Processing Flow

Processing Flow#

High-level overview of all processing steps executed by lsp.pipeline().

Registration#

Step	Description	Outputs	Disable
Rigid Registration	Correct XY frame shifts via phase correlation	`data.bin` (registered), `ops["yoff"]`, `ops["xoff"]`	`ops["do_registration"] = 0`
Non-rigid Registration	Correct local warping via block-wise shifts	`data.bin` (registered)	`ops["nonrigid"] = False`
Compute meanImg	Average all registered frames	`ops["meanImg"]`	N/A
Compute meanImgE	Enhanced mean via spatial high-pass	`ops["meanImgE"]`	N/A

Detection Preprocessing#

These steps are applied to the registered movie before ROI detection:

Step	Description	Affects	Disable
Temporal Binning	Average consecutive frames to reduce noise	Detection movie, `ops["max_proj"]`	`ops["nbinned"] = nframes`
Temporal High-Pass	Remove slow baseline drift	Detection movie, `ops["max_proj"]`	`ops["high_pass"] = 0`*

*The mbo fork allows disabling high_pass; standard Suite2p forces a value.

What each projection image contains#

Image	Source	Binned?	Temporally HP filtered?
`meanImg`	Average of registered frames	No	No
`meanImgE`	Spatial HP of meanImg	No	No
`max_proj`	Max of detection movie	Yes	Yes

Detection: Anatomical Mode (Cellpose)#

Cellpose segments one static image, chosen by cellpose_settings.img. Suite2p 1.1.0 accepts three values:

`img`	Image	Notes
`"max_proj / meanImg"`	`log(max_proj / meanImg)`	default; log ratio emphasizes active regions
`"meanImg"`	mean image	raw mean intensity
`"max_proj"`	max projection	binned + temporally high-pass filtered

The enhanced mean image (meanImgE) is no longer a cellpose input — the old anatomical_only=3 was removed upstream.

Selecting the image#

Recommended — enable cellpose with algorithm="cellpose" and pick img:

ops = {"algorithm": "cellpose", "img": "max_proj / meanImg"}  # default
ops = {"algorithm": "cellpose", "img": "meanImg"}
ops = {"algorithm": "cellpose", "img": "max_proj"}

Legacy — the integer anatomical_only still works and maps to img:

ops = {"anatomical_only": 1}   # img = "max_proj / meanImg" (default)
ops = {"anatomical_only": 2}   # img = "meanImg"
ops = {"anatomical_only": 4}   # img = "max_proj"

anatomical_only=0 (or algorithm="sparsery"/"sourcery") uses functional detection. anatomical_only=3 warns and falls back to 1.

Step	Description	Disable
PCA Denoise	Spatial denoising via block-wise PCA	`ops["denoise"] = 0`
Spatial High-Pass	Sharpen the chosen image before cellpose	`ops["spatial_hp_cp"] = 0`
Cellpose Segmentation	Run Cellpose model on the image	N/A

Spatial high-pass: `spatial_hp_cp` (upstream `cellpose_settings.highpass_spatial`)#

Integer, default 0 (off). When > 0, the chosen img is normalized to [0, 1] and a Gaussian-blurred copy (sigma = diameter × spatial_hp_cp) is subtracted twice, removing illumination gradients and sharpening cell boundaries. It applies to whichever img you select and does not modify the stored ops["meanImg"] or ops["max_proj"].

Detection: Functional Mode#

When ops["anatomical_only"] = 0:

Step	Description	Disable	Notes
Spatial High-Pass	High-pass filter for activity detection	`ops["spatial_hp_detect"] = 0`	sparse only
Compute Vcorr	Activity correlation images	N/A	sparse only
Seed Detection	Find candidate ROI centers from peaks	N/A	sparse only
ROI Growing	Expand seeds based on correlation	N/A	sparse only
SVD Decomposition	Reduce dimensionality for source detection	N/A	sourcery only (`sparse_mode=False`)
Source Detection	Find ROIs via iterative NMF-like method	N/A	sourcery only
ROI Overlap Removal	Merge/remove overlapping ROIs	`ops["max_overlap"] = 1.0`	all functional

Post-Detection (All Modes)#

Step	Description	Disable
Extract Fluorescence (F)	Compute raw fluorescence from ROI pixels	N/A
Compute Neuropil (Fneu)	Estimate neuropil signal from surround	`ops["neuropil_extract"] = False`
Classifier	Apply ROI classifier (cell vs not-cell)	`ops["classifier_path"] = ""`
Spike Deconvolution	Infer spike times via OASIS	`ops["spikedetect"] = False`
Apply Filters	Size/shape filters on detected ROIs	`cell_filters=[]` in `lsp.pipeline()`

lsp.pipeline() Additional Steps#

These are applied by lsp.pipeline() after suite2p completes:

Step	Description	Disable
Accept All Cells	Mark all suite2p-rejected ROIs as accepted	`accept_all_cells=False` (default)
Cell Filters	Size/shape filters (default 4-35 µm diameter)	`cell_filters=[]`
Compute dF/F	Rolling percentile baseline correction	N/A
Generate Figures	Diagnostic plots for each plane	N/A