Expand description
§DICOM-volume library
This crate serves a high-level API for handling multiple DICOM files as volumes
This library is part of the dicom-rs ecosystem and leverages its component
to provide a volume representation of multiple DICOM files.
Volumes can either be loaded from multiple FileDicomObject<InMemDicomObject> or from a
specified folder where each “.dcm” file is read from.
If the environment supports it the DICOM files are loaded in parallel
using rayon. The volume can be sliced in the three different medical axes:
- Axial
- Coronal
- Sagittal
Library consumers can chose whether the Coronal and Sagittal slices should be interpolated to preserve the aspect ratios between of the images. DICOM files are assumed to have the following attributes:
- Axial data set (Only Coronal and Sagittal axes are interpolated)
- No multiframe (always the first frame is used)
- Images from the same series (Series Instance UID) and acquisition (Acquisition Number)
Contributions are highly welcome!
§Roadmap
- GPU processor for interpolation using WGPU and compute shaders
- Trilinear interpolation
- Cubic interpolation
- Caching of images
§Examples
§Reading multiple DICOM files into a volume
To read all DICOM files from the dicom/ directory, sort them by InstanceNumber. Then get the image at the center of the volume in the Sagittal axis.
let volume = VolumeLoader::load_from_directory(&PathBuf::from("dicom"), SortBy::InstanceNumber)
.expect("should have loaded files from directory");
let image = volume
.get_image_from_axis(
volume.dim().2 / 2,
Orientation::Sagittal,
Interpolation::Bilinear(Processor::CPU),
)
.expect("should have returned image at center of volume");
image.save("result.png");