Raman spectroscopic grading of astrocytoma tissues: using soft reference information

  • Claudia BeleitesEmail author
  • Kathrin Geiger
  • Matthias Kirsch
  • Stephan B. Sobottka
  • Gabriele Schackert
  • Reiner Salzer
Original Paper


Gliomas are the most frequent primary brain tumours. During neurosurgical treatment, locating the exact tumour border is often difficult. This study assesses grading of astrocytomas based on Raman spectroscopy for a future application in intra-surgical guidance. Our predictive classification models distinguish the surgically relevant classes “normal tissue” and “low” and “high grade astrocytoma” in Raman maps of moist bulk samples (80 patients) acquired with a fibre-optic probe. We introduce partial class memberships as a strategy to utilize borderline cases for classification. Borderline cases supply the most valuable training and test data for our application. They are (a) examples of the sought boundary and (b) the cases for which new diagnostics are needed. Besides, the number of suitable training samples increases considerably: soft logistic regression (LR) utilizes 85% more spectra and 50% more patients than linear discriminant analysis (LDA). The predictive soft LR models achieve ca. 85, 67 and 84% (normal, low and high grade) sensitivity and specificity. We discuss the different heuristics of LR and LDA in the light of borderline samples. While we focus on prediction, the spectroscopic interpretation of the predictive models agrees with previous descriptive studies. Unsaturated lipids are used to differentiate between normal and tumour tissues, while the total lipid content prominently contributes to the determination of the tumour grade. The high-wavenumber region above 2,800 cm−1 alone did not allow successful grading. We give a proof of concept for Raman spectroscopic grading of moist astrocytoma tissues and propose to include borderline samples into classifier training and testing.


Gliomas Astrocytomas Grading Classification Tumour Raman spectroscopy Linear discriminant analysis Logistic regression Soft classification 



C. Beleites acknowledges financial support by a scholarship of Deutsche Telekom Stiftung, the IRCCS Burlo Garofolo and the Associazione per i Bambini Chirurgici del Burlo.

Supplementary material

216_2011_4985_MOESM1_ESM.pdf (486 kb)
ESM 1 Fig. S1. More detailed version of the diagram in Fig. 8: the stability of the predictions of the different models (columns) for each predicted class separately (rows). The boxes mark median and quartiles, the whiskers extend to the last value inside 1.5 IQR from the box, all further values are marked by points [47]. Fig. S2. Both MAE and RMSE have the same pattern across the models, and are very similar for all but the LR–highwn models. Low-grade tissues are the most difficult class as they lie in between the normal and high-grade tumour tissue: normal tissue is confused with low-grade tumours, but not with high-grade tissue, and vice versa. The soft LR models have slightly increased MAE for the predicted memberships of crisply labelled spectra, which is counterweighted by a decreased MAE of the soft spectra. Their RMSE does not show this increase in these cases, but rather an improvement. Also the improvement for soft samples is more pronounced. Together, these findings allow the conclusion that the LR–soft models have more small deviations from the reference, while the LR–crisp and LDA models have larger deviations for fewer spectra. This is in accordance with the LR–soft modelling smoother class transitions. The boxes mark median and quartiles, the whiskers extend to the last value inside 1.5 IQR from the box, all further values are marked by points [47]. Fig. S3. Upper part: detailed Versions of the contributions to the rotated LDA model: median and quartiles, lower part: spectra before (bottom; black: mean normal grey matter spectrum subtracted for “centering”) and after “centering” (second lowest row). Fig. S4. Performance of the LDA (continuous) and crisp LR (dashed) models (top row), and crisp and soft LR (lower row), respectively: median, 5th and 95th percentiles observed over the 125 iterations of the cross-validation, thresholds 0.01–0.99. Both LDA and crisp LR models have virtually the same performance for the low-grade tumours, and LDA reaches slightly higher sensitivities for normal tissue. The advantage of the LDA models is most pronounced for the high-grade tumours. Soft LR outperforms crisp LR for all classes, the improvement is most pronounced for normal tissue, but most important for A °II (PDF 485 kb)


  1. 1.
    VandenBerg SR (1992) Current diagnostic concepts of astrocytic tumors. J Neuropathol Exp Neurol 51(6):644–657CrossRefGoogle Scholar
  2. 2.
    Louis DN, Ohgaki H, Wiestler OD, Cavenee WK, Burger PC, Jouvet A, Scheithauer BW, Kleihues P (2007) The 2007 who classification of tumours of the central nervous system. Acta Neuropathol 114(2):97–109CrossRefGoogle Scholar
  3. 3.
    Kros JM, Gorlia T, Kouwenhoven MC, Zheng PP, Collins VP, Figarella-Branger D, Giangaspero F, Giannini C, Mokhtari K, Mørk SJ, Paetau A, Reifenberger G, van den Bent MJ (2007) Panel review of anaplastic oligodendroglioma from European organization for research and treatment of cancer trial 26951: assessment of consensus in diagnosis, influence of 1p/19q loss, and correlations with outcome. J Neuropathol Exp Neurol 66(6):545–551CrossRefGoogle Scholar
  4. 4.
    Duran I, Raizer JJ (2007) Low-grade gliomas: management issues. Expert Rev Anticancer Ther 7(12 Suppl):S15–S21CrossRefGoogle Scholar
  5. 5.
    Stupp R, Reni M, Gatta G, Mazza E, Vecht C (2007) Anaplastic astrocytoma in adults. Crit Rev Oncol Hematol 63(1):72–80CrossRefGoogle Scholar
  6. 6.
    Diener HC, Putzki N, Berlit P, Hacke W, Hufnagel A, Hufschmidt A, Mattle H, Meier U, Oertel W, Reichmann H, Rieckmann P, Schmutzhard E, Wallesch CW, Weller M (2008) Leitlinien für Diagnostik und Therapie in der Neurologie, 4th edn. Thieme, Stuttgart, URL Accessed 2011-01-08
  7. 7.
    Stummer W, Pichlmeier U, Meinel T, Wiestler OD, Zanella F, Reulen HJ, Group ALAGS (2006) Fluorescence-guided surgery with 5-aminolevulinic acid for resection of malignant glioma: a randomised controlled multicentre phase III trial. Lancet Oncol 7(5):392–401CrossRefGoogle Scholar
  8. 8.
    Sobottka SB, Geiger KD, Salzer R, Schackert G, Krafft C (2009) Suitability of infrared spectroscopic imaging as an intraoperative tool in cerebral glioma surgery. Anal Bioanal Chem 393(1):187–195CrossRefGoogle Scholar
  9. 9.
    Krafft C, Sergo V (2006) Biomedical applications of Raman and infrared spectroscopy to diagnose tissues. Spectroscopy 20:195–218Google Scholar
  10. 10.
    Krafft C, Steiner G, Beleites C, Salzer R (2009) Disease recognition by infrared and Raman spectroscopy. J Biophotonics 2(1–2):13–28CrossRefGoogle Scholar
  11. 11.
    Kendall C, Isabelle M, Bazant-Hegemark F, Hutchings J, Orr L, Babrah J, Baker R, Stone N (2009) Vibrational spectroscopy: a clinical tool for cancer diagnostics. Analyst 134(6):1029–1045CrossRefGoogle Scholar
  12. 12.
    Koljenović S, Choo-Smith LP, Schut TCB, Kros JM, van den Berge HJ, Puppels GJ (2002) Discriminating vital tumor from necrotic tissue in human lioblastoma tissue samples by raman spectroscopy. Lab Invest 82(10):1265–1277Google Scholar
  13. 13.
    Koljenović S, Schut TCB, Wolthuis R, de Jong B, Santos L, Caspers PJ, Kros JM, Puppels GJ (2005) Tissue characterization using high wave number Raman spectroscopy. J Biomed Opt 10(3):031116-1–031116-11Google Scholar
  14. 14.
    Krafft C, Sobottka SB, Schackert G, Salzer R (2005) Near infrared Raman spectroscopic mapping of native brain tissue and intracranial tumors. Analyst 130(7):1070–1077CrossRefGoogle Scholar
  15. 15.
    Amharref N, Beljebbar A, Dukic S, Venteo L, Schneider L, Pluot M, Manfait M (2007) Discriminating healthy from tumor and necrosis tissue in rat brain tissue samples by raman spectral imaging. Biochim Biophys Acta 1768(10):2605–2615CrossRefGoogle Scholar
  16. 16.
    Krafft C, Kirsch M, Beleites C, Schackert G, Salzer R (2007) Methodology for fiber-optic Raman mapping and FTIR imaging of metastases in mouse brains. Anal Bioanal Chem 389(4):1133–1142CrossRefGoogle Scholar
  17. 17.
    Beljebbar A, Dukic S, Amharref N, Manfait M (2010) Ex vivo and in vivo diagnosis of c6 glioblastoma development by Raman spectroscopy coupled to a microprobe. Anal Bioanal ChemGoogle Scholar
  18. 18.
    Kirsch M, Schackert G, Salzer R, Krafft C (2010) Raman spectroscopic imaging for in vivo detection of cerebral brain metastases. Anal Bioanal Chem 398(4):1707–1713CrossRefGoogle Scholar
  19. 19.
    Krafft C, Sobottka SB, Schackert G, Salzer R (2004) Analysis of human brain tissue, brain tumors and tumor cells by infrared spectroscopic mapping. Analyst 129(10):921–925CrossRefGoogle Scholar
  20. 20.
    Beleites C, Salzer R (2008) Assessing and improving the stability of chemometric models in small sample size situations. Anal Bioanal Chem 390(5):1261–1271CrossRefGoogle Scholar
  21. 21.
    Utzinger U, Richards-Kortum RR (2003) Fiber optic probes for biomedical optical spectroscopy. J Biomed Opt 8(1):121–147CrossRefGoogle Scholar
  22. 22.
    Santos LF, Wolthuis R, Koljenović S, Almeida RM, Puppels GJ (2005) Fiber-optic probes for in vivo Raman spectroscopy in the high-wavenumber region. Anal Chem 77(20):6747–6752CrossRefGoogle Scholar
  23. 23.
    Prayson RA, Agamanolis DP, Cohen ML, Estes ML, Kleinschmidt-DeMasters BK, Abdul-Karim F, McClure SP, Ebek BA, Vinay R (2000) Interobserver reproducibility among neuropathologists and surgical pathologists in fibrillary astrocytoma grading. J Neurol Sci 175(1):33–39CrossRefGoogle Scholar
  24. 24.
    Louis DN, Cavenee WK, Ohgaki H, Wiestler OD (eds) (2007) WHO Classification of tumors of the central nervous system, Word Health Organization, chap Astrocytic tumorsGoogle Scholar
  25. 25.
    Wrensch M, Rice T, Miike R, McMillan A, Lamborn KR, Aldape K, Prados MD (2006) Diagnostic, treatment, and demographic factors influencing survival in a population-based study of adult glioma patients in the San Francisco Bay area. Neuro Oncol 8(1):12–26CrossRefGoogle Scholar
  26. 26.
    Hastie T, Tibshirani R, Friedman J (2001) The elements of statistical learning; data mining. Inference and prediction. Springer Verlag, New YorkGoogle Scholar
  27. 27.
    Beleites C, Baumgartner R, Bowman C, Somorjai R, Steiner G, Salzer R, Sowa MG (2005) Variance reduction in estimating classification error using sparse datasets. Chemom Intell Lab Syst 79:91–100CrossRefGoogle Scholar
  28. 28.
    Kowalski B, Wold S (1982) Pattern recognition in chemistry. In: Krishnajah PR, Kanal LN (eds) Pattern recognition and reduction of dimensionality, handbook of statistics, vol II. North-Holland, Amsterdam, pp 673–697Google Scholar
  29. 29.
    Huberty CJ (1994) Applied discriminant analysis. Wiley, New YorkGoogle Scholar
  30. 30.
    Vittinghoff E, McCulloch CE (2007) Relaxing the rule of ten events per variable in logistic and cox regression. Am J Epidemiol 165(6):710–718CrossRefGoogle Scholar
  31. 31.
    Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR (1996) A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol 49(12):1373–1379CrossRefGoogle Scholar
  32. 32.
    Foody GM (2002) Status of land cover classification accuracy assessment. Remote Sens Environ 80(1):185–201CrossRefGoogle Scholar
  33. 33.
    Sattlecker M, Bessant C, Smith J, Stone N (2010) Investigation of support vector machines and Raman spectroscopy for lymph node diagnostics. Analyst 135(5):895–901CrossRefGoogle Scholar
  34. 34.
    Hedegaard M, Krafft C, Ditzel HJ, Johansen LE, Hassing S, Popp J (2010) Discriminating isogenic cancer cells and identifying altered unsaturated fatty acid content as associated with metastasis status, using k-means clustering and partial least squares-discriminant analysis of Raman maps. Anal Chem 82(7):2797–2802CrossRefGoogle Scholar
  35. 35.
    Venables WN, Ripley BD (2002) Modern Applied Statistics with S, 4th edn. Springer, New York, URL Accessed 2010-02-21
  36. 36.
    Teh S, Zheng W, Ho K, Teh M, Yeoh K, Huang Z (2009) Near-infrared Raman spectroscopy for gastric precancer diagnosis. J Raman Spectrosc 40(8):908–914, cited By (since 1996) 4CrossRefGoogle Scholar
  37. 37.
    da Silva Martinho H, de Oliveira Monteiro da Silva CM, Yassoyama MCBM, de Oliveira Andrade P, Bitar RA, do Espírito Santo AM, Arisawa EAL, Martin AA (2008) Role of cervicitis in the Raman-based optical diagnosis of cervical intraepithelial neoplasia. J Biomed Opt 13(5):054–029Google Scholar
  38. 38.
    Teh SK, Zheng W, Ho KY, Teh M, Yeoh KG, Huang Z (2010) Near-infrared Raman spectroscopy for early diagnosis and typing of adenocarcinoma in the stomach. Br J Surg 97(4):550–557CrossRefGoogle Scholar
  39. 39.
    Nijssen A, Schut TCB, Heule F, Caspers PJ, Hayes DP, Neumann MHA, Puppels GJ (2002) Discriminating basal cell carcinoma from its surrounding tissue by Raman spectroscopy. J Invest Dermatol 119(1):64–69CrossRefGoogle Scholar
  40. 40.
    Majumder SK, Gebhart S, Johnson MD, Thompson R, Lin WC, Mahadevan-Jansen A (2007) A probability-based spectroscopic diagnostic algorithm for simultaneous discrimination of brain tumor and tumor margins from normal brain tissue. Appl Spectrosc 61(5):548–557CrossRefGoogle Scholar
  41. 41.
    Majumder SK, Keller MD, Boulos FI, Kelley MC, Mahadevan-Jansen A (2008) Comparison of autofluorescence, diffuse reflectance, and Raman spectroscopy for breast tissue discrimination. J Biomed Opt 13(5):054–009CrossRefGoogle Scholar
  42. 42.
    Kanter EM, Majumder S, Vargis E, Robichaux-Viehoever A, Kanter GJ, Shappell HHWJ III, Mahadevan-Jansen A (2009) Multiclass discrimination of cervical precancers using Raman spectroscopy. J Raman Spectrosc 40:204–211CrossRefGoogle Scholar
  43. 43.
    Kohavi R (1995) A study of cross-validation and bootstrap for accuracy estimation and model selection. In: Mellish CS (ed) Artificial Intelligence Proceedings 14th International Joint Conference, 20–25. Montréal, Québec, Canada, Morgan Kaufmann, USA, pp 1137–1145Google Scholar
  44. 44.
    Kohavi R (1995) Wrappers for performance enhancement and oblivious decision graphs. PhD thesis, Department of Computer Science, Stanford UniversityGoogle Scholar
  45. 45.
    Ellison S, Fearn T (2005) Characterising the performance of qualitative analytical methods: statistics and terminology. TrAC 24(6):468–476Google Scholar
  46. 46.
    Fedorov V, Mannino F, Zhang R (2009) Consequences of dichotomization. Pharm Stat 8:50–61CrossRefGoogle Scholar
  47. 47.
    R Development Core Team (2010) R: A Language and Environment for Statistical Computing. ISBN 3-900051-07-0Google Scholar
  48. 48.
    Wickham H (2009) ggplot2: elegant graphics for data analysis. Springer New York, URL Accessed 2010-07-08
  49. 49.
    Beleites C, Sergo V (2011) hyperSpec: a package to handle hyperspectral data sets in R. URL, R package version 0.96
  50. 50.
    Bengtsson H, Riedy J (2008) R.matlab: Read and write of MAT files together with R-to-Matlab connectivity. URL, Accessed 2009-07-10, R package version 1.2.4
  51. 51.
    Mevik BH, Wehrens R (2007) The pls package: principal component and partial least squares regression in R. J Stat Softw 18(2):1Google Scholar
  52. 52.
    Salzberg S (1997) On comparing classifiers: pitfalls to avoid and a recommended approach. Data Min Knowl Discov 1(3):317–328CrossRefGoogle Scholar
  53. 53.
    Köhler M, Machill S, Salzer R, Krafft C (2009) Characterization of lipid extracts from brain tissue and tumors using Raman spectroscopy and mass spectrometry. Anal Bioanal Chem 393(5):1513–1520CrossRefGoogle Scholar
  54. 54.
    Sing T, Sander O, Beerenwinkel N, Lengauer T (2005) ROCR: visualizing classifier performance in R. Bioinformatics 21(20):3940–3941, Google Scholar
  55. 55.
    Kendall C, Stone N, Shepherd N, Geboes K, Warren B, Bennett R, Barr H (2003) Raman spectroscopy, a potential tool for the objective identification and classification of neoplasia in Barrett’s oesophagus. J Pathol 200(5):602–609CrossRefGoogle Scholar
  56. 56.
    Barker M, Rayens W (2003) Partial least squares for discrimination. J Chemom 17(3):166–173CrossRefGoogle Scholar
  57. 57.
    Carr D, Lewin-Koh N, Maechler M (2010) hexbin: Hexagonal Binning Routines. URL Accessed 2010-07-08, R package version 1.22.0, ported by Nicholas Lewin-Koh and Martin Maechler
  58. 58.
    Beleites C (2003) Chemometrische Auswertung von IR-Images und -Maps. Master’s thesis, Technische Universität DresdenGoogle Scholar
  59. 59.
    Krafft C, Neudert L, Simat T, Salzer R (2005) Near infrared Raman spectra of human brain lipids. Spectrochim Acta A Mol Biomol Spectrosc 61(7):1529–1535CrossRefGoogle Scholar
  60. 60.
    Beljebbar A, Amharref N, Lévèques A, Dukic S, Venteo L, Schneider L, Pluot M, Manfait M (2008) Modeling and quantifying biochemical changes in c6 tumor gliomas by Fourier transform infrared imaging. Anal Chem 80(22):8406–8415CrossRefGoogle Scholar
  61. 61.
    Socrates G (2001) Infrared and Raman Characteristic Group Frequencies, 3rd edn. WileyGoogle Scholar
  62. 62.
    Diem M, Griffiths PR, Chalmers JM (eds) (2008) Vibrational spectroscopy for medical diagnosis. WileyGoogle Scholar

Copyright information

© Springer-Verlag 2011

Authors and Affiliations

  • Claudia Beleites
    • 1
    • 2
    Email author
  • Kathrin Geiger
    • 3
  • Matthias Kirsch
    • 3
  • Stephan B. Sobottka
    • 3
  • Gabriele Schackert
    • 3
  • Reiner Salzer
    • 2
  1. 1.CENMAT and DI3University of TriesteTriesteItaly
  2. 2.Analytical ChemistryDresden University of TechnologyDresdenGermany
  3. 3.University Hospital Carl-Gustav CarusDresden University of TechnologyDresdenGermany

Personalised recommendations