Advertisement

Automated Prostate Cancer Localization with Multiparametric Magnetic Resonance Imaging

  • Yusuf Artan
  • Imam Samil Yetik
  • Masoom A. Haider
Chapter

Abstract

Prostate cancer is a leading cause of cancer death for men in the world. Fortunately, the survival rate for early-diagnosed patients is relatively high. Therefore, in vivo imaging plays an important role for the detection and treatment of the disease. Accurate prostate cancer localization with noninvasive imaging can be used to guide biopsy, radiotherapy, and surgery as well as to monitor disease progression. Magnetic resonance imaging (MRI) performed with an endorectal coil provides higher prostate cancer localization accuracy, when compared to transrectal ultrasound (TRUS). However, in general, a single type of MRI is not sufficient for reliable tumor localization. As an alternative, multispectral MRI, i.e., the use of multiple MRI-derived datasets, has emerged as a promising noninvasive imaging technique for the localization of prostate cancer; however, almost all studies are with human readers. There is a significant inter- and intra-observer variability for human readers, and it is substantially difficult for humans to analyze the large dataset of multispectral MRI. To solve these problems, this study presents an automated localization method. We first perform tests to see the best performing combination of multiparametric MRI, then develop localization methods using cost-sensitive support vector machines (SVMs), and show that this method results in improved localization accuracy than classical SVM. Additionally, we develop a new segmentation method by combining conditional random fields (CRF) with a cost-sensitive framework and show that our method further improves cost-sensitive SVM (C-SVM) results by incorporating spatial information. We test SVM, C-SVM, and the proposed cost-sensitive CRF on multispectral MRI datasets acquired from 21 biopsy-confirmed cancer patients. Our results show that multispectral MRI helps to increase the accuracy of prostate cancer localization when compared to single MR images and that using advanced methods such as C-SVM as well as the proposed cost-sensitive CRF can boost the performance significantly when compared to SVM. We finally discuss potentially effective methods of localization using texture as the next steps of research.

Prostate cancer is a leading cause of cancer death for men in the world. Fortunately, the survival rate for early-diagnosed patients is relatively high. Therefore, in vivo imaging plays an important role for the detection and treatment of the disease. Accurate prostate cancer localization with noninvasive imaging can be used to guide biopsy, radiotherapy, and surgery as well as to monitor disease progression. Magnetic resonance imaging (MRI) performed with an endorectal coil provides higher prostate cancer localization accuracy, when compared to transrectal ultrasound (TRUS). However, in general, a single type of MRI is not sufficient for reliable tumor localization. As an alternative, multispectral MRI, i.e., the use of multiple MRI-derived datasets, has emerged as a promising noninvasive imaging technique for the localization of prostate cancer; however, almost all studies are with human readers. There is a significant inter- and intra-observer variability for human readers, and it is substantially difficult for humans to analyze the large dataset of multispectral MRI. To solve these problems, this study presents an automated localization method. We first perform tests to see the best performing combination of multiparametric MRI, then develop localization methods using cost-sensitive support vector machines (SVMs), and show that this method results in improved localization accuracy than classical SVM. Additionally, we develop a new segmentation method by combining conditional random fields (CRF) with a cost-sensitive framework and show that our method further improves cost-sensitive SVM (C-SVM) results by incorporating spatial information. We test SVM, C-SVM, and the proposed cost-sensitive CRF on multispectral MRI datasets acquired from 21 biopsy-confirmed cancer patients. Our results show that multispectral MRI helps to increase the accuracy of prostate cancer localization when compared to single MR images and that using advanced methods such as C-SVM as well as the proposed cost-sensitive CRF can boost the performance significantly when compared to SVM. We finally discuss potentially effective methods of localization using texture as the next steps of research.

Keywords

Support Vector Machine Filter Bank Classical Support Vector Machine Prostate Cancer Localization Conditional Random Field 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

References

  1. 1.
    American Cancer Society (2010) Surveillance and health policy research. American Cancer Society, AtlantaGoogle Scholar
  2. 2.
    Villeirs GM, Verstraete L, De Neve WJ, De Meerleer GO (2005) Magnetic resonance imaging anatomy of the prostate and periprostatic area: a guide for radiotherapists. Radiother Oncol 76:99–106PubMedCrossRefGoogle Scholar
  3. 3.
    Ito H, Kamoi K, Yokoyama K, Yamada K, Nishimura T (2003) Visualization of prostate cancer using dynamic contrast-enhanced MRI. Comparison with transrectal power doppler ultrasound. Br J Radiol 76(909):617–624PubMedCrossRefGoogle Scholar
  4. 4.
    van Dorsten FA, van der Graaf M, Engelbrecht MR, van Leenders GJ, Verhofstad A, Ripkema M, de la Rosette JJ, Heerschap A (2004) Combined quantitative enhanced MR imaging and (1)H MR spectroscopic imaging of human prostate cancer. J Magn Reson Imaging 20(2):279–287PubMedCrossRefGoogle Scholar
  5. 5.
    Futterer J, Heijmink S, Scheenen TW, Veltman J, Huisman HJ, Vos P, Hulsbergen-Van de Kaa CA, Witjes JA, Krabbe PF, Heerschap A, Barentsz JO (2006) Prostate cancer localization with dynamic contrast-enhanced MR imaging and proton MR spectroscopic imaging. Radiology 241(2):449–458PubMedCrossRefGoogle Scholar
  6. 6.
    Turkbey B, Albert PS, Kurdziel K, Choyke PL (2009) Imaging localized prostate cancer. Current approaches and new developments. Am J Roentgenol 192(6):1471–1480CrossRefGoogle Scholar
  7. 7.
    Futterer JJ, Barentsz J, Heijmink S (2009) Imaging modalities for prostate cancer. Expert Rev Anticancer Ther 9(7):923–937PubMedCrossRefGoogle Scholar
  8. 8.
    Ocak I, Bernardo M, Metzger G, Barrett T, Pinto P, Albert PS, Choyke PL (2007) Dynamic contrast-enhanced MRI of prostate cancer at 3 T. A study of pharmacokinetic parameters. Am J Roentgenol 189(4):192–200CrossRefGoogle Scholar
  9. 9.
    Girouin N, Mege-Lechevallier F, Senes S, Bissery S, Rabilloud M, Colombel M, Lyonnet D, RouviÔøΩre O (2007) Prostate dynamic contrast-enhanced MRI with simple visual diagnostic criteria. Is it reasonable? Eur J Radiol 17(6):1498–1509CrossRefGoogle Scholar
  10. 10.
    Haider M, van der Kwast TH, Tanguay J, Evans AJ, Hashmi AT, Lockwood G, Trachtenberg J (2007) Combined T2-weighted and diffusion weighted MRI for localization of prostate cancer. Am J Roentgenol 189(2):323–328CrossRefGoogle Scholar
  11. 11.
    Scheidler J, Hricak H, Vigneron DB et al (1999) Prostate cancer. Localization with three dimensional proton MR spectroscopic imaging-clinicopathologic study. Radiology 213:473–480PubMedCrossRefGoogle Scholar
  12. 12.
    Yoshikazo T, Wada A, Hayashi T, Uchida K, Sumura M, Uchida N, Kitagaki H, Igawa M (2008) Usefulness of diffusion-weighted imaging and dynamic contrast enhanced magnetic resonance imaging in the diagnosis of prostate transition-zone cancer. Acta Radiol 49(10):1208–1213Google Scholar
  13. 13.
    Davenport MA, Baraniuk RG, Scott CD (2006) Controlling false alarms with support vector machines. In: Proceedings of IEEE international conference on Acoustics, Speech, and Signal Processing (ICASSP), MIT Press, Cambridge, MA, pp 589–593Google Scholar
  14. 14.
    Liu X, Yetik IS, Wernick M et al (2009) Prostate cancer segmentation with simultaneous estimation of the MRF parameters and the class. IEEE Trans Med Imaging 28:906–915PubMedCrossRefGoogle Scholar
  15. 15.
    Chan I, Wells W, Mulkern RV, Haker S, Zhang J, Zou KH, Maier SE, Tempany CM (2003) Detection of prostate cancer by integration of line-scan diffusion, T2-mapping and T2-weighted magnetic resonance imaging; a multichannel statistical classifier. Med Phys 30(9):2390–2398PubMedCrossRefGoogle Scholar
  16. 16.
    Artan Y, Yetik IS, et al (2009) Prostate cancer segmentation with multispectral MRI using cost sensitive conditional random fields. In: Proceedings of IEEE International Symposium on Biomedical Imaging (ISBI), pp 226–230Google Scholar
  17. 17.
    Vapnik V (1998) Statistical learning theory. Wiley, New YorkGoogle Scholar
  18. 18.
    Ricci E, Perfetti R (2007) Retinal blood vessel segmentation using line operators and support vector classification. IEEE Trans Med Imaging 26:1357–1365PubMedCrossRefGoogle Scholar
  19. 19.
    El-Naqa I, Yang Y, Wernick M, Galatsanos NP, Nishikawa P (2002) A support vector machine approach for detection of microcalcifications. IEEE Trans Med Imaging 21:1552–1563PubMedCrossRefGoogle Scholar
  20. 20.
    Liang L, Cherkassy V, Rottenberg DA (2006) Spatial SVM for feature selection and fMRI activation detection. In: International Joint Conference on Neural Networks (IJCNN), pp 1463–1469Google Scholar
  21. 21.
    Vishvanathan SV, Schaudarolph NN, Schmidt M, Murphy KP (2006) Accelerated training of conditional random fields with stochastic gradient methods. In: Proceedings of International Conference on Machine Learning (ICML), pp 969–976Google Scholar
  22. 22.
    Kumar S, Hebert M (2004) Discriminative random fields. A discriminative framework for contextual interaction in classification. In: Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp 1150–1157Google Scholar
  23. 23.
    Wang Y, Rajapakse J (2006) Contextual modelling of functional MR images with conditional random fields. IEEE Trans Med Imaging 25:804–812PubMedCrossRefGoogle Scholar
  24. 24.
    Lee CH, Greiner R, Schmidt M, (2005) Support vector random fields for spatial classification. Lecture notes in computer science 3721:121–132, SpringerGoogle Scholar
  25. 25.
    Bhattacharyya C, Grate LR, Rizki A et al (2002) Simultaneous relevant feature identification and classification in high-dimensional spaces: an application to molecular profiling data. Signal Process 83:729–743CrossRefGoogle Scholar
  26. 26.
    Carrol CL, Somer FG, McNeal JE, Stammey TA (1987) The abnormal prostate. MR imaging at 1.5-T with histopathologic correlation. Radiology 163(2):521–525PubMedGoogle Scholar
  27. 27.
    Chew HG, Bogner RE, Lim CC (2001) Dual-v support vector machines and applications in multi-class image recognition. In: Proceedings of IEEE international conference on Acoustics, Speech, and Signal Processing (ICASSP), pp 1269–1272Google Scholar
  28. 28.
    Scholkopf B, Smola AJ, Williamson RC, Bartlett PL (2000) New support vector algorithms. Neural Comput 12:1207–1245PubMedCrossRefGoogle Scholar
  29. 29.
    Steinwart I (2003) On the optimal parameter choice for $\nu$-support vector machine. IEEE Trans Pattern Anal Mach Intell 25:1274–1284CrossRefGoogle Scholar
  30. 30.
    Chalimourda A, Scholkopf B, Smola AJ (2004) Experimentally optimal $\nu$-support vector regression for different noise models and parameter settings. Neural Netw 17(1):127–141PubMedCrossRefGoogle Scholar
  31. 31.
    Lafferty J, McCallum A, Pereira F (1986) Conditional rando fields: a probabilistic model. J R Stat Soc 48(3):259–302Google Scholar
  32. 32.
    Besag J (1986) On the statistical analysis of dirty pictures. J R Stat Soc 48(3):259–302Google Scholar
  33. 33.
    Weiss Y (2001) Comparing mean field method and belief propagation for approximate inference in MRFs. Advanced mean field inference methods. MIT Press, Cambridge, MAGoogle Scholar
  34. 34.
    Brix G, Semmler W, Port R, Schad LR, Layer G, Lorenz WJ (1991) Pharmacokinetic parameters in CNS Gd-DTPA enhanced MR imaging. J Comput Assist Tomogr 15(4):621–628PubMedCrossRefGoogle Scholar
  35. 35.
    Tofts PS (1997) Modeling tracer kinetics in dynamic Gd-DTPA MR imaging. J Magn Reson Imaging 7(1):91–101PubMedCrossRefGoogle Scholar
  36. 36.
    Fisher RA (1954) Statistical methods for research workers. Oliver and Boyd, EdinburghGoogle Scholar
  37. 37.
    Metz CE, Herman BA, Shen JH (1998) Maximum likelihood estimation of receiver operating characteristic curves from continuously distributed data. Stat Med 17:1033–1053PubMedCrossRefGoogle Scholar
  38. 38.
    van Rijsbergen CJ (1979) Information retrieval. Butterworth-Heinemann, Newton, MAGoogle Scholar
  39. 39.
    Malik J, Belongie S, Leung T, Shi J (2001) Contour and texture analysis for image segmentation. Int J Comput Vis 43:7–27CrossRefGoogle Scholar
  40. 40.
    Artan Y, Haider MA, Langer DL, Yetik IS (2010) Semi-supervised prostate cancer segmentation with multiparametric MRI. In: Proceedings of 2010 International Symposium on Biomedical Imaging (ISBI), pp 648–651Google Scholar
  41. 41.
    Schlemmer HP, Merkle J, Grobholz R (2004) Can operative contrast-enhanced dynamic MR imaging for prostate cancer predict microvessel density in prostatectomy specimens? Eur Radiol 14:309–317PubMedCrossRefGoogle Scholar
  42. 42.
    Yu KK, Hricak H (2000) Imaging prostate cancer. Radiol Clin North Am 38(1):59–85PubMedCrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media New York 2014

Authors and Affiliations

  • Yusuf Artan
    • 1
  • Imam Samil Yetik
    • 2
  • Masoom A. Haider
    • 3
  1. 1.Medical Imaging Research CenterIllinois Institute of TechnologyChicagoUSA
  2. 2.Department of Electrical and Electronics EngineeringTOBB Economy and Technological UniversityAnkaraTurkey
  3. 3.Institute of Medical Science, Sunnybrook Research InstituteUniversity of TorontoTorontoCanada

Personalised recommendations