CT图像中肿大淋巴结肺癌转移分类方法

刘露; 刘宛予; 楚春雨; 吴军; 周洋; 张红霞; 鲍劼

doi:10.3724/SP.J.1146.2009.00699

CT图像中肿大淋巴结肺癌转移分类方法

doi: 10.3724/SP.J.1146.2009.00699

基金项目:

国家国际科技合作重大专项(2007DFB30320)，国家自然科学基金(60777004)和黑龙江省教育厅科技计划项目(11531048)资助课题

计量
- 文章访问数: 4096
- HTML全文浏览量: 58
- PDF下载量: 852
- 被引次数: 0
出版历程
- 收稿日期: 2009-05-11
- 修回日期: 2009-07-13
- 刊出日期: 2009-10-19

Classification of Tumid Lymph Nodes Metastases and Non-Metastases from Lung Cancer in CT Image

摘要

摘要: 为解决肺癌N分期中胸部CT难于对肿大淋巴结是否癌转移进行评价的问题，寻求能够有效表示淋巴结病理特性的图像特征，实现对肿大淋巴结癌转移快速准确地判别。该文采取交互式分割从CT图像中提取出肿大淋巴结；直接计算淋巴结的多分辨率直方图得到200维空间信息特征样本集；利用具有处理高维数据集优势的支持向量机(SVM)构造分类器；用测试集对经训练的SVM分类器进行测试以评价分类性能。经96例病例实验结果表明：100个淋巴结图像的200维特征计算用时1.91 s，SVM分类器训练测试用时1.36 s，敏感性76%，特异性64%，准确度70%，接受者操作特性曲线(ROC)下面积(AUC)0.6525。高维图像空间信息特征能够有效表示淋巴结特性；没有考虑医学征象进行肿大淋巴结癌转移定性诊断的准确度就达到了70%，同时分类速度比传统纹理算法提高了约10倍。
- 肺癌N分期; CT图像; 肿大淋巴结; 多分辨率直方图; 支持向量机(SVM)
Abstract: In order to solve the low accuracy diagnosis of metastases and non-metastases tumid lymph nodes in the lung cancer N stage with chest CT images, effective image features of lymph nodes need to be found for quickly and accurately differentiating metastases and non-metastases tumid lymph nodes. First, tumid lymph nodes are extracted from chest CT images using interactive segmentation. Second, the multi-resolution histograms of tumid lymph nodes are directly calculated to receive a high-dimensional features sample set with spatial information. Then the classifier for differentiating metastases and non-metastases tumid lymph nodes is constructed with making full use the advantage of SVM which is good at dealing with high dimensional data sets. Finally, the performance of classification is evaluated by testing the trained SVM with the test sample set. The test results by 96 cases show that it takes 1.91 s for computing 200 dimensional features of 100 lymph nodes, 1.36 s for training and testing the SVM classifier. Receiver Operating Characteristic (ROC) analysis of the classification performance shows that the sensitivity is 76%, specificity is 64%, accuracy is 70%, and the Area Under Curve (AUC) is nearly 0.6525. Image spatial information can effectively express the characteristics of lymph nodes, the classification accuracy of metastases and non-metastases tumid lymph nodes is up to 70% without medical signs, and the classification speed is about 10 times than traditional texture methods. It provides a feasible, simple, objective method for improving the accuracy of the lung cancer N stage in medical imaging diagnosis.

HTML全文

参考文献(1)

Rankin S C. The role of positron emission tomography instaging of non-small cell lung cancer [J].Targeted Oncology.2008, 3(3):149-159[2]An Y S, Sun J S, and Park K J, et al.. Diagnosticperformance of 18F-FDG PET/CT for lymph node staging inpatients with operable non-small-cell lung cancer andinflammatory lung disease [J].Lung.2008, 186(5):327-336[3]Smulders S A, Gundy C M, and Lingen A, et al.. ObserverVariation of 2-Deoxy-2-[F-18]fluoro-D-Glucose-PositronEmission Tomography in Mediastinal Staging of Non-SmallCell Lung Cancer as a Function of Experience, and itsPotential Clinical Impact [J].Molecular Imaging and Biology.2007, 9(5):318-322[4]张晓颖, 韩冰, 侯跃芳, 等. 非小细胞肺癌淋巴结分期的PET/CT 准确性评价Meta 分析[J]. 中国临床医学影像杂志, 2009, 20(3): 184-188.Zhang X Y, Han B, and Hou Y F, et al.. Accuracy of PET/CTon lymph nodes staging in non-small cell lung cancerAMeta-analysis [J]. Journal of China Clinic Medical Imaging,2009, 20(3): 184-188.[5]党亚萍, 王琦. 18F-FDG PET/CT 在纵隔淋巴结鉴别诊断中的价值[J]. 中国医学影像技术, 2008, 24(5): 781-784.Dang Y P and Wang Q. Differential diagnosis of mediastinallymph node by PET/CT [J]. Chinese Journal of MedicalImaging Technology, 2008, 24(5): 781-784.[6]杨光, 姚树展, 黄葵红, 等.18F-PET-CT 显像在肺癌纵隔淋巴结转移中临床应用价值[J]. 山东大学学报(医学版), 2007,45(8): 846-857.Yang G, Yao S Z, and Huang K H, et al.. 18F-FDG PET-CT indetecting lymph nodes metastasis of lung cancer [J]. Journalof Shandong University (Health Sciences), 2007, 45(8):846-857.[7]张成琪, 王广丽, 宋吉清, 等. 18F-FDG PET-CT 评价非小细胞肺癌淋巴结分期的价值[J]. 山东大学学报(医学版), 2007,45(10): 1038.Zhang C Q, Wang G L, and Song J Q, et al.. 18F-FDGPET-CT for lymph node staging in non-small cell lungcancer[J]. Journal of Shan-dong University (Health Sciences),2007, 45(10): 1038.[8]Sone S, Li F, and Takashima Y, et al.. Characteristics ofsmall cancer invisible on conventional chest radiography anddetected by population based screening using spiral CT [J].Br J Radiol, 2000, 73: 137-145.[9]MCloud T C, Bourgouin P M, and Greenberg R W, et al..Brochogenic carcinoma: analysis of staging in themediastinum with CT by correlative lymph node mappingand sampling [J]. Radiology, 1992, 182(2): 319-323.[10]郑列, 吴沛宏, 莫运仙, 等. 非小细胞肺癌纵隔淋巴结分期：螺旋CT 和病理检查对照[J]. 癌症, 2006, 25(11): 1384-1388.Zheng L, Wu P H, and Mo Y X, et al.. Mediastinal lymphnode staging in non-small cell lung cancer: Comparisonbetween spiral Computed Tomographic(CT) findings andpathology [J]. Chinese Journal of Cancer, 2006, 25(11):1384-1388.[11]Kiernan P D, Sheridan M J, and Lamberti J, et al..Mediastinal staging of non-small cell lung carcinoma usingcomputed and positron-emission tomography [J]. South MedJ, 2002, 95(10): 1168-1172.[12]Takamochi K, Nagai K, and Yoshida J, et al.. The role ofcomputed tomographic scanning in diagnosing mediastinalnode involvement in non-small cell lung cancer [J]. J ThoracCardiovasc Surg, 2000, 119(6): 1135-1140.[13]Quint L E, Francis I R, and Wahl R L, et al.. Preoperativestaging of non-small-cell carcinoma of the lung: imagingmethods [J]. AJR Am J Roentgenol, 1995, 164(6): 1349-1359.[14]Keller S M, Adak S, and Wagner H, et al.. Mediastinal lymphnode dissection improves survival in patients with stages IIand IIIa non-small cell lung cancer [J].Ann Thorac Surg.2000, 70:358-365[15]Takamochi K, Nagai K, and Suzuki K, et al.. Clinicalpredictors of N2 disease in non-small cell lung cancer [J].Chest.2000, 117:1577-1582[16]Kunio D. Computer-aided diagnosis in medical imaging:Historical review, current status and future potential [J].Computerized Medical Imaging and Graphics.2007, 31:198-211[17]Liu L, Liu W Y, and Sun X M. Automated detection ofpulmonary nodules in CT images with support vectormachines [C]. Proceedings of the Fifth InternationalSymposium on Instrumentation Science and Technology,Shenyang, China, Sep. 15-18, 2008, Vol.1: 550-555.[18]Campadelli P, Casiraghi E, and Valentini G. Support vectormachines for candidate nodules classification[J].NeuroComputing.2005, Vol.68:281-288[19]Leef J L and Klein J S. The solitary pulmonary nodule[J].Radiol Clin North Am.2002, 40(1):123-143[20]Sortini D, Maravegias K, and Sortini A. Difficulty of earlydiagnosis in patients with solitary pulmonary nodule [J].Thorac Cardiovasc Surg, 2005, 129(5): 1196.[21]Chapelle O, Haffner P, and Vapnik V. SVMs forhistogram-based image classification[J].IEEE Transactionson Neural Networks.1999, 10(5):1055-1065[22]Belikova T P, Yashunskaya N I, and Kogan E A. Computeraideddifferential diagnosis of small solitary pulmonarynodules[J].Computers and Biomedical Research.1996, 29(1):48-62[23]Hadjidemetriou E, Grossberg M D, and Nayar S K. Multiresolutionhistogram and their use for recognition[J].IEEETransactions on Pattern Analysis and Machine Intelligence.2004, 26(7):831-847[24]刘欣悦, 黄廉卿. 利用多分辨率直方图特征分类数字X 光乳腺图象[J]. 光学精密工程, 2006, 14(2): 327-332.Liu X Y and Huang L Q. Classification of digitalmammograms using multi-resolution histogram feature [J].Optics and Precision Engineering, 2006, 14(2): 327-332.[25]Hadjidemetriou E, Grossberg M D, and Nayar S K. Spatialinformation in multi-resolution histograms[C]. Proceedings ofIEEE Conference on Computer Vision and PatternRecognition, 2001, Vol.1: 702-709.[26]Vapnik V. Statistical learning theory [M]. New York: Wiley,1998.[27]El-naqa I, Yang Y, and Wernick M N, et al.. A supportvector machine approach for detection of micro-calcifications[J].IEEE Transactions on Medical Imaging.2002, 21(12):1552-1563[28]Chang R F, Wu W J, and Moon W K, et al.. Support vectormachines for diagnosis of breast tumors on US images [J].Academic Radiology.2003, 10(2):189-197

施引文献

资源附件(0)

访问统计