A novel hybrid atlas-free hierarchical graph-based segmentation of newborn brain MRI using wavelet filter banks

Feasibility of Automatic Seed Generation Applied to Cardiac MRI Image Analysis

We present a method of using interactive image segmentation algorithms to reduce specific image segmentation problems to the task of finding small sets of pixels identifying the regions of interest. To this end, we empirically show the feasibility of automatically generating seeds for GrowCut, a popular interactive image segmentation algorithm. The principal contribution of our paper is the proposal of a method for automating the seed generation method for the task of whole-heart segmentation of MRI scans, which achieves competitive unsupervised results (0.76 Dice on the MMWHS dataset). Moreover, we show that segmentation performance is robust to seeds with imperfect precision, suggesting that GrowCut-like algorithms can be applied to medical imaging tasks with little modeling effort.

Impacts of skull stripping on construction of three-dimensional T1-weighted imaging-based brain structural network in full-term neonates

BACKGROUND

Skull stripping remains a challenge for neonatal brain MR image analysis. However, little is known about the accuracy of how skull stripping affects the neonatal brain tissue segmentation and subsequent network construction. This paper therefore aimed to clarify this issue by comparing two automatic (FMRIB Software Library's Brain Extraction Tool, BET; Infant Brain Extraction and Analysis Toolbox, iBEAT) and a semiautomatic (iBEAT with manual correction) processes in constructing 3D T1-weighted imaging (T1WI)-based brain structural network.

METHODS

Twenty-two full-term neonates (gestational age, 37-42 weeks; boys/girls, 13/9) without abnormalities on MRI who underwent brain 3D T1WI were retrospectively recruited. Two automatic (BET and iBEAT) and a semiautomatic preprocessing (iBEAT with manual correction) workflows were separately used to perform the skull stripping. Brain tissue segmentation and volume calculation were performed by a Johns Hopkins atlas-based method. Sixty-four gray matter regions were selected as nodes; volume covariance network and its properties (clustering coefficient, C_p; characteristic path length, L_p; local efficiency, E_local; global efficiency, E_global) were calculated by GRETNA. Analysis of variance (ANOVA) was used to compare the differences in the calculated volume between three workflows.

RESULTS

There were significant differences in volumes of 50 brain regions between the three workflows (P < 0.05). Three neonatal brain structural networks presented small-world topology. The semiautomatic workflow showed remarkably decreased C_p, increased L_p, decreased E_local, and decreased E_global, in contrast to the two automatic ones.

CONCLUSIONS

Imperfect skull stripping indeed affected the accuracy of brain structural network in full-term neonates.