Skip to main content
Fig. 2 | Radiation Oncology

Fig. 2

From: Pretreatment patient-specific quality assurance prediction based on 1D complexity metrics and 3D planning dose: classification, gamma passing rates, and DVH metrics

Fig. 2

Network architecture utilized in this study. (a) The overall architecture of the combined model. (b) Reprocess block of the input image, including rescale and a convolution layer for all backbones. (c) The simplified Swin-transformer block of 3D version, consisting of two successive layers. (d), (e) The 3D residual blocks and U-net Encoder utilized in this study for comparison. The number of n or 2n means output channels of the convolution layer and n is the input channel of the block, the s means stride of the convolution kernel or pooling kernel. Norm: Normalization layer, in (b), (d) and (e) means Batch Normalization, in (c) means Layer Normalization. MSA: Multi-head Self-attention. MLP: Multilayer Perceptron

Back to article page

Radiation Oncology

ISSN: 1748-717X

Contact us