Registration

1. Registration#

Note

The terms motion-correction and registration are often used interchangeably. Similarly, non-rigid and piecewise-rigid are often used interchangeably. Here, piecewise-rigid registration is the method used to correct for non-rigid motion.

1.1. Overview#

We use suite2p:registration to ensure spatial alignment across all frames in a movie. This means a neuron that appears in one location in frame 0 remains in the same spot in frame N.

Registration uses either a single-channel movie (default: channel 1), or an alternate channel if ops['align_by_chan'] = 2.

The result of registration is a binary file (data.bin) aligned to a reference frame. Suite2p creates this reference frame by selecting the top correlated subset of frames and refining their alignment iteratively. You can customize this via ops['nimg_init'].

from suite2p.registration import register
refImg = register.pick_initial_reference(ops)

Use ops['refImg'] to inspect the reference if needed.

1.2. Rigid and Non-Rigid Registration#

Suite2p first runs rigid registration (frame-wide shifts using phase correlation), followed by optional non-rigid registration (local shifts in blocks).

Set ops['nonrigid'] = True to enable block-based correction.

1.2.1. Key Parameters#

Parameter

Description

align_by_chan

Channel to align by (1 or 2).

nonrigid

Whether to apply non-rigid registration.

block_size

Size of blocks in [Y, X] for nonrigid. Default [128, 128].

maxregshift

Maximum fraction of FOV allowed to shift (rigid).

maxregshiftNR

Max shift in pixels for non-rigid registration.

smooth_sigma

Gaussian smoothing sigma. Improves template clarity.

batch_size

Number of frames per registration batch.

Recommended tuning

  • Increase nimg_init if your template looks noisy or blurry

  • If there is remaining motion after rigid, try nonrigid = True

  • If registration looks unstable, try decreasing maxregshift or maxregshiftNR

1.3. Output Files#

  • data.bin: motion-corrected movie (channel 1)

  • data_chan2.bin: registered channel 2 (if nchannels = 2)

  • ops['xoff'], ops['yoff']: rigid shifts for each frame

  • ops['corrXY']: phase-correlation scores frame vs reference

1.4. Metrics#

Suite2p computes PC-based registration metrics via registration:

from suite2p.registration import metrics
ops = metrics.get_pc_metrics(ops, use_red=False)

Metric

Description

regPC

Spatial principal components used to compute metrics.

tPC

Time courses of principal components used for alignment QC.

regDX

Shift distance between split PCs. Lower values = better registration.

You can visualize these with Suite2p’s GUI or custom plotting tools. Even with good metrics, always visually inspect registered frames.

Important

Visual inspection trumps all. Even if regDX is low, double check the video. And if it looks good but metrics are high, trust your eyes.

1.5. CLI for Registration Metrics#

You can also use the command-line:

reg_metrics path/to/data --nplanes 1 --smooth_sigma 1.2

To see available CLI options:

reg_metrics --help

Example output:

Avg_Rigid: 0.000000     Avg_Average NR: 0.028889        Avg_Max NR: 0.120000
Max_Rigid: 0.000000     Max_Average NR: 0.044444        Max_Max NR: 0.200000

These values are from ops['regDX'], and should be close to zero in well-registered data.

Contents