DIPY

1.12.1 verified Fri May 01 auth: no python

DIPY is a Python library for diffusion MRI analysis, featuring tractography, reconstruction, registration, and signal processing. Current version is 1.12.1. It has a moderate release cadence with major versions every 1-2 years.

pip install dipy

Common errors

error ModuleNotFoundError: No module named 'dipy.dti' ↓

cause Import path changed after DIPY 1.0.0; old top-level submodules no longer exist.

fix

Use from dipy.reconst.dti import TensorModel instead.

error AttributeError: module 'dipy' has no attribute 'data' ↓

cause dipy.data is not automatically imported; need to import explicitly.

fix

Try from dipy.data import get_data or import dipy.data as dpd.

error ValueError: bvals and bvecs must have same number of volumes ↓

cause bvals and bvecs files have mismatched lengths or are not aligned properly.

fix

Read both files with read_bvals_bvecs and verify shapes. Ensure they correspond to the same DW images.

Warnings

breaking DIPY 1.0.0 dropped support for Python 2 and changed many submodule paths. Some previously top-level functions moved to submodules. ↓

fix Update imports: e.g., from dipy.reconst.dti import TensorModel instead of dipy.dti.TensorModel.

deprecated The `dipy.data` module contains some deprecated fetchers. Prefer using `nibabel` to load data. ↓

fix Use `nib.load()` directly instead of `dipy.data.get_data()`.

gotcha gradient_table expects b-values in s/mm² and b-vectors as (N,3) array. Mixing units or shapes causes silent errors. ↓

fix Ensure bvals are numeric, bvecs are normalized, and use correct order: gradient_table(bvals, bvecs, b0_threshold=50).

gotcha DIPY's tractography algorithms (e.g., EuDX) require specific input formats. Incorrect seed masks can lead to empty streamlines. ↓

fix Check seed mask is binary and in same space as data. Use dipy.tracking.utils seeds_from_mask.

Install

conda install -c conda-forge dipy

Imports

dipy
```
import dipy
```
Top-level import works, but specific submodules must be imported explicitly (e.g., from dipy.io import read_bvals_bvecs)

gtab

wrong

from dipy.io.gradients import read_bvals_bvecs (not for creating gtab)

correct

from dipy.core.gradients import gradient_table

gradient_table is in core.gradients

Quickstart

Minimal DTI tensor fitting example. Replace file paths with actual data.

import numpy as np
from dipy.io import read_bvals_bvecs
from dipy.core.gradients import gradient_table
from dipy.reconst.dti import TensorModel
# Load data (example: replace with own files)
# data = nib.load('dwi.nii.gz').get_fdata()
# bvals, bvecs = read_bvals_bvecs('bvals.txt', 'bvecs.txt')
# gtab = gradient_table(bvals, bvecs)
# tenmodel = TensorModel(gtab)
# tenfit = tenmodel.fit(data)
# fa = tenfit.fa
print('DIPY DTI example ready')