张璐坚

【摘 要】为了解决传统磁共振成像技术一次扫描只能得到一种加权像，且成像时间过长，扫描过程产生噪声等问题，提出了“磁共振指纹”成像方法。人脑有多种组织组成，每种组织的磁共振扫描都有自己不同特征信號该信号被称为“指纹”。“磁共振指纹”成像方法一次扫描不仅可同时得到 T1图，T2图 及 质子密度图像;而且大大缩短了扫描时间。该新技术在成像算法上也与传统磁共振算法完全不同。基于字典的匹配方法，首先建立指纹“字典”;之后将每个体素的检测信号与“字典”中的所有信号进行逐条匹配;然后确定是该信号的参数组合;最终得到多种参数的参数图像。本实验采用大脑数据进行模拟采集和匹配过程。结果表明，该方法可以得到各种组织参数图像，且确实会大大降低磁共振扫描所需的时间。

【关键词】磁共振指纹;字典;参数图像;大脑

中图分类号： R318 文献标识码： A 文章编号： 2095-2457（2019）08-0001-003

DOI：10.19694/j.cnki.issn2095-2457.2019.08.001

【Abstract】There are some defects in the traditional magnetic resonance imaging methods. For example， only one weighted image can be obtained in one scan， and the scanning time is too long and the scanning process produces noise. Therefore， the "magnetic resonance fingerprint" imaging method was proposed. The human brain is made up of a variety of tissues， each tissue has its own characteristic signal during MR scans that is called a" fingerprinting". MRf imaging method can obtain T1 map， T2 map and proton density map at the same time. It also shortens the scanning time. The new technique is also different from the traditional magnetic resonance algorithm. Firstly， a fingerprint "dictionary" is established based on the dictionary matching method. Then the detection signal of each individual element is matched with all signals in the "dictionary" one by one. Then the parameter combination of the signal is determined. Finally， the parametric images of various parameters are obtained. This experiment uses brain data to simulate collection and matching process. The results show that this method can obtain various tissue parameter images and reduce the time required for MR scanning.

【Key words】Magnetic resonance fingerprinting; Dictionary; Parameter map; Brain

0 引言

核磁共振成像技术（Magnetic Resonance Imaging， MRI）近些年在临床上作为重要的临床诊断方法之一被广泛使用[1]，但是核磁共振成像技术仍需要解决较多问题。问题之一就是核磁共振成像是定性的检测技术。它能通过不同物质表现的信号强度不同，成像时通过相对亮度的不同，虽然能够准确检测出每种组织的特性， 但它产生的信号受到许多其它外部条件的影响比如扫描的类型，探测器等，而且对组织内部信号强度不同的区域不敏感。问题二就是它进行多参数加权成像，如T1加权图T2加权图等[2]，对于图像的灰度信息却不能给出其生理特质。……