Automatic Segmentation of Eight Tissue Classes in Neonatal Brain MRI

Abstract

PurposeVolumetric measurements of neonatal brain tissues may be used as a biomarker for later neurodevelopmental outcome. We propose an automatic method for probabilistic brain segmentation in neonatal MRIs.Materials and MethodsIn an IRB-approved study axial T1- and T2-weighted MR images were acquired at term-equivalent age for a preterm cohort of 108 neonates. A method for automatic probabilistic segmentation of the images into eight cerebral tissue classes was developed: cortical and central grey matter, unmyelinated and myelinated white matter, cerebrospinal fluid in the ventricles and in the extra cerebral space, brainstem and cerebellum. Segmentation is based on supervised pixel classification using intensity values and spatial positions of the image voxels. The method was trained and evaluated using leave-one-out experiments on seven images, for which an expert had set a reference standard manually. Subsequently, the method was applied to the remaining 101 scans, and the resulting segmentations were evaluated visually by three experts. Finally, volumes of the eight segmented tissue classes were determined for each patient.ResultsThe Dice similarity coefficients of the segmented tissue classes, except myelinated white matter, ranged from 0.75 to 0.92. Myelinated white matter was difficult to segment and the achieved Dice coefficient was 0.47. Visual analysis of the results demonstrated accurate segmentations of the eight tissue classes. The probabilistic segmentation method produced volumes that compared favorably with the reference standard. ConclusionThe proposed method provides accurate segmentation of neonatal brain MR images into all given tissue classes, except myelinated white matter. This is the one of the first methods that distinguishes cerebrospinal fluid in the ventricles from cerebrospinal fluid in the extracerebral space. This method might be helpful in predicting neurodevelopmental outcome and useful for evaluating neuroprotective clinical trials in neonates.

Neonatal brain MRI segmentation: A review

“…So, a majority of the algorithms perform only a limited numeric evaluation of their results. For instance, Anbeek et al [5] performed manual annotation and subsequent quantitative validation only on seven of the total 108 neonates in the study. Some approaches also use selective slices per subject for the validation.…”

“…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. In instances where follow-up papers have been published by the participating groups, namely Gui et al [43], Egekher et al [33] and Anbeek et al [5] corresponding to methods E, F and G of the challenge, the extended studies are described.…”

“…The work of Anbeek et al [5] involved the creation of an average brain image that was used as a common coordinate system for transferring all subject images. For this, an iterative registration and summation of all subject scans was done.…”

“…Quantitative validation metrics Prastawa et al [78] Dice similarity coefficient, Cohen's kappa coefficient [119] Dice coefficient, false positive, false negative, boundary segmentation error, Surface reconstruction error Song [100] Dice coefficient Weisenfeld and Warfield [115] Dice coefficient [120] Dice coefficient, correct estimation index, over estimation index, under estimation index Shi et al [91,92,93] Dice coefficient [54] Dice coefficient, surface reconstruction error Melbourne et al [64] Dice coefficient Makropoulos et al [59] Dice coefficient, 95th-percentile Hausdorff distance Gui et al [43] Dice coefficient Anbeek et al [5] Dice coefficient, sensitivity, specificity Gousias et al [40] Dice coefficient [33] Dice coefficient, 95th-percentile Hausdorff distance Cardoso et al [17] Dice coefficient Wang et al [108] Dice coefficient, surface distance error, Hausdorff distance Wang et al [109] Dice coefficient, surface distance error, landmark curve distance error, local and global gyrification index Makropoulos et al [60] Dice coefficient works have included both neonatal and older infant/pediatric subjects in their studies; thus, such works also need to be included for the sake of completeness of the review. For instance, the 'LABEL' approach i.e.…”

Patent Ductus Arteriosus and Brain Volume

“…We used the algorithm of Anbeek et al 27 The proposed segmentation algorithm is based on supervised pixel classification. T1-and T2-weighted images provided intensity information, and voxel position (ie, x-, y-, and z-coordinates in the coordinate system of the average brain) gave spatial characteristics.…”