A sample of DSDS spleen dataset: (a) Raw image; (b) Ground truth (spleen region is indicated as green color); (c) 3D display. od Mehmet Zahid Genc (22683526)

Daljnji autori:

The architecture of the SVNC-Net. od Mehmet Zahid Genc (22683526)

Daljnji autori:

Impacts of pruning and quantization on the model performances. od Mehmet Zahid Genc (22683526)

Daljnji autori:

Qualitative comparison of spleen segmentation results on the MSD dataset: (a) Input CT slice, (b) Ground Truth (red boundary), (c) UPerNet prediction, (d) EMANet prediction, (e) CC... od Mehmet Zahid Genc (22683526)

Daljnji autori:

Segmentation accuracy of the models. od Mehmet Zahid Genc (22683526)

Daljnji autori:

The process of standard convolution and depthwise separable convolution. od Mehmet Zahid Genc (22683526)

Daljnji autori:

Training and implementation details. od Mehmet Zahid Genc (22683526)

Daljnji autori:

A sample of MSD spleen dataset: (a) Raw image; (b) Ground truth (spleen region is indicated as green color); (c) 3D display. od Mehmet Zahid Genc (22683526)

Daljnji autori:

Qualitative comparison of spleen segmentation results on the DSDS: (a) Input CT slice, (b) Ground Truth (red boundary), (c) UPerNet prediction, (d) EMANet prediction, (e) CCNet pre... od Mehmet Zahid Genc (22683526)

Daljnji autori:

Fold assignments for each evaluation round. od Mehmet Zahid Genc (22683526)

Daljnji autori:

Studies on the use of 2D CNN-based models for spleen segmentation from 3D CT images. od Mehmet Zahid Genc (22683526)

Daljnji autori:

Computational efficiency of the models. od Mehmet Zahid Genc (22683526)

Daljnji autori:

Summary of key architectural modifications and hyperparameter adjustments from the U-Net to SVNC-Net. od Mehmet Zahid Genc (22683526)

Daljnji autori: