透過您的圖書館登入
IP:18.117.152.251
  • 學位論文

利用電腦斷層掃描影像評估髖骨骨質密度

Estimation of Hip Bone Mineral Density by Computed Tomography Images

指導教授 : 黃詠暉
共同指導教授 : 陳泰賓(Tai-Been Chen)

摘要


目前臨床可用來測量骨質密度的儀器很多,包括X光、電腦斷層掃描、核子醫學及超音波等,其中僅有雙能量X光檢測儀(Dual-energy X-ray Absorptiometry, DXA)測量結果可為黃金標準,但對於年長者掃描髖部的檢查姿勢較不舒服,且需要固定同一姿勢長達10分鐘之久。而電腦斷層掃描(Computed Tomography, CT)具備非侵入性和掃描速度快的優勢,提供身體不同組織結構的三維立體影像,更可透過CT影像參數評估骨質密度。因此,本研究主要目的是利用CT影像萃取值及生理參數(如性別、年齡和身體質量指數(Body Mass Index, BMI)),評估髖骨的骨質密度。 本研究回溯性收集104名(58名男性,46名女性),從20到70歲,平均年齡42.3±9.5歲)皆有接受骨盆腔電腦斷層掃描和雙能量X光檢測儀之受檢者,取得骨盆腔CT影像和髖骨骨質密度(Bone Mineral Density, BMD)。使用影像處理軟體圈選感興趣區域(Region Of Interest, ROI)範圍為左、右邊的近端股骨、股骨頸、大轉子、股骨頭和股骨幹,其計算體積以及萃取Hounsfield Unit (HU)的平均值(Mean)、標準差(Standard Deviation, SD)、最小值(Minimum, Min)和最大值(Maximum, Max),接著,利用統計分析得到最佳迴歸模型。 左髖骨 (Left Hip) BMD與HUmean具有高度正相關性(迴歸模型為左髖骨BMD = 0.202+ 0.016BMI + 0.002L't HUmean(p <0.001,R2 = 0.581)),而右髖骨 (Right Hip) BMD與HUmean之間為正相關性(迴歸模型為右髖骨 BMD = 0.3 + 0.015BMI + 0.001R't HUmean(p <0.001,R2 = 0.621))。左髖骨T-Score和HUmean的接收者操作特徵曲線 (Receiver Operating Characteristic, ROC) 曲線,顯示左髖骨HUmean的曲線下面積 (Area Under the Curve, AUC) 為0.6059,說明薄弱的鑑別能力,左髖骨HUmean切截點 (Cut-Off Point) 為174.3,敏感度為79.5%,特異性為76.2%;右髖骨T-Score和HUmean的ROC曲線,顯示右髖骨HUmean的AUC為0.6885,說明薄弱的鑑別能力,右髖骨HUmean切截點為183.9,敏感度為72.3%,特異性為95.2%。 評估BMD可藉用CT快速和立即性的影像,作為篩選BMD之工具,然而,HU和BMD之轉換模型仍有改進空間,應該要更多的臨床數據來評定,所提出的方法的準確性和可靠性,以期許取得更好的預測模型,提早預知骨質密度與骨骼結構。

並列摘要


Currently, in clinical practice, there are many instruments used to measure bone density, includes X-ray, computed tomography, nuclear medicine and ultrasound, etc. Among those, the Dual-energy X-ray Absorptiometry (DXA) measurement is regarded as the gold standard, but for the older subjects feel uncomfortable for the need to stabilize in the same position for at least ten minutes during hip scanning. Computed Tomography (CT) not only provides the advantages of being non-invasive and fast scanning, but also provide three dimensional images and evaluates bone mineral density by Hounsfield Unit (HU). The purpose of this study is using HU and physiological parameters (gender, age and Body Mass Index (BMI) to estimate the bone mineral density of hip. A total number of 104 cases (58 male and 46 female) were collected age from 20 to 80 years old, average age 42.3±9.5 years old. Pelvis CT and hip Bone Mineral Density (BMD) were obtained for retrospective study. Meanwhile, the CT image was used to calculus the image-features of Region of Interest (ROIs) located left and right proximal femora, femur neck, trochanter, femur head and femur shaft on CT images. The features were volume, HU mean, Standard Deviation (SD), minimum and maximum in ROIs. The personal physical parameter and images-features were analyzed by correlation coefficients and regression model. Then, using statistical analysis to obtain the best regression model. The left hip BMD estimated by DXA was positive correlation with left HUmean calculated by CT image (i.e., L’t Hip BMD=0.202+0.016BMI+0.002L’t HUmean (p<0.001, R2=0.581)). The left hip BMD estimated by DXA was positive correlation with right HU mean calculated by CT image (i.e., R’t Hip BMD=0.3+0.015 BMI+0.001R’t HUmean (p<0.001, R2=0.621)). The Receiver Operating Characteristic (ROC) curve of the L't Hip T-Score and the HUmean showed that the Area Under the Curve (AUC) was 0.6059, indicating a weak discriminating power, the L't HUmean cut-off point was 174.3. The sensitivity was 79.5% and the specificity was 76.2% for L’t Hip BMD. R't Hip T-Score and the HUmean showed that the AUC was 0.6885, indicating a weak discriminating power, the R't HUmean cut-off point was 183.9. R't Hip the sensitivity was 72.3% and the specificity was 95.2%. The BMD may be estimated by CT images with fast and quickly as a tool for screening BMD. However, HU and BMD of the transformation model still space for improvement should be more clinical data to estimate accuracy and reliability of presented method. In order to obtain a better related model for expected early to predict bone density and structure.

並列關鍵字

BMD CT Hip

參考文獻


[1] C. Celenk, P. Celenk, "Bone Density Measurement Using Computed Tomography," InTech, p.127, 2012.
[2] R.M. Lorente Ramos, J. Azpeitia Armán, N. Arévalo Galeano, A. Mu˜noz Hernández, J.M. García Gómez, J. Gredilla Molinero, "Dual energy X-ray absorptimetry Fundamentals, methodology, " Radiología, vol. 54(5): , pp.410-423, 2011.
[3] J. Pirayesh Islamian, I. Garoosi, K. Abdollahi Fard, and M. R. Abdollahi, "How much intravenous contrast media affect bone mineral density (BMD) assessed by routine computed tomography (CT)," The Egyptian Journal of Radiology and Nuclear Medicine, vol. 47, pp. 571-575, 2016.
[4] L. A. Binkovitz and M. J. Henwood, "Pediatric DXA: technique and interpretation," Pediatr Radiol, vol. 37, pp. 21-31, Jan 2007.
[5] Goldman LW : Principles of CT and CT technology. J Nucl Med Tech-nol 35 : 115-128; quiz 129-130, 2007.

延伸閱讀