Synthesis of 3D MRI Brain Images With Shape and Texture Generative Adversarial Deep Neural Networks
Bibliographic record
Abstract
Generative Adversarial Networks (GAN) are emerging as an exciting training paradigm which promises a step improvement to the impressive feature learning capabilities of deep neural networks. Unlike supervised learning approaches, GAN learns generalizable features without requiring labeled images to achieve new capabilities like distinguishing previously unseen anomalies, creating novel instances of data and factorizing learned features into explainable dimensions in fully unsupervised fashion. The advanced feature learning property of GAN will enable the next generation of computational image understanding tasks. However, GAN models are difficult to train to converge towards good models, especially for high resolution and high dimensional datasets like image volumes. We develop a GAN approach to learn a generative model of T1-contrast 3D MRI image volumes of the healthy human brain by training on 1112 MRI images from the Human Connectome Project. Our method utilizes a first unconditional Super-Resolution GAN, dubbed the shape network, to learn the 3D shape variations in adult brains and a second conditional pix2pix GAN, dubbed the texture network, to upgrade image slices with realistic local contrast patterns. Novel 3D MRI images are synthesized by first applying the 3D voxel-wise deformation map which is generated from the shape network to deform the Montreal Neurological Institute (MNI) brain template and subsequently performing style transfer on axial-wise slices using the texture network. The Maximum Mean Discrepancy (MMD) and Multi-scale Structural Similarity Index Measure (MS-SSIM) scores of MRI image volumes synthesized using our GAN approach are competitive with state-of-art GAN methods. Our work establishes the feasibility of an alternative approach to high-dimensional GAN learning - splitting the type of information content learned among several GANs can be an effective form of regularization and complementary to latent code shaping or super-resolution approaches in state-of-the-art methods.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".