Gervais Julien, Périé Delphine, Aubin Carl-Éric
a Department of Mechanical Engineering , Ecole Polytechnique , Montréal , Québec , Canada.
Comput Methods Biomech Biomed Engin. 2014;17(12):1383-90. doi: 10.1080/10255842.2012.748756. Epub 2013 Feb 22.
There is a lack of early biomarkers of intervertebral disc (IVD) degeneration. Thus, the authors developed the analysis of magnetic resonance signal intensity distribution (AMRSID) method to analyse the 3D distribution of the T2-weighted MR signal intensity within the IVD using normalised histograms, weighted centres and volume ratios. The objective was to assess the sensitivity of the AMRSID method to the segmentation process and data normalisation. Repetition of the semi-automatic segmentation by the same operator did not influence the quality of the contour or our new MR distribution parameters whereas the skills of the operator influenced only the MR distribution parameters, and the instructions given prior to the segmentation influenced both the quality of the contour and the MR distribution parameters. Bone normalisation produces an index that jointly highlights IVD and bone health, whereas cerebrospinal fluid normalisation only suppresses the effect of the acquisition gain. This robust AMRSID method has the potential to improve the diagnostic with earlier biomarkers and the prognosis of evolution.
目前缺乏椎间盘（IVD）退变的早期生物标志物。因此，作者开发了磁共振信号强度分布分析（AMRSID）方法，使用归一化直方图、加权中心和体积比来分析IVD内T2加权磁共振信号强度的三维分布。目的是评估AMRSID方法对分割过程和数据归一化的敏感性。同一操作员重复半自动分割不会影响轮廓质量或我们新的磁共振分布参数，而操作员的技能仅影响磁共振分布参数，分割前给出的指令则会影响轮廓质量和磁共振分布参数。骨归一化产生一个共同突出IVD和骨骼健康的指标，而脑脊液归一化仅抑制采集增益的影响。这种稳健的AMRSID方法有可能通过早期生物标志物改善诊断和疾病进展的预后。
