Development of CT image-based cranial bone markers of intra-cranial hypertension

项目总结 在儿童神经科和神经外科中，颅内高压(ICH)的识别是必不可少的。 评估。如果不治疗，可能会导致脑损伤、发育迟缓甚至死亡。然而，目前的非 脑出血的侵袭性评估是基于与其相关性较差的临床症状和影像。 调查结果具有很高的变数和主观性。因此，对脑出血的非侵入性评估是具有挑战性的，而且通常 没有定论。此外，侵入性颅内压监测仪可以帮助识别脑出血，但它们通常 避免，因为它们往往提供不确定的结果，并带有出血和感染的重要风险。 因此，脑出血在人群中仍未得到充分诊断。在这个项目中，我们的目标是创建非侵入性 方法利用头颅CT图像自动识别儿童脑出血的存在。 这个项目的主要假设是，颅内压升高会导致骨骼减少。 矿化，可以从CT图像中定量。这一假设得到了以下临床报道的支持 血压升高或脑生长受限的患者表现为颅骨更薄、密度更低，以及 出现异常增厚的颅骨的低颅压患者。 在这个项目中，我们将：(1)创建一个特定性别和年龄的颅骨规范统计模型 使用受试者的大型回溯性CT图像数据集研究出生至10年之间的厚度和密度 无颅脑病理；(2)量化脑出血患者的骨异常并创建机器学习 使用需要手术治疗的压力患者的回顾数据集来识别它们的模型 释放；以及(3)进行一项先导性研究，以识别和量化患有高血压的患者的局部血压升高迹象 颅缝融合导致的颅骨生长受限。这个项目完成后，我们将拥有 建立颅骨矿化的标准化参考标准，鉴定和量化骨矿化 与非物质文化遗产相关的变化，并创建了自动工具来识别它们。另外，如果我们的探索性研究 在颅缝融合成功的患者中，这将促使对这些患者是否存在脑出血进行更大规模的临床研究 患者可以通过及时地将其临床管理转化为显著改善 干预措施，特别是在年龄较大的儿童和复发患者中。

Project summary The identification of intracranial hypertension (ICH) is essential during pediatric neurological and neurosurgical evaluation. If untreated, it may lead to brain injury, developmental delay or even death. However, current non- invasive evaluation of ICH is based on clinical symptoms that have poor correlation with it, and imaging findings are highly variable and subjective. Hence, non-invasive evaluation of ICH is challenging and often inconclusive. Additionally, invasive intracranial pressure monitors can help identify ICH but they are usually avoided because they often offer inconclusive results and carry important risks of hemorrhage and infections. Consequently, ICH remains underdiagnosed in the population. In this project, we aim to create non-invasive methods to automatically identify the presence of ICH in pediatric patients using CT images of the head. The main hypothesis of this project is that an elevated intracranial pressure results into a decrease of bone mineralization, which can be quantified from CT images. This hypothesis is supported by clinical reports of patients with elevated pressure or brain growth constraints showing thinner and less dense cranial bones, and of patient with intracranial hypotension presenting an abnormally thick cranial bone. During this project, we will: (1) create a sex- and age-specific normative statistical model of cranial bone thickness and density between birth and 10 years using a large retrospective CT image dataset of subjects without cranial pathology; (2) quantify bone abnormalities in patients with ICH and create machine learning models to identify them using a retrospective dataset of patients who required surgical treatment for pressure release; and (3) perform a pilot study to identify and quantify local signs of elevated pressure in patients with cranial growth constraints caused by craniosynostosis. Upon completion of this project, we will have established normative references of cranial bone mineralization, identified and quantified bone mineralization changes associated with ICH, and created automatic tools to identify them. In addition, if our exploratory study in patients with craniosynostosis succeeds, it will prompt a larger clinical study of the presence of ICH in these patients that could translate in a substantial improvement of their clinical management through timely interventions, especially in older children and relapsing patients.