Learning Conditional Deformable Templates with Convolutional Networks

Authors: Adrian Dalca, Marianne Rakic, John Guttag, Mert Sabuncu

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate the usefulness of this method on a variety of domains, with a special focus on neuroimaging. This is particularly useful for clinical applications where a pre-existing template does not exist, or creating a new one with traditional methods can be prohibitively expensive. Our code and atlases are available online as part of the Voxel Morph library at http://voxelmorph.csail.mit.edu. We present two main sets of experiments. The first set uses image-based datasets MNIST and Google Quick Draw, with the goal of providing a picture of the capabilities of our method. While deformable templates in these data are not a real-world application, these are often-studied datasets that provide a platform to analyze aspects of deformable templates. In contrast, the second set of experiments is designed to demonstrate the utility of our method on a task of practical importance, analysis of brain MRI.
Researcher Affiliation Academia Adrian V. Dalca CSAIL, MIT MGH, HMS adalca@mit.edu Marianne Rakic D-ITET, ETH CSAIL, MIT mrakic@mit.edu John Guttag CSAIL, MIT guttag@mit.edu Mert R. Sabuncu ECE and BME, Cornell msabuncu@cornell.edu
Pseudocode No The paper describes the probabilistic model and network architecture in detail with equations and diagrams (Figure 2), but it does not include any explicit pseudocode or algorithm blocks.
Open Source Code Yes Our code and atlases are available online as part of the Voxel Morph library at http://voxelmorph.csail.mit.edu.
Open Datasets Yes We use the MNIST dataset, consisting of small 2D images of hand-written digits [49] and 11 classes from the Google Quick Draw dataset [39]. We use a large dataset of 7829 T1-weighted 3D brain MRI scans from publicly available datasets: ADNI [58], OASIS [52], ABIDE [24], ADHD200 [54], MCIC [29], PPMI [53], HABS [16], and Harvard GSP [34].
Dataset Splits Yes The datasets are split into train, validation and test sets. The dataset is split into 7329 training volumes, 250 validation and 250 test.
Hardware Specification Yes Training the model requires approximately a day on a Titan XP GPU.
Software Dependencies No The paper mentions software components like 'convolutional neural networks', 'U-Net like architecture', and references 'Free Surfer [27]' for pre-processing. However, it does not provide specific version numbers for any of these software dependencies.
Experiment Setup Yes Hyperparameters. Model hyperparameters have intuitive effects on the sharpness of templates, the spatial smoothness of registration fields, and the quality of alignments. In practical settings, they should be chosen based on the desired goal of a given task. In these experiments, we tune hyperparameters by visually assessing deformations on validation data, starting from γ = 0.01, λd = 0.001, λa = 0.01, and σ = 1 for training on the D-class data.