Neuroglancer

Neuroglancer is a WebGL-based viewer designed for volumetric data visualization. It supports the following visualization types:

• Volumetric Data: Arbitrary (non-axis-aligned) cross-sectional views.
• 3-D Meshes: Rendering of complex 3D surfaces.
• Line-segment based models: Visualization of skeletons or similar structures.

To access non-public Google Cloud Storage (GCS) buckets within Neuroglancer, you must use an ngauth server. This server acts as a bridge to provide authenticated access to private GCS resources.

For detailed implementation and configuration instructions, refer to the ngauth_server documentation.

The Python interface allows you to control the Neuroglancer web viewer with the following capabilities:

• Data Visualization: View in-memory NumPy arrays or other array-like types (e.g., HDF5 arrays via h5py).
• State Management: Read and write the Neuroglancer viewer state directly from Python.
• Input Customization: Change Neuroglancer key and mouse bindings.
• Python Callbacks: Define actions triggered by key or mouse bindings that invoke Python functions.
• Mesh Generation: Generate mesh representations of the surface of in-memory segmentation volumes on-demand as requested by the client.

Depending on the sharding configuration in the info file, indices are stored in one of two ways:

• Unsharded uint64 index: Data for a uint64 ID is stored in a file named <id> (the base-10 string representation of the ID) within the directory specified by the key.
• Sharded uint64 index: The uint64 ID is used directly as the key within the sharded representation within the directory specified by the key.

The #uicontrol directive allows you to add interactive widgets to the Neuroglancer UI that can be used within your shader code.

Common types include:

• color: A color picker (e.g., vec3 or vec4).
• slider: A continuous range control (e.g., float). Use min, max, step, and default parameters.
• checkbox: A boolean toggle (e.g., bool).
• invlerp: An inverse linear interpolation control for remapping annotation properties to a 0-1 range.
• uint: A discrete integer slider.

// Slider example
#uicontrol float red slider(min=0.0, max=1.0, step=0.05, default=1.0)
void main() {
  vec3 mycolor = vec3(red, 0.0, 0.0);
  setColor(mycolor);
}

// Checkbox example
#uicontrol bool showBorders checkbox(default=true)
void main() {
  if (showBorders) {
    // ... logic
  }
}

Neuroglancer supports associating a collection of property values with uint64 segment IDs (typically used for segmentation volumes, meshes, or skeletons).

Currently, only inline properties are supported. This means the entire mapping of segment IDs to their associated property values must be stored within a single info JSON file.

To use this format, your data must be organized into a directory containing exactly one info JSON file.

The precomputed:// data source format is used for static collections of files served over HTTP (e.g., Google Cloud Storage or Amazon S3). It does not require special serving infrastructure, but you must either host the Neuroglancer client on the same server as the data or enable CORS access on the data server.

Supported data types include:

• Single-resolution or multi-resolution image/segmentation volumes
• Single-resolution or multi-resolution object surface meshes (keyed by uint64 object ids)
• Object skeleton representations (keyed by uint64 object ids)
• Collections of point, line, bounding box, or ellipsoid annotations
• Segment property maps

To use a standard precomputed dataset, provide the URL to the directory containing the info metadata file using the precomputed:// scheme.

precomputed://FILE_URL

The Python API provides interfaces to manipulate the current state of the Neuroglancer viewer. This includes:

• Segment sets: Managing which segment sets are currently loaded.
• URL representation: Interacting with the state as represented in the viewer's URL.
• Coordinate space: Controlling the viewer's coordinate system.
• Layers: Managing volumetric or segmentation layers.
• Tools: Controlling active viewer tools (e.g., selection tools, measurement tools).