF) repeatability, reproducibility, and inter-reader agreement was evaluated. RESULTS: Intra-class correlation coefficients between the two readers ranged from good to excellent (0.75 to 0.90). The inferior level consistently had the highest concordance. The lower values of scan-rescan variability for all DKI parameters were found for the inferior level. Statistically significant differences in kurtosis values were not found in the lateral white-matter bundles of the spinal cord. CONCLUSION: The integration of DKI and PSIR sequences in a clinical MR acquisition to explore the regional microstructure of the cSC in healthy subjects is feasible, and the results obtainable are reproducible. Further investigation will be required to verify the possibility to translate this method to a clinical setting to study patients with SC involvement especially in the absence of MRI abnormalities on standard sequences.

Comparing 3T T1-weighted sequences in identifying hyperintense punctate lesions in preterm neonates.

BACKGROUND AND PURPOSE: The loss of contrast on T1-weighted MR images at 3T may affect the detection of hyperintense punctate lesions indicative of periventricular leukomalacia in preterm neonates. The aim of the present study was to determine which 3T T1-weighted sequence identified the highest number of hyperintense punctate lesions and to explore the relationship between the number of hyperintense punctate lesions and clinical outcome. MATERIALS AND METHODS: The presence of hyperintense punctate lesions was retrospectively evaluated in 200 consecutive preterm neonates on 4 axial T1-weighted sequences: 3-mm inversion recovery and spin-echo and 1- and 3-mm reformatted 3D-fast-field echo. Statistically significant differences in the number of hyperintense punctate lesions were evaluated by using a linear mixed-model analysis. Logistic regression analysis was used to assess the relation between the number of hyperintense punctate lesions and neuromotor outcome at 3 months. RESULTS: Thirty-one neonates had at least 1 hyperintense punctate lesion indicative of periventricular leukomalacia in at least 1 of the 4 sequences. The 1-mm axial reformatted 3D-fast-field echo sequence identified the greatest number of hyperintense punctate lesions (P < .001). No statistically significant differences were found among the 3-mm T1-weighted sequences. The greater number of hyperintense punctate lesions detected by the 1-mm reformatted T1 3D-fast-field echo sequence in the central region of the brain was associated with a worse clinical outcome. CONCLUSIONS: At 3T, the 1-mm axial reformatted T1 3D-fast-field echo sequence identified the greatest number of hyperintense punctate lesions in the central region of preterm neonate brains, and this number was associated with neuromotor outcome.