A Stacked Generalization of 3D Orthogonal Deep Learning Convolutional Neural Networks for Improved Detection of White Matter Hyperintensities in 3D FLAIR Images

Introduction

Accurate and reliable detection of white matter hyperintensities (WMH) and its volume quantification can provide valuable clinical information to assess neurological disease progression. In this work, a stacked generalization ensemble of orthogonal 3D Convolutional Neural Networks (CNNs), StackGen-Net, is explored for improving automated detection of WMH in 3D T2-FLAIR images.

Architecture

Figure shows the stacked generalization ensemble framework (StackGen-Net) used in this work. Individual CNNs (DeepUNET3D) in StackGen-Net were trained on 2.5D patches from orthogonal reformatting of 3D-FLAIR volumes in the training set to yield WMH posteriors. A Meta-CNN was trained to learn the functional mapping from orthogonal WMH posteriors to the final WMH prediction. Additional model architecture and training details are available in the manuscript referenced in the Citation section.

Predictions

Following assumptions are made regarding the test FLAIR volume:

1. Volumes are 3D (approximately isotropic resolution to facilitate reformatting without interpolation).
2. Volumes are in nifti format.
3. Volumes are pre-processed (brain extraction and N4 bias correction)
4. Volumes are oriented axially.

Figure shows axial and sagittal views of WMH mask predicted by StackGen-Net on a test FLAIR volume. Manual annotations are shown for reference.

Description

The following are available in this repository

1. Pretrained Orthogonal CNN models for predicting WMHs on orthogonal reformatting of 3D FLAIR volumes
2. An evaluation script for testing new 3D FLAIR volume
3. Utilities

Environment

Tensorflow 1.4 python 3.5 keras 2.2 nibabel 2.3

Citation

If you use this CNN model in your work, please cite the following: Umapathy, L., G. G. Perez-Carrillo, M. B. Keerthivasan, J. A. Rosado-Toro, M. I. Altbach, B. Winegar, C. Weinkauf, A. Bilgin, and Alzheimer’s Disease Neuroimaging Initiative. "A Stacked Generalization of 3D Orthogonal Deep Learning Convolutional Neural Networks for Improved Detection of White Matter Hyperintensities in 3D FLAIR Images." American Journal of Neuroradiology, 2021 (http://doi.org/10.3174/ajnr.A6970).

Remarks: RESEARCH USE ONLY, NOT APPROVED FOR CLINICAL USE

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.