Mode
ℹ️
DBSCAN Results
Explore resulting clusters of the modified DBSCAN clustering algorithm.
Cosine Similarity
Plots the cosine similarities between a comparison vector and the voxel-wise model parameters.
If images are selected, the comparison vector is the mean of the brain-decoded CLIP embeddings of the selected images.
If the image selection is empty, the comparison vector is the model parameter vector for a particular voxel chosen by clicking the brain surface.
Model
Subject
ℹ️
Sets the currently selected NSD participant. After a voxel has been selected by clicking the brain surface, the top
representative images for that voxel are displayed in the bottom panel.
Layer
ℹ️
Sets the brain surface layer. Pial is the outermost layer, white is the boundary between grey and white matter, and
layerBX are 3 intermediate layers.
Space
ℹ️
Display the results on the fsaverage atlas surface, or the subject specific surface mesh.
Representative Images
ℹ️
Display either the positive or negative representative images for the current cluster in the right panel.
Cluster
ℹ️
Sets the currently selected cluster.
Parent Cluster
ℹ️
Click to select the parent of the currently selected cluster.
Child Clusters
ℹ️
Click to select any of the child clusters of the currently selected clusters.
No buttons are displayed if the current cluster is a leaf node in the hierarchy.