基于纹理分析的骨质疏松计量诊断

摘要

Objective To detect and analyze bone tissue spicemen using texture image analysis.Methods Twenty 3-month female SPF Sparague-Dawley rats were randomly divided into 2 groups:the sham-operated ( SHAM) group and the ovariectomized ( OVX) group. At the experimental endpoint, the undecalicifiled sections of the proximal tibia were performed for bone histomorphometric analysis and texture analysis.After extracting from gray level co-occurrence matrix and run length matrx, the texture analysis was performed.Results The static parameters were significantly different between groups (P<0.05), indicating that the establishment of osteoporosis rat model was successful.In texture analysis, the area in SHAM group and OVX group was 282552.1 ±11896.32 and 133418.0 ±37934.50, respectively (P<0.05).The angular second moment in SHAM group and OVX group was 0.06 ±0.01 and 0.08 ±0.02, respectively (P<0.05).The contrast in SHAM group and OVX group was 2.92 ±0.52 and 1.99 ±0.67, respectively (P<0.05).The correlate in SHAM group and OVX group was 0.54 ±0.02 and 0.48 ±0.02, respectively (P<0.05).The entropy in SHAM group and OVX group was 1.54 ±0.08 and 1.33 ±0.11, respectively (P<0.05). The DifVarnc in SHAM group and OVX group was 1.54 ±0.08 and 1.06 ±0.36, respectively (P<0.05).The RLNonUni in SHAM group and OVX group was 30932.94 ±12790.38 and 11116.16 ±1714.33, respectively (P<0.05).The GLevNonU in SHAM group and OVX group was 11896.84 ±4737.95 and 6497.72 ±1823.29, respectively ( P<0.05) .Conclusion The bone tissue image in normal rats and the ovariectomized osteoporosis rats can be distinguished by texture analysis.%目的：用图像纹理分析方法对骨组织切片进行初步识别和分析。方法3月龄雌性SD大鼠20只，随机分为假手术组（ SHAM），去卵巢组（ OVX）。实验结束后，制作骨组织切片，并测量骨形态计量学静态参数。切片于显微镜下拍照，获得图像后进行纹理分析。采用游程长矩阵和灰度共生矩阵模型来提取纹理特征。结果 OVX组的静态参数与SHAM组相比有显著差异（ P ＜0．05），表明去卵巢所致的骨质疏松大鼠模型造模成功。纹理参数中，面积（ area ）：SHAM 组：（282552．1±11896．32）；OVX组：（133418．0±37934．50）（P＜0．05）；角二阶矩：SHAM组：（0．06±0．01）；OVX 组：（0．08±0．02）（P＜0．05）；对比度（contrast）：SHAM组：（2．92±0．52）OVX组：（1．99±0．67）（P＜0．05）；相关性（correlat）：SHAM组：（0．54±0．02）；OVX组：（0．48±0．02）（P＜0．05）；熵（Entropy）：SHAM组：（1．54±0．08）；OVX组：（1．33±0．11）（P＜0．05）；差方差（ DifVarnc）：SHAM组：（1．54±0．08）；OVX 组：（1．06±0．36）（ P ＜0．05）；游程长不均匀度因子（ RLNonUni ）：SHAM 组：（30932．94±12790．38）；OVX组：（11116．16±1714．33）（P＜0．05）；灰度不均匀度因子（GLevNonU）：SHAM组：（11896．84± 4737．95）；OVX组：（6497．72±1823．29）（P＜0．05）。结论采用纹理分析的方法，可将去卵巢骨质疏松大鼠的骨组织切片图像与正常大鼠图像明显区分。