Skip to main content

GSplit LBI: Taming the Procedural Bias in Neuroimaging for Disease Prediction

Part of the Lecture Notes in Computer Science book series (LNIP,volume 10435)


In voxel-based neuroimage analysis, lesion features have been the main focus in disease prediction due to their interpretability with respect to the related diseases. However, we observe that there exist another type of features introduced during the preprocessing steps and we call them “Procedural Bias”. Besides, such bias can be leveraged to improve classification accuracy. Nevertheless, most existing models suffer from either under-fit without considering procedural bias or poor interpretability without differentiating such bias from lesion ones. In this paper, a novel dual-task algorithm namely GSplit LBI is proposed to resolve this problem. By introducing an augmented variable enforced to be structural sparsity with a variable splitting term, the estimators for prediction and selecting lesion features can be optimized separately and mutually monitored by each other following an iterative scheme. Empirical experiments have been evaluated on the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database. The advantage of proposed model is verified by improved stability of selected lesion features and better classification results.


  • Voxel-based structural magnetic resonance imaging
  • Procedural bias
  • Split Linearized Bregman Iteration
  • Feature selection

This is a preview of subscription content, access via your institution.

Fig. 1.
Fig. 2.
Fig. 3.
Fig. 4.


  1. 1.

    Here \(D_G:\mathbb {R}^V \rightarrow \mathbb {R}^E\) denotes a graph difference operator on \(G=(V,E)\), where V is the node set of voxels, E is the edge set of voxel pairs in neighbour (e.g. 3-by-3-by-3), such that \(D_G(\beta )(i,j):=\beta (i)-\beta (j)\).

  2. 2.

  3. 3.

    For logit model, \(\alpha < \nu / \kappa (1 + \nu \Lambda _{H}^2 + \nu \Lambda _{X}^2)\) since \(\Lambda _{X} > \Lambda _{H}\).

  4. 4.

    In this experiment, comparable prediction result will be given for \(\nu \in (0.1,10)\).

  5. 5.

    0 corresponds to logistic regression model.

  6. 6.

    In [12], \(mDC := \frac{10 | \cap _{k=1}^{10} S(k) | }{\sum _{k=1}^{10} | S(k) |}\) where S(k) denotes the support set of \(\beta _{les}\) in k-th fold.


  1. Ashburner, J.: A fast diffeomorphic image registration algorithm. Neuroimage 38(1), 95–113 (2007)

    CrossRef  Google Scholar 

  2. Ashburner, J., Friston, K.J.: Why voxel-based morphometry should be used. Neuroimage 14(6), 1238–1243 (2001)

    CrossRef  Google Scholar 

  3. Dai, Z., Yan, C., Wang, Z., Wang, J., Xia, M., Li, K., He, Y.: Discriminative analysis of early alzheimer’s disease using multi-modal imaging and multi-level characterization with multi-classifier. Neuroimage 59(3), 2187–2195 (2012)

    CrossRef  Google Scholar 

  4. Dice, L.R.: Measures of the amount of ecologic association between species. Ecology 26(3), 297–302 (1945)

    CrossRef  Google Scholar 

  5. Grosenick, L., Klingenberg, B., Katovich, K., Knutson, B., Taylor, J.E.: Interpretable whole-brain prediction analysis with graphnet. Neuroimage 72, 304–321 (2013)

    CrossRef  Google Scholar 

  6. Huang, C., Sun, X., Xiong, J., Yao, Y.: Split lbi: An iterative regularization path with structural sparsity. In: Advances In Neural Information Processing Systems, pp. 3369–3377 (2016)

    Google Scholar 

  7. Osher, S., Ruan, F., Xiong, J., Yao, Y., Yin, W.: Sparse recovery via differential inclusions. Appl. Comput. Harmonic Anal. 41(2), 436–469 (2016)

    MathSciNet  CrossRef  Google Scholar 

  8. Peng, J., An, L., Zhu, X., Jin, Y., Shen, D.: Structured sparse kernel learning for imaging genetics based alzheimer’s disease diagnosis. In: International Conference on Medical Image Computing and Computer-Assisted Intervention, pp. 70–78 (2016)

    CrossRef  Google Scholar 

  9. Rudin, L.I., Osher, S., Fatemi, E.: Nonlinear total variation based noise removal algorithms. Physica D 60(1–4), 259–268 (1992)

    MathSciNet  CrossRef  MATH  Google Scholar 

  10. Tibshirani, R.: Regression shrinkage and selection via the lasso. J. Roy. Stat. Soc.: Ser. B (Methodol.) 58, 267–288 (1996)

    MathSciNet  MATH  Google Scholar 

  11. Tibshirani, R.J., Taylor, J.E., Candes, E.J., Hastie, T.: The solution path of the generalized lasso. Ann. Stat. 39(3), 1335–1371 (2011)

    MathSciNet  CrossRef  MATH  Google Scholar 

  12. Xin, B., Hu, L., Wang, Y., Gao, W.: Stable feature selection from brain smri. In: AAAI, pp. 1910–1916 (2014)

    Google Scholar 

  13. Zou, H., Hastie, T.: Regularization and variable selection via the elastic net. J. Roy. Stat. Soc. Ser. B (Statistical Methodology) 67(2), 301–320 (2005)

    MathSciNet  CrossRef  MATH  Google Scholar 

Download references


This work was supported in part by 973-2015CB351800, 2015CB85600, 2012CB825501, NSFC-61625201, 61370004, 11421110001 and Scientific Research Common Program of Beijing Municipal Commission of Education (No. KM201610025013).

Author information

Authors and Affiliations


Corresponding authors

Correspondence to Lingjing Hu or Yuan Yao .

Editor information

Editors and Affiliations

1 Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary material 1 (pdf 802 KB)

Rights and permissions

Reprints and Permissions

Copyright information

© 2017 Springer International Publishing AG

About this paper

Cite this paper

Sun, X., Hu, L., Yao, Y., Wang, Y. (2017). GSplit LBI: Taming the Procedural Bias in Neuroimaging for Disease Prediction. In: Descoteaux, M., Maier-Hein, L., Franz, A., Jannin, P., Collins, D., Duchesne, S. (eds) Medical Image Computing and Computer Assisted Intervention − MICCAI 2017. MICCAI 2017. Lecture Notes in Computer Science(), vol 10435. Springer, Cham.

Download citation

  • DOI:

  • Published:

  • Publisher Name: Springer, Cham

  • Print ISBN: 978-3-319-66178-0

  • Online ISBN: 978-3-319-66179-7

  • eBook Packages: Computer ScienceComputer Science (R0)