Manual segmentation of the dentate nucleus of the cerebellum

• Grant number: 25/24312-9
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: December 01, 2025
• End date: November 30, 2026
• Field of knowledge: Health Sciences - Medicine - Medical Radiology
• Principal Investigator: Thiago Junqueira Ribeiro de Rezende
• Grantee: Giovanna Lobo Costa
• Host Institution: Faculdade de Medicina de Ribeirão Preto (FMRP). Universidade de São Paulo (USP). Ribeirão Preto , SP, Brazil
• Associated research grant: 25/00186-4 - Development of Advanced Computational Tools for In Vivo Analysis of Dorsal Root Ganglia and Spinal Cord Integrity in Sensory Neuronopathy, AP.JP

Abstract

Hereditary ataxias (SCA) are a group of diseases that share cerebellar ataxia as a common clinical finding; however, non-motor symptoms and cognitive deficits may also be present. Pathologically, SCAs are characterized by severe neuronal loss in the cerebellum and brainstem, especially in the dentate nuclei of the cerebellum. Unfortunately, there are few imaging studies that aim to assess this structure. In this sense, there are no automatic segmentation techniques that can identify such structures. Machine learning algorithms, called convolutional neural networks, have been successfully applied in medical images to automatically segment some brain structures. However, to make this possible, the creation of a training database is necessary. Therefore, manually segmented images are used to create this database and teach the algorithm. Nevertheless, the manual segmentation process is not simple and is often time-consuming.Thus, the main objective of this project is to create a database of dentate nucleus images to supply the training database that will be used in the computational tool created by us to segment these structures. This project is extremely important, as it is a fundamental step of this young researcher project, as it will save time for the PhD student, allowing them to focus on developing the learning algorithm and program. Furthermore, the undergraduate research student will gain important knowledge in radiological neuroanatomy and image processing.Regarding the clinical impact of this project, the creation of the segmented image database will contribute to the development of the computational tool that will automatically segment the dentate nucleus, adding important information to the pathophysiology of SCAs. This tool can also be applied to various other diseases, such as sensory neuronopathies and motor neuron diseases. Therapeutic implications may arise from the results obtained, since a better understanding of the disease mechanism allows for a more adequate structuring of the therapies applied.