📋 Note: This is the project landing page, not the full documentation site. For complete guides, CLI reference, Python API docs, and real-world examples, use the Full Documentation button below.
🧊

VoxelKit

Inspect, preview, and QA-check multidimensional imaging data from one unified Python and CLI workflow.

⚡ Get started in seconds

pip install voxelkit

Requires Python ≥ 3.9  ·  MIT License  ·  Extras: pip install voxelkit[gui] for Streamlit GUI  ·  100% local — no cloud uploads

Features

Everything you need for day-to-day imaging dataset triage

🔍

File Inspection

Extract shape, dtype, and metadata from any supported format instantly — no format-specific boilerplate required.

🖼️

PNG Slice Previews

Generate crisp 2D previews from 3D/4D volumes with configurable plane, slice index, and colormap in a single command.

📊

Per-File QA Reports

Compute statistics and auto-detect problems — NaNs, Infs, constant arrays, near-zero volumes — for any single file.

📁

Batch QA Reporting

Aggregate QA across an entire directory with report-batch. Emit JSON for pipelines or a self-contained HTML report with thumbnails for sharing.

🩻

DICOM with PHI Stripping

Inspect single .dcm files or whole series directories. Patient identifiers are stripped by default; --phi opts in with a stderr warning.

🛡️

Batch Anonymisation

Scrub PHI from every .dcm under a directory tree with voxelkit anonymise. Preserves pixel data, modality, and series grouping.

🔄

DICOM → NIfTI Conversion

Convert a single slice or full series directory to NIfTI with voxelkit convert. Builds the affine from DICOM headers with the LPS → RAS flip applied.

🧬

Embedding Analysis

Specialized QA for 2D embedding matrices — detect dead dimensions, outlier samples, and per-dimension norm statistics.

🔌

Unified Python API

Four clean functions — inspect_file, preview_file, report_file, report_batch — work identically across all formats.

🌐

REST API

FastAPI-powered HTTP endpoints for remote inspection, preview generation, and QA reporting — drop into any stack.

🖥️

Optional Local GUI

Streamlit-based offline interface for point-and-click workflows. Launch in seconds with voxelkit gui.

Supported Formats

One toolkit, five imaging ecosystems

🧠 .nii  /  .nii.gz 🗄️ .h5  /  .hdf5 🔢 .npy  /  .npz 🖼️ .tif  /  .tiff 🩻 .dcm  /  DICOM series

Who Can Benefit

VoxelKit fits wherever multidimensional data needs triage

🔬

Neuroscience Researchers

Quickly inspect and QA fMRI, DTI, and structural MRI datasets in NIfTI format — no format-specific boilerplate.

🏥

Medical Imaging Teams

Run batch QA across large clinical datasets to catch constant arrays, NaN-contaminated volumes, or zero-dominated scans before analysis.

🤖

ML / AI Practitioners

Validate imaging tensors and embedding matrices before training. Catch dead dimensions and outliers early to prevent silent failures.

📐

Data Scientists

Rapidly explore unfamiliar multidimensional datasets from collaborators without reading format-specific documentation first.

🏗️

Data Engineers

Integrate VoxelKit's REST API or Python interface into automated pipelines to enforce data quality gates at ingestion time.

🌍

Geospatial Engineers

Inspect and QA multi-band raster stacks, satellite imagery tiles, and elevation models stored as NumPy, HDF5, or TIFF arrays.

🎓

Educators & Students

Use the GUI and CLI to interactively explore neuroimaging and scientific datasets in teaching and research settings.

Ready to triage your data?

The full documentation covers installation, CLI reference, Python API, QA warning types, and real-world examples.