preview_dicom¶

from voxelkit import preview_dicom

png_bytes = preview_dicom("scan.dcm")
with open("preview.png", "wb") as f:
    f.write(png_bytes)

Returns a PNG as raw bytes. For a single .dcm file you get the 2D pixel data rendered as an image. For a series directory VoxelKit assembles the 3D volume first and then extracts one slice from it.

Signature¶

def preview_dicom(
    file_path: str | Path,
    *,
    axis: int = 0,
    slice_index: int | None = None,
    as_array: bool = False,
) -> bytes | np.ndarray

Parameters¶

Parameter	Type	Default	Description
`file_path`	`str` or `Path`	required	Path to a `.dcm` file or series directory
`axis`	`int`	`0`	Which axis to slice along. Ignored for single 2D files.
`slice_index`	`int` or `None`	`None`	Index of the slice to extract. Defaults to the centre slice.
`as_array`	`bool`	`False`	When `True`, returns a `uint8` NumPy array instead of PNG bytes.

Examples¶

from voxelkit import preview_dicom

# Single .dcm file
png = preview_dicom("scan.dcm")

# Series directory, default axis (0 = slice axis)
png = preview_dicom("./series/")

# Coronal view
png = preview_dicom("./series/", axis=1)

# Pick a specific slice
png = preview_dicom("./series/", axis=0, slice_index=64)

# Get a NumPy array for embedding in your own report or GUI
arr = preview_dicom("./series/", as_array=True)
# arr.shape == (rows, cols), dtype uint8

with open("preview.png", "wb") as f:
    f.write(png)

The output image is grayscale and normalised to 0-255 regardless of the original bit depth or window level.