A Novel Approach for Manual Segmentation of the Amygdala and Hippocampus in Neonate MRI

• Published in: Frontiers in neuroscience 2019-09, Vol.13, p.1025-1025
• Main Authors: Hashempour, Niloofar, Tuulari, Jetro J., Merisaari, Harri, Lidauer, Kristian, Luukkonen, Iiris, Saunavaara, Jani, Parkkola, Riitta, Lähdesmäki, Tuire, Lehtola, Satu J., Keskinen, Maria, Lewis, John D., Scheinin, Noora M., Karlsson, Linnea, Karlsson, Hasse
• Format: Article
• Language: English
• Subjects: Adults; Amygdala; automated segmentation; Automation; Babies; Brain; Child & adolescent psychiatry; Hippocampus; Image processing; Infants; Magnetic resonance imaging; manual segmentation; Neonates; Neuroscience; Newborn babies; NMR; Nuclear magnetic resonance; Segmentation; Software; Studies
• ISSN: 1662-453X; 1662-4548; 1662-453X
• DOI: 10.3389/fnins.2019.01025

The gross anatomy of the infant brain at term is fairly similar to the adult brain, but structures are immature, and the brain undergoes rapid growth during the first two years of life. Neonate MR images have different contrasts as compared to adult images, and automated segmentation of brain MRI can thus be considered challenging as less software are available. Despite this, most anatomical regions are identifiable and thus amenable to manual segmentation. In the current study, we developed a protocol for segmenting the amygdala and hippocampus in T2-weighted neonatal MR images. The participants are 31 healthy infants between 2 and 5 weeks of age. Intra-rater reliability was measured in 12 randomly selected MR images where six MR images were segmented at 1-month interval between the delineations, and another six MR images at 6-months interval. The protocol was also tested by two independent raters in 20 randomly selected T2-weighted images, and finally with T1 images. Intra-class correlation and dice coefficient for intra-rater, inter-rater, and T1 vs. T2 comparisons were conducted. Moreover, 10 T2-weighted manually segmented images were compared to the same 10 T2-weighted automated segmentations obtained from the iBEAT toolbox. The intra-rater reliability was high ICC ≥ 0.91, DSC ≥ 0.89, the inter-rater reliabilities were satisfactory ICC ≥ 0.90, DSC ≥ 0.75 for hippocampus and DSC ≥ 0.52 for amygdalae. Segmentations for T1 vs. T2-weighted images showed high consistency ICC ≥ 0.90, DSC ≥ 0.74. The manual segmentation and iBEAT segmentations did not agree DSC ≥ 0.39. In conclusion, there is a clear need for improving and developing procedures for automated segmentation of infant brain MR images, and we present one such protocol.