Variable Importance in Nonlinear Kernels (VINK): Classification of Digitized Histopathology

详细信息    查看全文

• 作者：Shoshana Ginsburg ; Sahirzeeshan Ali ; George Lee ; Ajay Basavanhally ; Anant Madabhushi
• 刊名：Lecture Notes in Computer Science
• 出版年：2013
• 出版时间：2013
• 年：2013
• 卷：8150
• 期：1
• 页码：246-253
• 全文大小：2132KB
• 参考文献：1. Saeys, Y., Inza, I., Larranaga, P.: A Review of Feature Selection Techniques in Bioinformatics. Bioinformatics?23, 2507-517 (2007) p://dx.doi.org/10.1093/bioinformatics/btm344">CrossRef
  2. Yan, H., et al.: Correntropy Based Feature Selection using Binary Projection. Pattern Recogn.?44, 2834-842 (2011) p://dx.doi.org/10.1016/j.patcog.2011.04.014">CrossRef
  3. Ham, J., et al.: A Kernel View of the Dimensionality Reduction of Manifolds. Max Planck Institute for Biological Cybernetics, Technical Report No. TR-110 (2002)
  4. Shi, J., Luo, Z.: Nonlinear Dimensionality Reduction of Gene Expression Data for Visualization and Clustering Analysis of Cancer Tissue Samples. Computers Biol. Med.?40, 723-32 (2010) p://dx.doi.org/10.1016/j.compbiomed.2010.06.007">CrossRef
  5. Ginsburg, S., Tiwari, P., Kurhanewicz, J., Madabhushi, A.: Variable Ranking with PCA: Finding Multiparametric MR Imaging Markers for Prostate Cancer Diagnosis and Grading. In: Madabhushi, A., Dowling, J., Huisman, H., Barratt, D. (eds.) Prostate Cancer Imaging 2011. LNCS, vol.?6963, pp. 146-57. Springer, Heidelberg (2011) p://dx.doi.org/10.1007/978-3-642-23944-1_15">CrossRef
  6. Tenenbaum, J.B., de Silva, V., Langford, J.C.: A Global Geometric Framework for Nonlinear Dimensionality Reduction. Science?290, 2319-323 (2000) p://dx.doi.org/10.1126/science.290.5500.2319">CrossRef
  7. Belkin, M., Niyogi, P.: Laplacian Eigenmaps for Dimensionality Reduction and Data Representation. Neural Comput.?15, 1373-396 (2003) p://dx.doi.org/10.1162/089976603321780317">CrossRef
  8. Esbensen, K.: Multivariate Data Analysis—In Practice: An Introduction to Multivariate Data Analysis and Experimental Design. CAMO, Norway (2004)
  9. Chong, I.G., Jun, C.H.: Performance of Some Variable Selection Methods when Multicollinearity is Present. Chemometr. Intell. Lab?78, 103-12 (2005) p://dx.doi.org/10.1016/j.chemolab.2004.12.011">CrossRef
  10. Golugula, A., et al.: Supervised Regularized Canonical Correlation Analysis: Integrating Histologic and Proteomic Measurements for Predicting Biochemical Recurrence Following Prostate Surgery. BMC Bioinformatics?12, 483-95 (2011) p://dx.doi.org/10.1186/1471-2105-12-483">CrossRef
  11. Basavanhally, A., et al.: Multi–Field–of–View Framework for Distinguishing Tumor Grade in ER+ Breast Cancer from Entire Histopathology Slides. IEEE Trans. Biomed. Eng. (Epub ahead of print) (PMID: 23392336)
  12. Ali, S., et al.: Cell Cluster Graph for Prediction of Biochemical Recurrence in Prostate Cancer Patients from Tissue Microarrays. In: Proc. SPIE Medical Imaging: Digital Pathology (2013)
  
• 作者单位：Shoshana Ginsburg (21)
  Sahirzeeshan Ali (21)
  George Lee (22)
  Ajay Basavanhally (22)
  Anant Madabhushi (21)
  
  21. Department of Biomedical Engineering, Case Western Reserve University, USA
  22. Department of Biomedical Engineering, Rutgers University, USA
  
• ISSN：1611-3349

文摘

Quantitative histomorphometry is the process of modeling appearance of disease morphology on digitized histopathology images via image–based features (e.g., texture, graphs). Due to the curse of dimensionality, building classifiers with large numbers of features requires feature selection (which may require a large training set) or dimensionality reduction (DR). DR methods map the original high–dimensional features in terms of eigenvectors and eigenvalues, which limits the potential for feature transparency or interpretability. Although methods exist for variable selection and ranking on embeddings obtained via linear DR schemes (e.g., principal components analysis (PCA)), similar methods do not yet exist for nonlinear DR (NLDR) methods. In this work we present a simple yet elegant method for approximating the mapping between the data in the original feature space and the transformed data in the kernel PCA (KPCA) embedding space; this mapping provides the basis for quantification of variable importance in nonlinear kernels (VINK). We show how VINK can be implemented in conjunction with the popular Isomap and Laplacian eigenmap algorithms. VINK is evaluated in the contexts of three different problems in digital pathology: (1) predicting five year PSA failure following radical prostatectomy, (2) predicting Oncotype DX recurrence risk scores for ER+ breast cancers, and (3) distinguishing good and poor outcome p16+ oropharyngeal tumors. We demonstrate that subsets of features identified by VINK provide similar or better classification or regression performance compared to the original high dimensional feature sets.