Multimodal 3D Brain Tumor Segmentation with Adversarial Training and Conditional Random Field

TY - UNPB

T1 - Multimodal 3D Brain Tumor Segmentation with Adversarial Training and Conditional Random Field

AU - Jiang, Lan

AU - Zheng, Yuchao

AU - Yu, Miao

AU - Zhang, Haiqing

AU - Aladwani, Fatemah

AU - Perelli, Alessandro

PY - 2024/11/21

Y1 - 2024/11/21

N2 - Accurate brain tumor segmentation remains a challenging task due to structural complexity and great individual differences of gliomas. Leveraging the pre-eminent detail resilience of CRF and spatial feature extraction capacity of V-net, we propose a multimodal 3D Volume Generative Adversarial Network (3D-vGAN) for precise segmentation. The model utilizes Pseudo-3D for V-net improvement, adds conditional random field after generator and use original image as supplemental guidance. Results, using the BraTS-2018 [12] dataset, show that 3D-vGAN outperforms classical segmentation models, including U-net, Gan, FCN and 3D V-net, reaching specificity over 99.8%.

AB - Accurate brain tumor segmentation remains a challenging task due to structural complexity and great individual differences of gliomas. Leveraging the pre-eminent detail resilience of CRF and spatial feature extraction capacity of V-net, we propose a multimodal 3D Volume Generative Adversarial Network (3D-vGAN) for precise segmentation. The model utilizes Pseudo-3D for V-net improvement, adds conditional random field after generator and use original image as supplemental guidance. Results, using the BraTS-2018 [12] dataset, show that 3D-vGAN outperforms classical segmentation models, including U-net, Gan, FCN and 3D V-net, reaching specificity over 99.8%.

KW - Multimodal Segmentation

KW - Generative Adversarial Network

KW - Brain tumor

U2 - 10.48550/arXiv.2411.14418

DO - 10.48550/arXiv.2411.14418

M3 - Preprint

BT - Multimodal 3D Brain Tumor Segmentation with Adversarial Training and Conditional Random Field

PB - arXiv

ER -