1. Drop the study
Individual .dcm files, a whole folder, or a .zip archive. Headers are read first, so nothing large is loaded yet.
Free · private · in your browser
DICOM to NIfTI converter
Turn a CT, MRI or PET study into .nii.gz volumes with a correct affine, ready for FSL, SPM, 3D Slicer or your own pipeline. The conversion runs on your computer. Your scan is never uploaded.
Drop your DICOM files, study folder or .zip here
Nothing is uploaded. Files are read and converted on this computer.
100% local: your files are read and converted in your browser and never uploaded.
How it works
Three steps, all of them on your own machine.
2. Check the series
Each series is listed with its dimensions, voxel spacing and data type. Pick the ones you want.
3. Save the volumes
One .nii.gz per series, written out as it is produced. Several series come back as a .zip holding a folder of volumes.
Format
Why convert to NIfTI
DICOM stores a study as hundreds of separate files, each one slice with a large header. NIfTI stores the same acquisition as a single 3D array plus a compact header describing where it sits in space. Almost every neuroimaging and research tool expects the latter.
Research pipelines
FSL, SPM, AFNI, FreeSurfer, ANTs and nipype all read NIfTI natively. It is the format registration, segmentation and statistics tools are written against.
3D visualisation
3D Slicer, ITK-SNAP and MRIcroGL open a .nii.gz as one volume, so you can reslice, render and segment without importing a folder of files.
Machine learning
nnU-Net, MONAI and TorchIO expect NIfTI. A single compressed array per scan is far easier to version, ship and load than a directory of slices.
Privacy by design
Your scan never leaves your device
Parsing, decoding and writing all happen locally in your browser. Your files are never uploaded to DocOrbit or any third party, and nothing is stored on a server. If you later request a second opinion, your imaging is encrypted in transit and at rest and isnever used to train AI models.
No upload
Files are opened directly by your browser. They never touch our servers.
No account required
No sign-up, no login, no email, no limit on how many studies you convert.
Works offline
Once the page has loaded you can disconnect entirely and it still converts.
DICOM to NIfTI FAQ
Is my scan uploaded anywhere?
No. Every part of the conversion happens on your own computer: your browser reads the files, decodes the pixel data and writes the NIfTI. Nothing is sent to DocOrbit or any third party, and nothing is stored on a server. You can disconnect from the internet after the page loads and it will still work.
Can it handle a whole study or a huge series?
Yes, and that is what it is built for. The converter reads each file's header first to work out the slice order, then streams the volume out one slice at a time into a compressed file. It never assembles the whole volume in memory, so a multi-gigabyte series converts in roughly the memory a single slice takes. In Chrome and Edge the output is written straight to the file you choose as it is produced.
Which DICOM files are supported?
CT, MRI, PET, and any other image series. Uncompressed DICOM, RLE, JPEG baseline and lossless, JPEG-LS, JPEG 2000 and HTJ2K are all decoded. Drop individual .dcm files, a whole study folder, or a .zip archive; series are separated automatically by SeriesInstanceUID.
Is the orientation correct?
Yes. The affine is built from ImageOrientationPatient, ImagePositionPatient and PixelSpacing and written to both the sform and qform, converted from DICOM's LPS convention to NIfTI's RAS. Slice spacing comes from the measured distance between consecutive slice positions rather than SliceThickness, which is wrong whenever an acquisition has a gap or overlap. Volumes load correctly oriented in FSL, SPM and 3D Slicer.
What happens to Hounsfield units and rescale values?
Stored pixel values are written unchanged and the RescaleSlope and RescaleIntercept are carried in the NIfTI header's scl_slope and scl_inter fields, which is what dcm2niix does. Your analysis tool applies them and you get exact Hounsfield units. If slices within a series disagree on the rescale, the volume is written as float32 with the values already applied instead.
What about diffusion or multi-echo series?
A 4D series stores several volumes at the same slice positions. The converter detects the repeated positions, writes the first volume and tells you it did. It does not produce bvec/bval files, so for diffusion pipelines dcm2niix remains the right tool.
Can I convert a .zip of a study?
Yes. Drop the .zip and it is read in place: the archive index is parsed and each file inside is decompressed only when it is needed, one at a time. The archive is never extracted, so a zipped study uses no more memory than a folder of loose files. Password-protected archives are not supported.
Can I get a radiologist to read the scan?
Yes. DocOrbit provides independent, AI-supported second opinions on radiology and nuclear medicine imaging, verified by a board-certified, subspecialty-matched radiologist, typically within 24 to 48 hours.
More free tools
All of them run in your browser. Nothing is uploaded.
Get a second read
Concerned about a finding?
Get an independent, AI-supported second opinion from a board-certified radiologist, with results in 24 to 48 hours.