Fig. 1: Combining computationally assisted imaging and machine learning-based analysis allows large-scale studies of normal healthy volunteer pregnancies over a long gestational range with a high yield. | Nature Communications

Fig. 1: Combining computationally assisted imaging and machine learning-based analysis allows large-scale studies of normal healthy volunteer pregnancies over a long gestational range with a high yield.

From: Motion corrected MRI differentiates male and female human brain growth trajectories from mid-gestation

Fig. 1

Accurate motion correction (a) of fast multi-planar multi-slice MRI data27,56 allows high spatial resolution (0.5 mm cubic voxels) 3D image (b) reconstruction29. Databases of expert delineated 3D images allow accurate and fully automated example-based tissue segmentation (c) into age-consistent classes, together with parcellation into lobe regions that can be tracked over long periods of development. Creation of surface representations at high resolution (d) allows mapping of subtle surface properties, such as local area and curvature to accurately quantify cortical development. Mixed-effects modeling of repeated measures (e) from fetuses allows fitting of non-linear growth models and analysis of complex relative growth trajectories between local and global regions (f).

Back to article page