地中海贫血是一类严重威胁人类健康的遗传性血液病，在我国长江以南为高发区，治疗中反复输血使铁在患者体内实质器官沉积并造成严重病变，准确地测量肝脏铁沉积对于指导与监控地贫患者的除铁治疗具有重要的临床意义。 基于磁共振成像T2*参数的肝脏铁沉积测量方法具有速度快易推广等优点，但仍存在一定局限：肝铁沉积严重时信号T2*值较小，信号衰减过快导致T2*不能准确测得；反映肝铁空间分布的R2*（1/T2*）图受噪声影响严重尚不能准确计算；全肝T2*测量可重复性更好但目前尚无简便可靠的肝实质提取方法。 本项目拟采用超短回波时间（TE）技术结合项目组最近提出的噪声补偿拟合模型以实现较小T2*的准确测量；拟结合非局部均值滤波与快速噪声补偿以实现R2*图的准确估计；并在此基础上提出用模糊聚类结合形态学操作实现全肝测量中肝实质的准确提取。 本项目研究将为临床地贫患者肝脏铁沉积测量提供更准确且重复性更好的方法。

Thalassemia is a kind of hereditary hematonosis and has a high risk in the South of Yangtze River in China. Beta-thalassemia major results in a regular blood transfusion requirement that leads to tissue iron overload. Accurate and reliable liver iron contentration (LIC) is playing an increasingly important role in managing the iron chelation therapy, and monitoring the response to treatment. Magnetic resonance T2* relaxometry method is becoming popular in LIC quantification as it is fast and easy to implement, and hence cost-saving. However, the current T2* relaxometry for LIC quantificaiton has several limits. First, accurate measurement of very small T2* value (<1ms) is challenging due to the fast decay of measured signal. Second, the R2* (1/T2*) map, which reflects the spatial distribution of LIC, cannot be efficiently and accurately caculated due to the impact from noise. Third, there is no reliable and easy-to-implement liver parenchyma extraction method for the whole liver T2* measurement which indicates the total LIC and has been shown to be more reproducibile. In this project, we propose corresponding methods to overcome the above limits of the current liver T2* method. For the accurate measuremnt of very small T2*, we propose to acquire signal of nearly zero echo time (TE) by using the ultra-short TE technique and to correct the noise's impact on T2* relaxometry by the noise-corrected models recently developed by our research group. To acquire the accurate R2* map, we plan to denoise the images before pixel-by-pixel curve fitting by using the nonlocal means algorithm which has an excellent detail-preseving ability while denoisng images, and to develop the fast implementation algorithm of the noise-corrected curve fitting. For the liver parenchyma extraction in the whole liver T2* measurement, we propose to perform fuzzy c-means clustering on the R2* map (already noise-corrected), which in theory is insensitve to the inhomogeneity field. After that, we plan to perfom the morphological operation to exclude the pixels near the parenchyma and nonparenchyma tissues which are likely affected by the partial volume effect. This project would provide improved, reliable and easy-to-implement T2* measurment methods for the quantificaiton of LIC, and improve the therapy management of beta-thalassemia major patients.