Mobile Three-Dimensional Screening for Cranial Malformations

摘要 婴儿头部畸形的延迟识别正在造成不必要的医疗并发症和社会 成本。早期发现的一个关键挑战是儿科办公室缺乏可供执行任务的工具 健康儿童访视期间的定量头型评估。诊断的延迟限制了早期诊断的机会， 侵入性较小且有效的治疗选择。在这个快速通道SBIR项目中，PediaMetrix Inc. 与儿科医院和提供商一起开发和评估SoftSpotTM，这是第一个 托儿点婴儿颅骨畸形的三维数据采集与分析 婴儿时期的头部畸形可以是融合性的(即颅骨融合)或非融合性的(例如 变形的斜头畸形和短头畸形或DPB)。这两种情况都需要立即注意 并从早期治疗中受益，以避免长期的健康并发症。DPB的患病率增加 近年来戏剧性地从5%上升到大约20%-30%，导致这种情况被称为儿科 流行病。颅缝早闭不太常见，每2000名儿童中就有一名受到影响。改善对儿童的早期管理 为了防止更复杂的治疗和相关的疾病，有必要监测 婴儿头在托儿点的生长。 为了解决这一未得到满足的临床需求，我们将开发和评估一种移动数字工具，该工具将使 儿科医生捕捉和分析每个婴儿的3D扫描，以早期诊断和处理颅脑损伤 畸形。在这个项目的第一阶段，我们将开发一种新的技术来快速捕获和分析 几秒钟内就能得到头盖骨的3D数据。我们将使用机器学习方法自动计算 头部形状参数，包括每个儿童期间常规测量的头部周长 访问，但目前使用的是过时和不可靠的量尺。我们的技术将专为 在健康儿童探视期间对所有婴儿进行一般头颅评估。在第二阶段，我们将开发3D的方法 从智能手机重建和分析整个头盖骨。我们还将培训深度学习模型以 对颅缝融合和其他颅脑疾病进行分类，并进行临床评估和使用可行性 学习。 PediaMetrix的总体使命是为儿科健康提供准确的决策支持工具 护理点。这将通过机器学习和定量成像算法来实现 与智能手机技术进步相结合，将被包装为移动数字健康解决方案 儿科保健提供者可随时随地获得。SoftSpot3DTM的成功演示将 导致未经治疗的儿童颅脑疾病的数量显著减少 降低相关的医疗成本和社交焦虑。

ABSTRACT Delayed identification of infant head malformation is causing unnecessary medical complications and societal costs. A critical challenge in the early detection is the absence of tools available to pediatric offices to perform quantitative head shape assessment during well-child visits. Delays in diagnosis limit the opportunity for early, less invasive and effective treatment options. In this Fast-track SBIR project, PediaMetrix Inc. has joined forces with pediatric hospitals and providers to develop and evaluate SoftSpotTM, which is the first mobile digital tool for 3D data collection and analysis of infant cranial malformations at the point-of-care. Head malformations during infancy can be synostotic (i.e., craniosynostosis) or nonsynostotic (such as deformational plagiocephaly and brachycephaly or DPB). Both types of conditions require immediate attention and benefit from early treatment to avoid long-term health complications. The prevalence of DPB increased dramatically in recent years, from 5% to approximately 20%-30%, causing the condition to be called a pediatric epidemic. Craniosynostosis is less common affecting 1 in 2,000 children. To improve the early management of these conditions and to prevent more complex treatment and associated morbidities, it is essential to monitor the growth of the infant head at the point-of-care. To address this unmet clinical need, we will develop and evaluate a mobile digital tool that will enable pediatricians to capture and analyze 3D scans of every infant for the early diagnosis and management of cranial malformations. In the Phase I of this project, we will develop a novel technology to rapidly capture and analyze 3D data of the top of cranium in just seconds. We will use machine learning methods to automatically compute the head shape parameters, including the head circumference which is routinely performed during every child visit, but currently with an outdated and unreliable measuring tape. Our technology will be designed for the general cranial evaluation of all infants during well-child visits. In Phase II, we will develop methods for the 3D reconstruction and analysis of the full cranium from a smartphone. We will also train deep learning models to classify types of craniosynostosis and other cranial conditions and conduct clinical evaluation and user-feasibility studies. The overall mission of PediaMetrix is to provide accurate decision support tools for pediatric health at the point-of-care. This will be achieved through machine learning and quantitative imaging algorithms that in combination with smartphone technological advances will be packaged as mobile digital health solutions accessible to pediatric health providers at any time and location. Successful demonstration of SoftSpot3DTM will lead to a significant reduction of the number of children left with untreated cranial conditions in addition to lowering the associated healthcare costs and social anxiety.