Dynamic tuned GAN improves brain tumor segmentation accuracy
De-GAN - Dynamic Parameter Tuned GAN for 3D Medical Image Segmentation: A Step Towards Generalisation
Computer Vision and Pattern Recognition
Summary
Brain tumor segmentation, especially for parts of the tumor that look very similar to nearby tissue, is difficult because medical images vary a lot. The authors propose DE-GAN, a special AI tool that creates enhanced MRI images by adjusting itself dynamically based on the input. These improved images help a 3D brain scan analysis tool better identify and separate tumor parts. Testing on several tumor image collections showed DE-GAN generally improved tumor detection compared to previous methods.
What this means in practice
- •For medical imaging teams: Enhance MRI image quality dynamically to improve tumor region segmentation across different scanners and sites in brain tumor analysis.
- •For healthcare ai developers: Integrate adaptive GAN-generated images to increase accuracy of 3D convolutional networks in detecting subtle tumor features.
Authors
Zoha Usama, Azadeh Alavi
Abstract
Brain tumor segmentation remains difficult because enhancing tumor (ET) has low contrast and overlaps surrounding tissue, while scanner and site variation causes domain shift. We propose DE-GAN, a contrast-enhancing conditional GAN that combines input-adaptive dynamic convolutions, style-aware feature mixing, and coordinate encoding to synthesize slice-adaptive FLAIR images. A label-guided, class-conditional target separates tumor-core (TC) and ET intensities while preserving anatomy. The generated FLAIR is concatenated with the original MR modalities and used to train a 3D U-Net. Across BraTS 2015, 2018, and 2019, DE-GAN improves segmentation over the baseline and static EnhGAN replacement on most reported TC/ET metrics, with the largest gains from retaining both original and enhanced FLAIR. Code and pretrained models are available at https://github.com/zkhansuri-ui/DE-GAN.