CRII: SCH: Analysis of Population-Based Image Metamorphosis

CRII：SCH：基于人群的图像变形分析

• 批准号: 1755970
• 负责人: Sheng Li
• 金额: $ 17.48万
• 依托单位: University of Georgia Research Foundation Inc
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-08-15 至 2022-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1755970&HistoricalAwards=false
• 关键词: CRII; SCH; Analysis; Population; Based

Uncovering brain development and disease progression can ultimately advance our understanding of brain disorders, such as mental illnesses and brain cancer. However, the analysis of early brain growth and pathological anatomy changes in brain scans needs image metamorphosis that can jointly capture structural changes occurring in a developing brain and appearance changes introduced by brain myelination and tumor growth. This project aims to develop statistical and computational models to summarize these metamorphic changes from image pairs, to sequences, and to populations. By enriching our knowledge of the healthy brain, psychopathologies, and brain pathologies, this research has the potential to provide clues in devising treatment strategies for brain disorders.The focus of this project is the development of population-based image metamorphosis that accounts for brain structure and image appearance changes simultaneously. The essential component of the developed model is an advanced image-to-image metamorphosis that allows establishing brain mappings between image pairs. This project develops a novel metamorphic regression model to handle image sequences and estimate age-related brain changes in image time-series. The population analysis component leverages the regression model and statistical tools for classification and assessment of typical and atypical brain changes. The primary applications of the developed image metamorphosis models are studies of early brain development and brain tumor growth. The resulting models could be applied to understand the etiology of developmental disorders and psychiatric conditions, such as autism, schizophrenia, and bipolar disease. They could also monitor brain tumor growth through predicting tumor evolution and classifying tumor type and grade.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

揭示大脑发育和疾病进展最终可以增进我们对精神疾病和脑癌等大脑疾病的理解。然而，分析早期大脑生长和脑扫描中的病理解剖变化需要图像变形，可以共同捕获发育中大脑中发生的结构变化以及脑髓鞘形成和肿瘤生长引起的外观变化。该项目旨在开发统计和计算模型来总结从图像对到序列和群体的这些变质变化。通过丰富我们对健康大脑、精神病理学和脑病理学的了解，这项研究有可能为设计脑部疾病的治疗策略提供线索。该项目的重点是开发基于人群的图像变形，以解释大脑结构和图像外观同时变化。所开发模型的基本组成部分是先进的图像到图像的变形，允许在图像对之间建立大脑映射。该项目开发了一种新颖的变形回归模型来处理图像序列并估计图像时间序列中与年龄相关的大脑变化。群体分析组件利用回归模型和统计工具对典型和非典型大脑变化进行分类和评估。所开发的图像变形模型的主要应用是早期大脑发育和脑肿瘤生长的研究。由此产生的模型可用于了解发育障碍和精神疾病（例如自闭症、精神分裂症和双相情感障碍）的病因。他们还可以通过预测肿瘤演变以及对肿瘤类型和等级进行分类来监测脑肿瘤的生长。该奖项反映了 NSF 的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

ASC-Net: Adversarial-Based Selective Network for Unsupervised Anomaly Segmentation

ASC-Net：用于无监督异常分割的对抗性选择性网络

• DOI: 10.1007/978-3-030-87240-3_23
• 发表时间: 2021-09
• 期刊: International Conference on Medical Image Computing and Computer-Assisted Intervention
• 作者: Dey, Raunak;Hong, Yi
• 通讯作者: Hong, Yi

Hybrid Cascaded Neural Network for Liver Lesion Segmentation

用于肝脏病变分割的混合级联神经网络

• DOI: 10.1109/isbi45749.2020.9098656
• 发表时间: 2019-09-11
• 期刊: 2020 IEEE 17th International Symposium on Biomedical Imaging (ISBI)
• 作者: Raunak Dey;Yi Hong
• 通讯作者: Yi Hong

Diffeomorphic Image Registration using Lipschitz Continuous Residual Networks

使用 Lipschitz 连续残差网络进行微分同胚图像配准

• 发表时间: 2022-10
• 期刊: Proceedings of Machine Learning Research
• 作者: Joshi, Ankita;Hong, Yi
• 通讯作者: Hong, Yi

Efficient Diffeomorphic Image Registration using Multi-Scale Dual-Phased Learning

使用多尺度双阶段学习的高效微分同胚图像配准

• DOI: 10.1109/isbi52829.2022.9761693
• 发表时间: 2022-08
• 期刊: IEEE 19th International Symposium on Biomedical Imaging (ISBI
• 作者: Joshi, Ankita;Hong, Yi
• 通讯作者: Hong, Yi