| 摘要: | 研究目標：以微米級電腦斷層掃描(Micro-computer tomography, Micro-CT)分析骨小樑結構為基準，評估牙科錐狀射束電腦斷層掃描(Dental cone beam computer tomography，CBCT)影像，經影像前處理後，分析骨小樑結構的準確度。
研究材料與方法：本研究有30個去除皮質骨的牛脊椎骨塊。CBCT掃描機型為Asahi AZ 3000，解析度為155m與100m，掃描中同時拍攝標準骨礦物質密度(Bone mineral density, BMD)假體做為後續樣本BMD分析的依據。Micro-CT掃描機型為Skyscan 1076，解析度18m。所有影像均匯入ImageJ軟體進行對位與切割，影像分析感興趣範圍仿照人工牙根4mm直徑、10mm長度圓柱體積。CBCT影像首先記錄其灰階值與轉換後BMD值，Micro-CT影像則紀錄灰階值。CBCT影像接續使用本篇的影像前處理方法，無前處理影像紀錄為(group 1)，陸續加入三種前處理步驟：銳利化與去除雜訊濾鏡(group 2)、局部二元化(group 3)、排除空氣影像(group 4)。Micro-CT影像則無前處理。所有影像均匯入CTAn軟體進行四項骨小樑結構參數分析：骨體積總體積比(Bone volume total volume fraction, BV/TV)、骨小樑平均粗細(Trabecular thickness, Tb.Th.)、骨小樑平均數量(Trabecular number, Tb.N.)、骨小樑平均距離(Trabecular separation, Tb.Sp.)。本篇採用有母數統計探討兩種解析度，group1~4 CBCT影像分析結果與Micro-CT分析結果的差異與相關性。
研究結果：155m解析度CBCT BMD大於100m CBCT BMD，但兩組BMD與Micro-CT灰階值均有高度相關。使用三道前處理的Group 4分析結果與Micro-CT分析結果數值比較如後，BV/TV：18m Micro- CT <100m CBCT <155m CBCT；Tb.Th.：18m Micro-CT <100m CBCT <155m CBCT；Tb.N.：100m CBCT ?155m CBCT <18m Micro-CT；Tb.Sp.：18m Micro-CT <155m CBCT ?100m CBCT。兩種解析度CBCT影像隨著三道分析前處理加入，BV/TV、Tb.N.、Tb.Sp.分析結果與Micro-CT分析結果間相關度依序提高，在group4影像能達到高度相關，Tb.Th.無論前處理加入與否均呈現低度相關。
結論：100m與155m解析度CBCT影像骨小樑結構分析結果絕對數值與Micro-CT分析結果有差異。相關度部分，本篇提出的前處理方式能大幅提升使用CBCT量測BV/TV、Tb.N.、Tb.Sp.骨小樑結構參數的準確度。
Objective: Validate accuracy of dental cone beam computer tomography (CBCT) trabecular bone structure analysis in comparison with micro- computer tomography (Micro-CT).
Materials and methods: 30 de-cortical bone blocks from cattle vertebrate were scan with Asahi AZ3000 CBCT in resolution of 100m and 155m, Skyscan1076 Micro-CT of 18m. Standard bone mineral density (BMD) phantom also scan with specimens during CBCT scan. ImageJ software was used for image alignment and cropping. Region of interest was set with 4mm diameter, 10mm length cylinder volume. CBCT BMD and Micro-CT grayscale were assessed. Before analysis, new image processing method was used in CBCT images (without processing image record as group 1), with following 3 steps: image sharpening and de-speckable filter (group2), local threshold (group 3), and air bubble eliminated (group4). All images analysis used CTAn software and four trabecular structure parameters were measured: Bone volume total volume fraction (BV/TV), trabecular thickness (Tb.Th.), trabecular number (Tb.N.) and trabecular separation (Tb.Sp.). Parametric statistics used in this study.
Results: 155m resolution CBCT BMD was larger than 100m result, high correlation between CBCT BMD and Micro-CT grayscale. Compare group 4 CBCT trabecular bone analysis result to Micro-CT result, BV/TV were as follows: 18m Micro-CT <100m CBCT <155m CBCT; Tb.Th.: 18m Micro-CT<100m CBCT <155m CBCT; Tb.N.: 100m CBCT ?155m CBCT <18m Micro-CT; Tb.Sp.: 18m Micro-CT <155 m CBCT ?100m CBCT. With the image processing method used, BV/TV, Tb.N. and Tb.Sp. showed increasing correlation between CBCT and Micro-CT result, especially in group 4 high correlations was found. However, Tb.Th. always showed low correlation.
Conclusion: Significant different was found between CBCT and Micro- CT trabecular analysis result. However, we can increase the accuracy of CBCT trabecular bone analysis with our new image processing method. (BV/TV, Tb.N., Tb.Sp., except Tb.Th.) |
