RodentLimb

English

简体中文

English

简体中文

Appearance

RodentLimbLimb pose for motor and behavior tests, on iPhone

Track four paws, nose and tail base in every frame, then score tail suspension, cylinder and gait tests. It all runs on the phone, and nothing is uploaded.

Get started
Assays
Source on GitHub
中文版
RodentLimb app icon

Seven keypoints, every frame

A pose model trained on mouse videos finds the four paws, tail base, nose and mouth in every video frame, on the iPhone itself.

Pose model

Tail suspension

Immobility time and %, latency, bouts, per-minute profile, tail climbing, and an estimated hindlimb clasping score.

Tail suspension

Cylinder test

Left, right and both-paw wall contacts during rearing, with lesion-relative asymmetry. You can replay and relabel every contact.

Cylinder test

Gait analysis

Stance, swing, stride, duty factor, base of support, interlimb phase, gait pattern and whole-body sway. Values are also reported at a common reference speed.

Gait analysis

Change a parameter, see it instantly

The model runs once per video. Thresholds, windows and calibration can be changed afterwards, and results update immediately.

Parameters

Private and open source

No account, no analytics, no network connection. Exports are in English with a methods paragraph you can cite. The source code is public.

Source code

Language: English · 简体中文

How it works ​

The model reads each video once and stores the keypoints. Everything after that is quick: cleaning, the assay analysis and the results are recomputed whenever you change a parameter or a label.

From video to results

Assays ​

AssayMain measuresCamera
Tail suspensionImmobility (s, %), latency, bouts, clasping score, tail climbingFront or back of the hanging mouse
Cylinder testLeft / right / both wall contacts, contralateral use, asymmetry indexBelow the clear floor (or side view)
Gait analysisStride, stance, swing, duty factor, base of support, phase, pattern, swayBelow a clear walkway, ≥ 60 fps

Open source ​

RodentLimb is open source under the GNU AGPL-3.0. The analysis code, every metric definition and this manual are in one public repository, so you can read exactly how each number is computed.

github.com/seesaw-earth/RodentLimb

Where to start ​

Quick startFrom a video to your first result in a few minutes.Recording tipsCamera position, light and frame rate for good paw tracking.Assay referenceWhat each assay measures and how every value is defined.ValidationWhat has been checked against ground truth, and what you should check.