Skip to main content
Fig. 2 | Cancer Imaging

Fig. 2

From: Automated segmentation of lesions and organs at risk on [68Ga]Ga-PSMA-11 PET/CT images using self-supervised learning with Swin UNETR

Fig. 2

Architecture of the Swin UNTER used in this study. The PET/CT input images undergo processing in the encoder, consisting of four stages, with each stage connected to the decoder through skip connections. Within the encoder, the dimensions of the images progressively decrease at each stage until reaching the bottleneck. Subsequently, in the decoder, the dimensions of the features increase as they ascend through deconvolution layers. The network guidelines are outlined in Supplementary Fig. S1

Back to article page

Cancer Imaging

ISSN: 1470-7330

Contact us