一种基于磁共振图像的心肌梗死体积定量算法

Quantification algorithm of myocardial infarction volume based on MRI

目的心脏磁共振（CMR）钆对比剂延迟强化（LGE）图像在诊断冠心病心肌梗死（MI）方面的应用价值已得到公认。MI体积的计算对MI的诊断和治疗有重要意义。笔者在对比了多种定量MI体积计算方法的基础上提出了一种新的基于直方图分布的MI体积计算方法,并以病理切片所得MI体积作为标准进行评价。方法选3个月中华小型猪6只,雌雄不限,体质量20.8~24.1 kg,平均体质量21.5 kg。结扎冠状动脉,用于制备MI模型,1个月后扫描磁共振图像;随即取心脏制备切片并染色。运用数学软件MATLAB（2013b）编程对LGE心脏短轴图像行不同信号强度（SI）阈值分割,包括5s（标准差）阈值法及新算法,提取梗死部分信号,进而计算MI体积,并与染色切片得到的MI体积进行对比。结果病理切片得到的MI体积为（12.24±4.34）cm3。5s阈值法及新算法得到的MI体积分别为（12.54±5.38）cm3、（12.40±4.46）cm3。病理切片所得MI体积与5s阈值法及新算法得到的MI体积之间的相关系数分别为0.967、0.987。结论基于直方图分布的新算法在中华小型猪MI模型MI体积计算中与病理切片所得体积相关性好,且无统计学差异。基于直方图分布的新算法能准确计算MI体积。

Objective To study the new calculation algorithm of myocardial infarction（MI） volume based on histogram by comparison of several quantitative algorithms, and evaluate the standard of pathological sections in MI volume of Chinese miniature swine. Methods Six Chinese miniature swines aged 3-month were enrolled, which were performed coronary artery ligation（CAL） to simulate MI model. MRI scan was performed after 1-month, and the heart was removed and analyzed by histochemical staining. The MATLAB（2013b） software was used in threshold segmentation for different signal intensity（SI） of late gadolinium enhancement（LGE） cardiac short axis images, which included 5s semi automatic quantitative algorithms and new algorithm. The signal of infarction was extracted, the MI volume was calculated and compared with the MI volume obtained by histochemical staining. Results The volume of MI obtained from pathological sections was（12.24 ± 4.34） cm3, and that from 5s semi automatic quantitative algorithms and new algorithm was（12.54 ± 5.38） cm3and（12.40 ± 4.46） cm3, respectively. In addition, the correlation coefficient between the pathological sections and 5s semi automatic quantitative algorithms was 0.967, and correlation coefficient between the pathological sections and new algorithm was 0.987. Conclusion It is demonstrated that the correlation coefficient of MI volume between the pathological sections and new algorithm based on histogram is 0.987, and there is no statistical difference, the new algorithm could be considered as the reliable method for LGE quantification.