MICCAI Grand Challenge on Multi-domain Cross-time-point Infant Cerebellum MRI Segmentation 2022

Cerebellum is a rapidly developing and critical brain structure during the early postnatal stages. Cerebellar involvement has been implicated in the parthenogenesis of many neurodevelopmental disorders, e.g., autism, attention-deficit/hyperactivity disorder, and schizophrenia. Therefore, accurate segmentation of the cerebellum into white matter (WM), gray matter (GM), and cerebrospinal fluid (CSF) is essential to better understand cerebellar structure and function and assist in the diagnosis and treatment of neurodevelopmental disorders. Compared with adult cerebellum, there are very few works proposed for infant cerebellum segmentation. Infant cerebellum MRIs exhibit extremely low tissue contrast and severe partial volume effects in magnetic resonance
imaging (MRI), posing a huge challenge for manual and automated segmentation of the cerebellum. First, due to the low tissue contrast, the manual annotation is extremely challenging, especially for younger infant subjects (e.g., 6 months). Second, the collaborative use of multi-domain infant images (acquired from different imaging sites) makes the segmentation task more difficult. Third, there are often anatomical errors in the segmentation results. Therefore, by taking advantage of accurate manual labels from 24-month-old subjects, the aim of this challenge is to promote accurate segmentation algorithms on infant cerebellum MRIs from multiple domains.