Super-resolution Reconstruction of Fetal Craniofacial MRI

描述(由申请人提供)：该项目的总体目标是显著提高胎儿MRI诊断、分析和预测颅面发育障碍的能力。这些疾病存在于所有人类出生缺陷中的四分之三，大约每500名活产中就有一名受到影响。颅面部疾病可能会导致严重的长期问题，影响生活质量，并导致更高的疾病风险。提高对早期颅面发育和发育障碍的认识有助于更好的预后和治疗，更好的妊娠管理，更好的新生儿护理，以及更好的长期结果。非侵入性医学成像技术是获取胎儿宫内信息的主要工具。产前超声检查通常在妊娠中期进行，被认为是主要的诊断工具。然而，超声对唇腭裂、半面部巨大症、小颌症等复杂头面部疾病的诊断准确率极低。另一方面，磁共振成像(MRI)已经成为超声检查的一个很好的补充，可以对这些复杂的疾病进行准确的诊断和分析。尽管如此，胎儿MRI仅限于通过小胎儿器官提供的小信号进行二维采集，以及中断准确三维分析所需的空间编码的间歇性胎儿运动。近年来发展了一种新的图像处理技术，用于重建胎儿大脑的高分辨率三维MRI。这项技术已经显著改善了胎儿的神经成像。然而，这项技术不能直接或简单地适用于颅面结构的胎儿MRI；软组织、液体、颅颌面骨和半骨结构的非刚性和局部运动给体积重建带来了巨大的挑战。这项建议的具体目标是开发新的头面部结构中软组织、液体和骨骼的模型，并基于这些模型进行局部运动估计。这也将通过使用先进的图像正则化技术来考虑，而不需要显式的亚像素运动估计。这项建议的具体目标还包括重建高分辨率胎儿颅面MRI，并根据各种疾病对其进行分类。收集的数据将与波士顿儿童医院的放射科医生以及FaceBase财团上更多的颅面专家社区共享。

DESCRIPTION (provided by applicant): The overall objective of this project is to dramatically improve the capability of fetal MRI for diagnosis, analysis, and prognosis of craniofacial developmental disorders. These disorders present in three quarters of all human birth defects and affect approximately one in every 500 live births. Craniofacial disorders may cause serious long-term problems that affect the quality of life and result in higher disease risk. Improved knowledge of early craniofacial development and developmental disorders helps for better prognosis and treatment, better management of pregnancy, improved neonatal care, and better long-term outcomes. Non-invasive medical imaging techniques are the main tools to acquire information about the fetus in-utero. Prenatal sonography is routinely performed in the second trimester of pregnancy, and is considered to be the primary diagnostic tool. Nevertheless the diagnostic accuracy of sonography for complex craniofacial diseases such as cleft lip and cleft palate, hemifacial microsomia, micrognathia, etc. is extremely low. On the other hand magnetic resonance imaging (MRI) has become an excellent complement to sonography for accurate diagnosis and analysis of such complex diseases. Nonetheless, fetal MRI is limited to two-dimensional acquisitions by small signal available from the small fetal organs, and by intermittent fetal motion that disrupts spatial encoding necessary for accurate three-dimensional analysis. Novel image processing technology has recently been developed for the reconstruction of high-resolution three-dimensional MRI of the fetal brain. This technology has led to significant improvements in fetal neuroimaging. However, this technology cannot be directly or simply adapted to fetal MRI of craniofacial structures; the non-rigid and local movement of soft tissue, fluid, and craniomaxillofacial bones and semi-bony structures pose significant challenges in volume reconstruction. The specific aim of this proposal is the development of novel models of soft tissue, fluid, and bone in craniofacial structures and local motion estimation based on these models. This will also be considered through the use of advanced image regularization techniques without explicit sub-voxel motion estimation. The specific aims in this proposal also involve the reconstruction of high-resolution fetal craniofacial MRI and their classification based on various types of disorders. The collected data will be shared with radiologists at Children's Hospital Boston as well as with the greater community of craniofacial experts on FaceBase consortium.