Augment to Augment: Diverse Augmentations Enable Competitive Ultra-Low-Field MRI Enhancement
This work addresses the challenge of making MRI more accessible by improving image quality for medical applications, though it is incremental as it builds on existing image-to-image translation methods with enhanced augmentations.
The paper tackled the problem of enhancing ultra-low-field MRI images, which suffer from low signal-to-noise ratio and reduced resolution, by using diverse data augmentations to improve a deep learning model's performance despite scarce paired training data. The result was a third-place ranking in brain-masked SSIM on a validation leaderboard and fourth place on the final test leaderboard.
Ultra-low-field (ULF) MRI promises broader accessibility but suffers from low signal-to-noise ratio (SNR), reduced spatial resolution, and contrasts that deviate from high-field standards. Image-to-image translation can map ULF images to a high-field appearance, yet efficacy is limited by scarce paired training data. Working within the ULF-EnC challenge constraints (50 paired 3D volumes; no external data), we study how task-adapted data augmentations impact a standard deep model for ULF image enhancement. We show that strong, diverse augmentations, including auxiliary tasks on high-field data, substantially improve fidelity. Our submission ranked third by brain-masked SSIM on the public validation leaderboard and fourth by the official score on the final test leaderboard. Code is available at https://github.com/fzimmermann89/low-field-enhancement.