Do desenvolvimento normal à psicopatologia: o desafio da replicação em neuroimagem

• Ano de defesa: 2016
• Autor(a) principal: Zugman, Andre [UNIFESP]
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Paulo (UNIFESP)
• Palavras-chave em Português: Children; Development; Brain; Neurodesenvolvimento; Criança; cérebro
• Link de acesso: https://sucupira.capes.gov.br/sucupira/public/consultas/coleta/trabalhoConclusao/viewTrabalhoConclusao.jsf?popup=true&id_trabalho=3609118; http://repositorio.unifesp.br/handle/11600/46461

Introduction: It is critical to understanding the typical trajectory of brain development and how changes in brain measures during development are related to psychopathology and mental disorders. The replication of findings is being discussed intensely in the scientific literature in recent years. This study has the main objective of exploring neuroimaging findings related to age, sex and behavior and also check the reproducibility of these findings. We will explore the main objective in two specific studies: a) validate the findings of the sex and age effect on cortical thickness and cortical surface area. b) validation of brain-behavioral correlation, as assessed by CBCL (child behavior checklist) and cortical thickness. Method: 781 children were enrolled in the neuroimaging phase of the High Risk Cohort (HRC) study in two different cities. Of these 749 completed the examination of structural MRI. The images were processed using the FreeSurfer 5.1 software and statistical analyzes conducted in MATLAB software as appropriate for each specific objective. Results: The results described here suggest that there is a high potential for replicability for sex and age measures, while behavioral measures by CBCL changes seem more difficult to replicate. It is noteworthy that one of our analysis indicated a rate of 48% false positives, even with statistical methods of correction for multiple comparisons. Conclusion: The results described in this thesis indicate that the validation of findings in the sample is an effective way to control for false positives. We believe that the differences in the replicability between the findings relating to age, sex and behavioral changes is due to the nature of the measures. While age and sex are objective characteristics, the characteristics assessed by CBCL are subjective. Future neuroimaging studies should take into account not only the finding of location but also effect size measures, and include in its design a sample for confirmatory analyses.