Image feature evaluation in two new mammography CAD prototypes
- 127 Downloads
Breast cancer is a common but treatable disease for adult women. Improvements in breast cancer detection and treatment have helped to lower mortality, but there is still a need for further improvements, particularly for better computer-aided diagnosis (CADx) and computer-aided detection (CADe).
Two new CAD prototypes, one CADx and one CADe prototype, were evaluated. The core modules are segmentation of lesions, feature extraction, and classification. The evaluation of microcalcifications and mass lesions is based on the currently largest publicly available Digital Database for Screening Mammography (DDSM) with digitized film mammograms and a smaller data source with high-quality mammograms from digital mammography devices. Two different image analysis approaches used by the respective CAD prototypes were examined and compared. These include the ‘machine learning’ approach and the new ‘knowledge-driven’ approach. Particular emphasis is put on a profound discussion of statistical methods with recommendations for their proper application in order to avoid common errors including feature selection, model fitting, and sampling schemes.
The results show that the classification performance of the investigated CADx prototypes for microcalcifications produced a higher AUC =.777 for 44 machine learning features than for 10 knowledge-driven features (AUC =.657). A combination of both feature sets (53 features) did not substantially raise the classification performance (AUC =.771). These analyses were based on 1,347 and 1,359 ROIs, respectively. Evaluating the CADx prototype with 242 machine learning features on DDSM masses data resulted in an AUC of .862 using 1,934 ROIs. Furthermore, a CADe prototype was applied to three own databases giving information about the true positive detection rate for mass lesions. Depending on the definition of a true positive detection, it produced AUC values of .953, .818, and .954 using 12, 17, and 18 features, respectively.
The comparison of CAD prototypes revealed that the quality of results is highly dependent on the correct usage of statistical models, feature selection methods, and evaluation schemes.
KeywordsMammographie Feature selection Classification DDSM CAD Sampling Selection bias Stepwise regression SVM LDA Classification tree AIC BIC
Unable to display preview. Download preview PDF.
- 1.WHO (ed) (2008) World health statistics. WHO Press, GenevaGoogle Scholar
- 2.Levi F, Lucchini F, Negri E, Vecchia CL (2007) Continuing declines in cancer mortality in the European union. Ann Oncol 18(3):593–595, [Online]. Available: http://annonc.oxfordjournals.org/content/18/3/593.abstract Google Scholar
- 3.Thurfjell EL, Lernevall KA, Taube AA (1994) Benefit of independent double reading in a population-based mammography screening program. Radiology 191(1):241–244 [Online]. Available: http://radiology.rsna.org/content/191/1/241.abstract
- 4.Warren RML, Duffy W (1995) Comparison of single reading with double reading of mammograms, and change in effectiveness with experience. Br J Radiol 68(813):958–962 [Online]. Available: http://bjr.birjournals.org/cgi/content/abstract/68/813/958 Google Scholar
- 5.Harvey SC, Geller B, Oppenheimer RG, Pinet M, Riddell L, Garra B (2003) Increase in cancer detection and recall rates with independent double interpretation of screening mammography. Am J Roentgenol 180(5):1461–1467 [Online]. Available: http://www.ajronline.org/cgi/content/abstract/180/5/1461 Google Scholar
- 6.Taylor P, Potts H (2008) Computer aids and human second reading as interventions in screening mammography: two systematic reviews to compare effects on cancer detection and recall rate. Eur J Cancer 44(6):798–807, April 2008. [Online]. Available: doi:10.1016/j.ejca.2008.02.016
- 8.Rosado B, Menzies S, Harbauer A, Pehamberger H, Wolff K, Binder M, Kittler H (2003) Accuracy of computer diagnosis of melanoma: a quantitative meta-analysis. Arch Dermatol 139(3):361–367 [Online]. Available: http://archderm.ama-assn.org/cgi/content/abstract/139/3/361 Google Scholar
- 9.Heath M, Bowyer K, Kopans D, Moore R, Kegelmeyer WP (2001) The digital database for screening mammography. In: Yaffe M (ed) Proceedings of the fifth international workshop on digital mammography. Medical Physics Publishing, London, pp 212–218Google Scholar
- 10.Elter M, Horsch A, Schöulz-Wendtland R, Sittek H, Athelogou M, Schmidt G, Wittenberg T (2007) A modern benchmark case database for computer-aided diagnosis of breast cancer. Int J Comput Assist Radiol Surg (CARS 2007) 2(S1): 514Google Scholar
- 11.Schönmeyer R, Athelogou M, Sittek H, Ellenberg P, Feehan O, Schmidt G, Binnig G (2011) Cognition network technology prototype of a cad system for mammography to assist radiologists by finding similar cases in a reference database. Int J Comput Assist Radiol Surg 6:127–134, doi:10.1007/s11548-010-0486-8. [Online].
- 12.Athelogou M, Schmidt G, Schäpe A, Baatz M, Binnig G (2007) Cognition network technology—a novel multimodal image analysis technique for automatic identification and quantification of biological image contents. In: Shorte S, Frischknecht F (eds) Imaging cellular and molecular biological functions. Springer, pp. 407–422. [Online]. Available: http://www.springerlink.com/content/u74v217m0381420v
- 13.Horsch A (2011) Biomedical image processing, 1st edn. ch. Melanoma Diagnosis. Springer, HeidelbergGoogle Scholar
- 14.Elter M, Held C (2008) Semiautomatic segmentation for the computer aided diagnosis of clustered microcalcifications. In: Giger ML, Karssemeijer N (eds) Medical imaging 2008: computer-aided diagnosis 6915(1). SPIE, p 691524. [Online]. Available: http://link.aip.org/link/?PSI/6915/691524/1
- 15.Elter M, Bergen T (2009) Incorporating a segmentation routine for mammographic masses into a knowledge-based cadx approach. In: Karssemeijer N, Giger ML (eds) Medical imaging 2009: computer-aided diagnosis, 7260(1). SPIE, p 726025. [Online]. Available: http://link.aip.org/link/?PSI/7260/726025/1
- 16.Elter M, Held C (2010) An improved method for segmentation of mammographic masses. SPIE medical imaging 2010: computer-aided diagnosis (in press)Google Scholar
- 17.Hu MK (1962) Visual pattern recognition by moment invariants. IEEE Trans Inf Theory IT-8: 179–187Google Scholar
- 19.Roß T, Handels H, Busche H, Kreusch J, Wolf HH, Pöppl SJ (1995) Automatische klassifikation hochaufgelöster oberflächenprofile von hauttumoren mit neuronalen netzen. In: DAGM-Symposium pp 379–386Google Scholar
- 20.Galloway MM (1975) Texture analysis using gray level run lengths. Comput Graphics Image Process 4(2):172–179 [Online]. Available: http://www.sciencedirect.com/science/article/B7GXF-4S26XJR-7/2/5a606d689d2f1db4a428360031fd5dcf
- 27.R Development Core Team (2009) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, ISBN 3-900051-07-0. [Online]. Available: http://www.R-project.org
- 28.Metter RLV, Beutel J, Kundel HL (eds) (February 2000) Handbook of medical imaging, physics and psychophysics, corrected ed. Bellingham, SPIE PressGoogle Scholar
- 29.Hothorn T, Hornik K, Zeileis A (2006) Unbiased recursive partitioning. J Comput Graph Stat 15(3):651–674 [Online]. Available: http://pubs.amstat.org/doi/abs/10.1198/106186006X133933 Google Scholar
- 30.Hastie T, Tibshirani R, Friedman JH (2009) The elements of statistical learning, corrected ed. SpringerGoogle Scholar
- 31.Bamber D (1975) The area above the ordinal dominance graph and the area below the receiver operating characteristic graph. J Math Psych 12(4):387–415 [Online]. Available: http://www.sciencedirect.com/science/article/B6WK3-4D7JNKG-8D/2/752ed837f02a9523cda7e96258f5516c Google Scholar
- 32.Jaeger J, Sengupta R, Ruzzo W (2003) Improved gene selection for classification of microarrays. In: Proceedings of pacific symposium on biocomputing. pp 53–64Google Scholar
- 33.Boulesteix AL, Strobl C, Augustin T, Daumer M (2008) Evaluating microarray-based classifiers: an overview. Cancer Informat 6: 77–97Google Scholar
- 34.Efron B, Tibshirani RJ (1994) An introduction to the bootstrap. Chapman & Hall, New YorkGoogle Scholar
- 37.Ambroise C, McLachlan GJ (2002) Selection bias in gene extraction on the basis of microarray gene-expression data. Proc Natl Acad Sci USA. 99(10):6562–6566 [Online]. Available: doi:10.1073/pnas.102102699
- 38.Sing T, Sander O, Beerenwinkel N, Lengauer T (2005) ROCR: visualizing classifier performance in R. Bioinformatics 21(20):3940–3941 [Online]. Available: http://bioinformatics.oxfordjournals.org/cgi/content/abstract/21/20/3940 Google Scholar
- 39.Pirooznia M, Yang J, Yang MQ, Deng Y (2008) A comparative study of different machine learning methods on microarray gene expression data. BMC Genomics 9(Suppl 1):S13 [Online]. Available: http://www.biomedcentral.com/1471-2164/9/S1/S13
- 41.Slawski M, Daumer M, Boulesteix A-L (2008) Cma—a comprehensive bioconductor package for supervised classification with high dimensional data. BMC Bioinformatics 9(1):439 [Online]. Available: http://www.biomedcentral.com/1471-2105/9/439
- 43.Freund Y, Schapire RE (1995) A decision-theoretic generalization of on-line learning and an application to boosting. In: EuroCOLT ’95: Proceedings of the second European conference on computational learning theory. Springer, London, pp 23–37Google Scholar