Skip to main content
. 2024 Jun 28;55:101021. doi: 10.1016/j.neo.2024.101021

Fig. 1.

Fig 1

CUP-classifier model workflow. The beta values of non-squamous cancer-specific DMCs and squamous cancer-specific DMCs were converted into 68×34 and 30×15 array formats, respectively, and were followed by conversion to IGTD image. The squamous part image was resized to 68×34 and merged with non-squamous part image. Merged images were divided into 17×17 image patches and these patches were used to train the ViT model for classification of CUP.