期刊
EUROPEAN UROLOGY
卷 58, 期 4, 页码 602-608出版社
ELSEVIER SCIENCE BV
DOI: 10.1016/j.eururo.2010.06.002
关键词
Raman spectroscopy; Renal cell carcinoma; Diagnosis; Spectrum analysis
资金
- French National Association for Research (ANR)
Background: New optical techniques of spectroscopy have shown promising results in the evaluation of solid tumours. Objective: To evaluate the potential of Raman spectroscopy (RS) to assess renal tumours at surgery. Design, setting, and participants: Over a 5-mo period, Raman optical spectra were prospectively acquired on surgical renal specimens removed due to suspicion of cancer. Measurements: Raman measures were normalised to ensure comparison between spectra. A lower resolution signal was computed using a wavelet decomposition procedure to diminish the size of the signal and exploit the complete spectrum. A support vector machine (SVM) with a linear kernel and a sequential minimal optimisation solver was applied. A leave-one-out cross-validation technique was used to train and test the SVM. Results and limitations: There were 36 patients with 34 malignant tumours (27 clear-cell, 6 papillary, and 1 chromophobe) and 2 benign (1 oncocytoma and 1 metanephric cyst) tumours. A total of 241 analysable Raman spectra were obtained. The SVM was able to classify tumoural and normal tissue with an accuracy of 84% (sensitivity 82%, specificity 87%). High-grade and low-grade tumours were differentiated with a precision of 82% (sensitivity 84%, specificity 80%). Histologic subtype could be categorised with an accuracy of 93% (sensitivity 96%, specificity 87%). SVM could not be applied to classify benign and malignant tumours because of the restricted number of benign spectra. Conclusions: RS can accurately differentiate normal and tumoural renal tissue, low-grade and high-grade renal tumours, and histologic subtype of renal cell carcinoma. Larger prospective studies are needed to confirm these preliminary data. (C) 2010 European Association of Urology. Published by Elsevier B.V. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据