Fast automatic segmentation of thalamic nuclei from MP2RAGE acquisition at 7 Tesla

Abstract

Thalamic nuclei are largely invisible in conventional MRI due to poor contrast. Thalamus Optimized Multi-Atlas Segmentation (THOMAS) provides automatic segmentation of 12 thalamic nuclei using white-matter-nulled (WMn) Magnetization Prepared Rapid Gradient Echo (MPRAGE) sequence at 7T, but increases overall scan duration. Routinely acquired, bias-corrected Magnetization Prepared 2 Rapid Gradient Echo (MP2RAGE) sequence yields superior tissue contrast and quantitative T1 maps. Application of THOMAS to MP2RAGE has been investigated in this study. Methods: Eight healthy volunteers and five pediatric-onset multiple sclerosis patients were recruited at Children's Hospital of Philadelphia and scanned at Siemens 7T with WMn-MPRAGE and multi-echo-MP2RAGE (ME-MP2RAGE) sequences. White-matter-nulled contrast was synthesized (MP2-SYN) from T 1 maps from ME-MP2RAGE sequence. Thalamic nuclei were segmented using THOMAS joint label fusion algorithm from WMn-MPRAGE and MP2-SYN datasets. THOMAS pipeline was modified to use majority voting to segment bias corrected T1-weighted uniform (MP2-UNI) images. Thalamic nuclei from MP2-SYN and MP2-UNI images were evaluated against corresponding nuclei obtained from WMn-MPRAGE images using dice coefficients, volume similarity indices (VSIs) and distance between centroids. Results: For MP2-SYN, dice > 0.85 and VSI > 0.95 was achieved for five larger nuclei and dice > 0.6 and VSI > 0.7 was achieved for seven smaller nuclei. The dice and VSI were slightly higher, whereas the distance between centroids were smaller for MP2-SYN compared to MP2-UNI, indicating improved performance using the MP2-SYN image. Conclusions: THOMAS algorithm can successfully segment thalamic nuclei in MP2RAGE images with essentially equivalent quality as WMn-MPRAGE, widening its applicability in studies focused on thalamic involvement in aging and disease.

Automatic segmentation of deep grey nuclei using a high‐resolution 7T magnetic resonance imaging atlas—Quantification of T1 values in healthy volunteers

“…To evaluate inter‐expert variability and the integration of this variability in the construction of the consensus‐based 7TAMIbrain DGN atlas, three metrics were computed for each segmented region between the rater atlases 7TAMIbrain R1 , 7TAMIbrain R2 and 7TAMIbrain DGN . These included the Dice similarity coefficient (DSC), volumetric similarity index defined as the ratio of the absolute volume difference divided by the sum of the compared volumes (referred to as VOLSMTY in Taha & Hanbury, 2015 and equivalent to the VSI metric described in Datta et al, 2020) and the Balanced average Hausdorff distance (bAVD, Aydin et al, 2021, more suitable for rankings and quality assessment of segmentations than average Hausdorff distance).…”

“…Other high‐resolution approaches such as magnetization prepared with two rapid acquisition gradient echoes (MP2RAGE) (Marques et al, 2010) or MP2RAGEME (Caan et al, 2019) sequences have provided enhanced contrast based on quantitative relaxometry and susceptibility values for the characterization of grey matter structures. Datta et al (Datta et al, 2020) have very recently adapted THOMAS's algorithm to automatically segment 12 thalamic nuclei from MP2RAGE images and have demonstrated a comparable efficiency obtained with the dedicated white‐matter‐nulled MP‐RAGE images, with results on eight healthy volunteers and five paediatric‐onset multiple sclerosis.…”

Multi-scale structural alterations of the thalamus and basal ganglia in focal epilepsy as demonstrated by 7T MRI

“…To assess the impact of atlas choice on the thalamic results, thalamic nuclei were additionally segmented using THOMAS 33 . Since THOMAS is optimized to work with white matter-nulled (WMn) Magnetization Prepared with Rapid Acquisition Gradient Echoes (MPRAGE) images (i.e., enhanced thalamic-white matter contrast), MP2RAGE T 1 maps were first white matter-nulled using a recently proposed thresholding method based on the inversion recovery signal equation 38 : Here, a T 1 threshold of 700 ms. was found to provide most reliable performance as assessed through visual inspection of segmentation results using varying thresholds. The WMn T 1 maps were than used as input for the joint-fusion THOMAS pipeline (https://github.com/thalamicseg/thomas_new).…”

“…To assess the impact of atlas choice on the thalamic results, thalamic nuclei were additionally segmented using THOMAS 33 . Since THOMAS is optimized to work with white matter-nulled (WMn) Magnetization Prepared with Rapid Acquisition Gradient Echoes (MPRAGE) images (i.e., enhanced thalamic-white matter contrast), MP2RAGE T 1 maps were first white matter-nulled using a recently proposed thresholding method based on the inversion recovery signal equation 38 : …”

Memory, processing of emotional stimuli, and volume of limbic structures in pediatric-onset multiple sclerosis

“…Volumes of the hippocampus and amygdala for each hemisphere were calculated using special modules available in FreeSurfer ( Iglesias et al, 2015 , Saygin et al, 2017 ). Automatic segmentation of the thalamus was obtained on the T1-weighted images using published methods ( Datta et al, 2020 ).…”