GFCN-8s, the proposed geometric deep learning network with three pooling layers and eight-fold upsampling, can extract rich geometric information from computed tomography scans of the brain. (IMAGE)

SPIE--International Society for Optics and Photonics

GFCN-8s, the proposed geometric deep learning network with three pooling layers and eight-fold upsampling, can extract rich geometric information from computed tomography scans of the brain.

Caption

GFCN-8s, the proposed geometric deep learning network with three pooling layers and eight-fold upsampling, can extract rich geometric information from computed tomography scans of the brain. This enables it to outperform the state-of-the-art models (U-Net, FCN-8s, and PointNet++) in detecting stroke lesions.

Credit

Iporre-Rivas et al., doi 10.1117/1.JMI.10.4.044502.

Usage Restrictions

For media use.

License

Original content

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.