Evaluation of automatic neonatal brain segmentation algorithms: The NeoBrainS12 challenge

3D MR ventricle segmentation in pre-term infants with post-hemorrhagic ventricle dilatation (PHVD) using multi-phase geodesic level-sets

“…Wang et al (2014a) integrated this probabilistic atlas into a patch-driven level-set framework for more accurate segmentation of neonatal MR images. Three different image sets of preterm infants provided in the NeoBrainS12 study (http://neobrains12.isi.uu.nl) (Isgum et al, 2015) were set up to allow a comparison of 8 brain tissue segmentation methods. The results demonstrated that the participating methods were able to segment all tissue classes well, except myelinated white matter.…”

“…Most of previous work (Gui et al, 2012;Wang et al, 2011Wang et al, , 2014aShi et al, 2010a;Isgum et al, 2015) segmented neonatal brain MR images into different brain tissues, such as white matter (WM), gray matter (GM), and CSF, but did not focus on the ventricular system. Moreover, these methods were only validated on high-quality healthy neonate images generated for research purposes only while patients were sedated.…”

Neonatal brain MRI segmentation: A review

“…Apart from the web-based challenge, an on-site challenge was also held. The quantitative segmentation results of both the web-based and on-site events can be found in Isgum et al [48]. The works of Makropoulos et al [59] and Melbourne et al [64], corresponding to methods A and C 5 of the NeobrainS12 challenge, are described in Section 4.2.2.…”

“…The existing algorithms have limitations and inaccuracies that need to be addressed by further research. Isgum et al [48] have outlined some possible directions for future work, based on the NeoBrainS12 challenge; this includes extending the segmentation algorithms to involve patients with diseases, incorporating images obtained at various gestational ages in neonatal phase and including images obtained by different MR scanners and protocols (Isgum, 2015). The research gaps in the available literature are presented below.…”

“…Even if they do, the accuracy of segmentation of MWM is well below that of other tissues, as evident from its lower Dice values and larger outliers ( [116]; Anbeek, 2013; Egekher, 2013). Isgum et al [48], based on their evaluation of the segmentation approaches that participated in the NeoBrainS12 challenge, have also stated that MWM was poorly segmented and suffered from large mean surface distance errors. The low Dice values can be explained to some extent by the relatively low volume of myelin in the neonatal brain; this makes the Dice metric sensitive to even small differences between the segmented and reference volumes (Anbeek, 2013).…”