Statistical spatiotemporal model of anatomical and pathological structures and computer-aided diagnostic imaging

为了描述医学图像中解剖结构之间的空间关系，我们提出了一个统计空间模型。提出了一种基于该模型的多结构同时分割算法。随后提出了一种新的条件统计模型，该模型可以放宽条件，并根据训练数据条件的误差模型自动优化条件。进一步的贡献是开发了基于层的统计模型的多层分割算法和基于统计图像数据库的健康和病理结构的同时识别算法。

We proposed a statistical spatial model so as to describe the spatial relationship between anatomical structures in a medical imagery. A simultaneous segmentation algorithm of multiple structures based on the model was developed. Subsequently a novel conditional statistical model was proposed, in which a condition can be relaxed and it is automatically optimized based on an error model of the condition of training data. Further contributions are developments of a multi-layer segmentation algorithm with a statistical model of layers and simultaneous recognition algorithm of healthy and pathological structures based on an statistical image database.