Initial release: 5 MI sessions, 5 calibrator recordings, loader, reports

.gitattributes

@@ -58,3 +58,11 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 # Video files - compressed
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
+data/mi/cap32_20260722_135310_raw.fif filter=lfs diff=lfs merge=lfs -text
+data/mi/cap32_20260725_135441_mi_raw.fif filter=lfs diff=lfs merge=lfs -text
+data/mi/cap32_20260725_143756_hands-rest_raw.fif filter=lfs diff=lfs merge=lfs -text
+data/mi/cap32_20260725_163251_hands-feet-math_raw.fif filter=lfs diff=lfs merge=lfs -text
+docs/crosstalk_report.pdf filter=lfs diff=lfs merge=lfs -text
+docs/crosstalk_report_zh.pdf filter=lfs diff=lfs merge=lfs -text
+docs/hardware_acceptance.pdf filter=lfs diff=lfs merge=lfs -text
+docs/mi_pilot_report.pdf filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,247 @@
+---
+license: cc-by-4.0
+language:
+  - en
+  - zh
+tags:
+  - eeg
+  - motor-imagery
+  - brain-computer-interface
+  - bci
+  - dry-electrode
+  - ads1299
+  - biosignal
+pretty_name: "cap32 — 32-channel dry-electrode motor imagery EEG"
+size_categories:
+  - n<1K
+task_categories:
+  - time-series-forecasting
+---
+
+# cap32 — 32-channel dry-electrode motor imagery EEG
+
+One subject, one low-cost 32-channel dry-electrode cap (TI ADS1299, 250 Hz, WiFi/UDP).
+Motor imagery and cognitive-task sessions, plus a set of calibrator recordings used to
+characterise the amplifier.
+
+Code, analysis and reports: **[github.com/twu3202/EEG_MI](https://github.com/twu3202/EEG_MI)**
+
+> **Read §4 before analysing.** Two of the five MI sessions have defects that change what
+> you can conclude from them, and they are not visible in the file itself.
+
+---
+
+## 1. Quick start
+
+```python
+from load_cap32 import load, epochs, to_mne
+
+rec = load("data/mi/cap32_20260725_163251_hands-feet-math.npz")
+X, y, t = epochs(rec, tmin=-1.0, tmax=4.0)   # (75, 32, 1250) µV, relative to imagery onset
+print(X.shape, set(y))                       # {'hands', 'feet', 'math'}
+
+raw = to_mne(rec)                            # or read the paired _raw.fif directly
+raw.info["bads"] = ["F7"]                    # see §4.3
+raw.filter(1., 40.).notch_filter(50.).set_eeg_reference("average")
+```
+
+Pure MNE:
+
+```python
+import mne
+raw = mne.io.read_raw_fif("data/mi/cap32_20260725_163251_hands-feet-math_raw.fif", preload=True)
+events, event_id = mne.events_from_annotations(raw)
+ep = mne.Epochs(raw, events, event_id, tmin=-1., tmax=4., baseline=(-1., 0.))
+```
+
+`python load_cap32.py` summarises every session. The loader needs only numpy (MNE optional).
+
+## 2. Hardware and montage
+
+| | |
+|---|---|
+| Amplifier | TI ADS1299, 24 bit, gain 24 |
+| Scaling | `µV = counts × 0.02235`; full scale ±187500 µV |
+| Sampling | 250 Hz |
+| Electrodes | dry, 32 channels |
+| Reference | on-board floating reference (**not** re-referenced in these files) |
+| Transport | WiFi UDP, 105-byte frames with a 1-byte sequence counter |
+
+Channel order (the row order of `data`):
+
+```
+FP1 FP2 AF3 AF4 F3 F4 F7 F8 FC1 FC2 FC5 FC6 C3 C4 T7 T8
+CP1 CP2 CP5 CP6 P3 P4 P7 P8 PO3 PO4 O1 O2 FZ CZ PZ OZ
+```
+
+**The signal is raw: unfiltered, not re-referenced, in µV.** The CAR and 1–40 Hz band-pass
+shown in the acquisition GUI affect the display only and are never written to disk, so you
+choose the preprocessing. Dry electrodes drift hard — high-pass at 0.5–1 Hz before looking
+at anything.
+
+## 3. File format
+
+### `data/mi/*.npz` — the complete source
+
+| key | dtype | shape | meaning |
+|---|---|---|---|
+| `data` | float32 | (32, N) | raw µV, unfiltered, pre-CAR |
+| `fs` | float64 | () | 250.0 |
+| `ch_names` | U3 | (32,) | the order above |
+| `marker` | int32 | (N,) | **primary label track**; the task code at imagery onset, 0 elsewhere |
+| `trigger` | int32 | (N,) | hardware trigger echo (redundant path; all-zero when the firmware does not echo) |
+| `gap` | int8 | (N,) | 1 = this sample was linearly interpolated across a dropped UDP frame — **not EEG** |
+| `trial_*` | | (n_trials,) | trial table: `index/code/name/onset/cue_onset/end`, in sample indices |
+| `meta_json` | str | () | full snapshot of paradigm, timing, link statistics, acquisition settings |
+| `format_version` | int64 | () | 2 (with trial table); the two earliest recordings lack this field |
+
+Prefer the trial table over the label track — it also records where the cue ended and
+imagery began.
+
+### `data/mi/*_raw.fif` — MNE version
+
+Same signal in volts, with the `standard_1020` montage and task names as annotations.
+**It does not carry the `gap` track** — go back to the npz to exclude interpolated samples.
+
+### `data/mi/*.json` — sidecar
+
+A readable copy of `meta_json`, so paradigm and link stats can be inspected without opening
+the npz.
+
+### Task codes
+
+| code | name | kind | | code | name | kind |
+|---|---|---|---|---|---|---|
+| 1 | rest | rest | | 10 | math | cognitive (serial subtraction by 7) |
+| 2 | left | motor (left hand) | | 11 | words | cognitive (word association) |
+| 3 | right | motor (right hand) | | 12 | song | cognitive (auditory imagery) |
+| 4 | feet | motor (both feet) | | 13 | navigate | cognitive (spatial navigation) |
+| 5 | tongue | motor (tongue) | | 14 | rotation | cognitive (mental rotation) |
+| 6 | hands | motor (both hands) | | 15 | face | cognitive (familiar face) |
+
+Only `left / right / hands / rest / feet / math` actually occur here.
+
+### Trial timing
+
+Four-phase state machine, 9.0 s per trial: `fixation 1.5 → cue 1.5 → imagery 4.0 → rest 2.0`.
+`trial_onset` points at **imagery onset**; `trial_cue_onset` at cue onset. Imagery was
+kinesthetic (imagine the *feeling* of moving, not watching yourself move).
+
+## 4. Known issues — please read all of these
+
+### 4.1 The hands-rest session has only 19 usable trials
+
+`cap32_20260725_143756_hands-rest` lists 50 trials. **Only 19 are real.**
+
+The acquisition socket had no timeout, so when the board stopped sending, `recvfrom` blocked
+forever and the reader thread died silently. The paradigm kept running, and trials 20–50 were
+all logged at the same frozen sample index (the end of the file). They are kept in the table
+for provenance but carry no signal.
+
+`valid_trials()` in the loader filters them; never use `len(trials)`. (A 2 s watchdog with
+auto-reconnect fixed this afterwards; later sessions are unaffected.)
+
+### 4.2 The left/right session has a visual confound
+
+During `cap32_20260725_135441_mi`, a countdown digit on screen changed four times *during*
+the imagery window. The result: **the only significant effect in the whole session was
+occipital 13–30 Hz (+46 %, p = 0.023)** — that is the digit, not motor imagery. Central
+channels showed nothing (p = 0.41).
+
+Usable for studying visual/artifact responses; **not usable for evaluating MI decoding**.
+Later sessions keep the countdown inside the cue phase and hold the imagery screen static.
+
+### 4.3 F7 is an intermittent open circuit
+
+F7 is flagged bad in all four sessions here. But it worked in a separate SSVEP test and has
+gone green in impedance checks — so this is **a connector/lead fault, not scalp contact**.
+
+Treat F7 as a bad channel (interpolate or drop), but not as a permanently dead electrode.
+`cap32_20260722_135310` additionally has FC1 / O2 / OZ bad.
+
+### 4.4 The subject was MI-naive
+
+This affects interpretation: for an untrained subject, **the urge to suppress actual movement
+plausibly dominated the motor imagery itself**. Both engage somatotopically organised
+sensorimotor cortex, so this offline data **cannot separate** the two. Do not read any
+hands-vs-feet effect here as pure motor imagery.
+
+### 4.5 Interpolated samples
+
+`gap == 1` marks linear interpolation, not measurement. Per session: hands-rest 84 samples
+(0.198 %, 3 bursts, longest 40 frames = 160 ms); all others 0.
+
+Interpolating rather than skipping is deliberate — silently dropping missing frames compresses
+the time axis and shifts every downstream latency and frequency estimate. Exclude affected
+trials when epoching (the loader does by default), or at least know that you did not.
+
+## 5. Sessions
+
+| file | duration | tasks | usable trials | bad channels | frame loss |
+|---|---|---|---|---|---|
+| `cap32_20260721_224212` | 1 s | — | — | — | — |
+| `cap32_20260722_135310` | 31 s | — | — | F7, FC1, O2, OZ | — |
+| `cap32_20260725_135441_mi` | 270 s | left / right | 30 (15+15) | F7 | 0.108 % |
+| `cap32_20260725_143756_hands-rest` | 170 s | hands / rest | **19** of 50 | F7 | 0.308 % |
+| `cap32_20260725_163251_hands-feet-math` | 675 s | hands / feet / math | 75 (25×3) | F7 | 0.045 % |
+
+`sessions.csv` is the machine-readable version.
+
+## 6. Calibrator recordings
+
+`data/calibrator/` holds the amplifier characterisation: the cap off the head, an external
+generator's ground tied to board GND and REF, and a 7 Hz sine driven into **one** channel
+input with every other input left open.
+
+| file | driven | rate | duration |
+|---|---|---|---|
+| `cal_7hz_180s.npz` | CH29 (FZ), VHDCI pin 31 | 250 Hz | 179 s |
+| `cal_7hz_fp1_250sps_180s.npz` | CH1 (FP1), VHDCI pin 2 | 250 Hz | 180 s |
+| `cal_7hz_fp1_1000sps_180s.npz` | CH1 (FP1), VHDCI pin 2 | 1000 Hz | 180 s |
+| `cal_7hz_xt7hz_60s.npz`, `cal_7hz_probe11s.npz` | CH29 (FZ) | 250 Hz | 59 s, 11 s |
+
+Two things to know before using them. First, an open ADS1299 input rails: 25–28 of the 32
+channels sit pinned at +187500 µV and carry **no information at all**. Second, the tone is at
+**7.0226 Hz, not 7.000** — fitting the nominal frequency over a long record averages a steady
+24.5 µV tone down to 0.31 µV.
+
+The terminated channel is a clean characterisation of the front end: 0.092 µV rms
+(20–45 Hz), THD 0.22 %, 50 Hz at 0.0045 µV, and the two driven electrodes agree on amplitude
+to 0.05 %. No crosstalk was detectable into any other channel; the measurement bounds it at
+−24 dB, a bound set by the *open* inputs' noise rather than by the amplifier.
+Full analysis in `docs/crosstalk_report.pdf`.
+
+## 7. What has been found so far
+
+From `docs/mi_pilot_report.pdf`:
+
+- **left vs right is not decodable** (p = 0.41). A dry cap does not resolve C3 vs C4 finely
+  enough, and that session also carried the visual confound in §4.2.
+- **hands vs rest is decodable**, AUC 0.83–0.87 (n = 19, breadth search over 115
+  channel × band × pipeline combinations). An earlier 0.90 in the report was best-of-5-bands
+  selection bias and is corrected there.
+- **hands vs feet separates in mu (8–13 Hz) on 17 central+frontal channels**, AUC 0.704,
+  p = 0.040. Exploratory and selection-biased; turning it into a trustworthy result needs a
+  pre-registered re-recording.
+- **Seven EEG foundation models all lost to the classical pipeline** (CSP / Riemannian
+  tangent space + LR, 0.790). The initially top-ranked BENDR (0.733) was an artifact of zero
+  padding — it scored 0.303 once the input was fixed. A representation-health gate (padding
+  fraction, channel coverage, centred variation ratio, effective rank) now runs before any
+  probe score is trusted.
+
+## 8. Citation and license
+
+CC BY 4.0. If you use this data, please link back to
+[github.com/twu3202/EEG_MI](https://github.com/twu3202/EEG_MI).
+
+```bibtex
+@misc{cap32_mi_eeg_2026,
+  title  = {cap32: 32-channel dry-electrode motor imagery EEG},
+  author = {Twu},
+  year   = {2026},
+  url    = {https://huggingface.co/datasets/Twu31/cap32-mi-eeg}
+}
+```
+
+Recordings are from a single consenting adult subject (the author), released deliberately.
+No clinical or identifying information is included.

data/calibrator/cal_7hz_180s.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aea9e6081dfe7c6296a379288a478fdf395d77654978409574d57fc17c7fa9ca
+size 1210398

data/calibrator/cal_7hz_fp1_1000sps_180s.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23a40924d3ecada6372860c95feac7f4f5cc53350a9888a95291991b9c2b1b4f
+size 2736880

data/calibrator/cal_7hz_fp1_250sps_180s.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e48587910a599cf10841f6272d3e2b85391dc930cac422211390f6864ff22ef5
+size 676218

data/calibrator/cal_7hz_probe11s.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c62e9d95b16ac5e944365266715aa17b25046cb667259fe218343dfed91a42a
+size 80581

data/calibrator/cal_7hz_xt7hz_60s.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:152a2eacff3e17b7638dd60da765e25dea34b0617d48a05601f63154b711572d
+size 400591

data/mi/cap32_20260721_224212.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:282af146a8ba758809c88440f7fda394d3cb6565e686f6974ce1004864280f18
+size 23646

data/mi/cap32_20260722_135310.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72b41f79bd33bd1e2eed88a15befb2a5bbf4b891bd4a9e0676d43fdf6f199e15
+size 726089

data/mi/cap32_20260722_135310_raw.fif

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ba1c866f9e1a0e70771e9eb265773837b5c9427a3eb88df0924a4ae55e378f9
+size 998640

data/mi/cap32_20260725_135441_mi.json

@@ -0,0 +1,114 @@
+{
+ "paradigm": {
+  "kind": "motor-imagery",
+  "tasks": [
+   "left",
+   "right"
+  ],
+  "reps": 15,
+  "imagery_mode": "kinesthetic",
+  "sequence": [
+   "left",
+   "right",
+   "left",
+   "left",
+   "right",
+   "right",
+   "left",
+   "left",
+   "right",
+   "left",
+   "left",
+   "right",
+   "right",
+   "right",
+   "left",
+   "left",
+   "left",
+   "right",
+   "left",
+   "right",
+   "right",
+   "right",
+   "right",
+   "right",
+   "right",
+   "right",
+   "left",
+   "left",
+   "left",
+   "left"
+  ],
+  "timing": {
+   "fixation": 1.5,
+   "cue": 1.5,
+   "imagery": 4.0,
+   "rest": 2.0
+  }
+ },
+ "acquisition": {
+  "source": "udp",
+  "host": "192.168.4.1",
+  "port": 8086,
+  "sfreq": 250.0,
+  "car": true,
+  "deblink": false,
+  "display_filter": {
+   "low": "1",
+   "high": "40",
+   "notch50": true
+  },
+  "channels": [
+   "FP1",
+   "FP2",
+   "AF3",
+   "AF4",
+   "F3",
+   "F4",
+   "F7",
+   "F8",
+   "FC1",
+   "FC2",
+   "FC5",
+   "FC6",
+   "C3",
+   "C4",
+   "T7",
+   "T8",
+   "CP1",
+   "CP2",
+   "CP5",
+   "CP6",
+   "P3",
+   "P4",
+   "P7",
+   "P8",
+   "PO3",
+   "PO4",
+   "O1",
+   "O2",
+   "FZ",
+   "CZ",
+   "PZ",
+   "OZ"
+  ],
+  "dead_channels": []
+ },
+ "link": {
+  "frames_received": 73760,
+  "frames_lost": 80,
+  "loss_pct": 0.108,
+  "samples_filled": 0
+ },
+ "n_filled_samples": 0,
+ "filled_pct": 0.0,
+ "format_version": 2,
+ "saved": "20260725_135441",
+ "n_samples": 67459,
+ "n_channels": 32,
+ "sfreq": 250.0,
+ "n_trials": 30,
+ "duration_s": 269.84,
+ "units": "microvolts",
+ "notes": "RAW µV, pre-CAR, unfiltered"
+}

data/mi/cap32_20260725_135441_mi.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:800059aefa9c2ae07f7284d5098045a2af46cf2fce7ef7328bfb6f4f0a5f8562
+size 5699057

data/mi/cap32_20260725_135441_mi_raw.fif

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9608e0aa7f62bd1772300f39588acbd4f6a88fb0407107cdeb62ea54f2f65c1
+size 8644932

data/mi/cap32_20260725_143756_hands-rest.json

@@ -0,0 +1,134 @@
+{
+ "paradigm": {
+  "kind": "motor-imagery",
+  "tasks": [
+   "hands",
+   "rest"
+  ],
+  "reps": 25,
+  "imagery_mode": "kinesthetic",
+  "sequence": [
+   "hands",
+   "rest",
+   "rest",
+   "rest",
+   "hands",
+   "hands",
+   "hands",
+   "rest",
+   "rest",
+   "rest",
+   "hands",
+   "hands",
+   "hands",
+   "hands",
+   "rest",
+   "rest",
+   "rest",
+   "hands",
+   "hands",
+   "hands",
+   "rest",
+   "hands",
+   "hands",
+   "hands",
+   "rest",
+   "hands",
+   "rest",
+   "rest",
+   "hands",
+   "rest",
+   "rest",
+   "rest",
+   "rest",
+   "rest",
+   "rest",
+   "hands",
+   "rest",
+   "rest",
+   "hands",
+   "rest",
+   "rest",
+   "hands",
+   "hands",
+   "hands",
+   "hands",
+   "rest",
+   "hands",
+   "hands",
+   "rest",
+   "hands"
+  ],
+  "timing": {
+   "fixation": 1.5,
+   "cue": 1.5,
+   "imagery": 4.0,
+   "rest": 2.0
+  }
+ },
+ "acquisition": {
+  "source": "udp",
+  "host": "192.168.4.1",
+  "port": 8086,
+  "sfreq": 250.0,
+  "car": true,
+  "deblink": false,
+  "display_filter": {
+   "low": "1",
+   "high": "40",
+   "notch50": true
+  },
+  "channels": [
+   "FP1",
+   "FP2",
+   "AF3",
+   "AF4",
+   "F3",
+   "F4",
+   "F7",
+   "F8",
+   "FC1",
+   "FC2",
+   "FC5",
+   "FC6",
+   "C3",
+   "C4",
+   "T7",
+   "T8",
+   "CP1",
+   "CP2",
+   "CP5",
+   "CP6",
+   "P3",
+   "P4",
+   "P7",
+   "P8",
+   "PO3",
+   "PO4",
+   "O1",
+   "O2",
+   "FZ",
+   "CZ",
+   "PZ",
+   "OZ"
+  ],
+  "dead_channels": []
+ },
+ "link": {
+  "frames_received": 53040,
+  "frames_lost": 164,
+  "loss_pct": 0.308,
+  "samples_filled": 84
+ },
+ "n_filled_samples": 84,
+ "filled_pct": 0.1976,
+ "format_version": 2,
+ "saved": "20260725_143756",
+ "n_samples": 42520,
+ "n_channels": 32,
+ "sfreq": 250.0,
+ "n_trials": 50,
+ "duration_s": 170.08,
+ "units": "microvolts",
+ "notes": "RAW µV, pre-CAR, unfiltered"
+}

data/mi/cap32_20260725_143756_hands-rest.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da7ca19a73d718ddd3fddf9900039a323ac293e2977e131ce1be826951cb7e2f
+size 3591751

data/mi/cap32_20260725_143756_hands-rest_raw.fif

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:629b16d53c827f8b712f10461bc5529724b865c9ac69503322bd4a8b3f5f7ce7
+size 5450964

data/mi/cap32_20260725_163251_hands-feet-math.json

@@ -0,0 +1,161 @@
+{
+ "paradigm": {
+  "kind": "motor-imagery",
+  "tasks": [
+   "hands",
+   "feet",
+   "math"
+  ],
+  "reps": 25,
+  "imagery_mode": "kinesthetic",
+  "sequence": [
+   "math",
+   "math",
+   "hands",
+   "feet",
+   "math",
+   "math",
+   "math",
+   "feet",
+   "feet",
+   "hands",
+   "math",
+   "feet",
+   "hands",
+   "feet",
+   "feet",
+   "math",
+   "math",
+   "feet",
+   "math",
+   "feet",
+   "hands",
+   "hands",
+   "feet",
+   "hands",
+   "feet",
+   "math",
+   "math",
+   "feet",
+   "math",
+   "hands",
+   "hands",
+   "math",
+   "hands",
+   "feet",
+   "feet",
+   "math",
+   "feet",
+   "feet",
+   "hands",
+   "hands",
+   "feet",
+   "feet",
+   "math",
+   "feet",
+   "math",
+   "math",
+   "hands",
+   "feet",
+   "feet",
+   "hands",
+   "hands",
+   "hands",
+   "hands",
+   "math",
+   "math",
+   "math",
+   "feet",
+   "hands",
+   "math",
+   "hands",
+   "hands",
+   "hands",
+   "math",
+   "math",
+   "math",
+   "feet",
+   "feet",
+   "hands",
+   "hands",
+   "math",
+   "feet",
+   "hands",
+   "hands",
+   "hands",
+   "feet"
+  ],
+  "timing": {
+   "fixation": 1.5,
+   "cue": 1.5,
+   "imagery": 4.0,
+   "rest": 2.0
+  }
+ },
+ "acquisition": {
+  "source": "udp",
+  "host": "192.168.4.1",
+  "port": 8086,
+  "sfreq": 250.0,
+  "car": true,
+  "deblink": false,
+  "display_filter": {
+   "low": "1",
+   "high": "40",
+   "notch50": true
+  },
+  "channels": [
+   "FP1",
+   "FP2",
+   "AF3",
+   "AF4",
+   "F3",
+   "F4",
+   "F7",
+   "F8",
+   "FC1",
+   "FC2",
+   "FC5",
+   "FC6",
+   "C3",
+   "C4",
+   "T7",
+   "T8",
+   "CP1",
+   "CP2",
+   "CP5",
+   "CP6",
+   "P3",
+   "P4",
+   "P7",
+   "P8",
+   "PO3",
+   "PO4",
+   "O1",
+   "O2",
+   "FZ",
+   "CZ",
+   "PZ",
+   "OZ"
+  ],
+  "dead_channels": []
+ },
+ "link": {
+  "frames_received": 177560,
+  "frames_lost": 80,
+  "loss_pct": 0.045,
+  "samples_filled": 0,
+  "stalls": []
+ },
+ "n_filled_samples": 0,
+ "filled_pct": 0.0,
+ "format_version": 2,
+ "saved": "20260725_163251",
+ "n_samples": 168776,
+ "n_channels": 32,
+ "sfreq": 250.0,
+ "n_trials": 75,
+ "duration_s": 675.1,
+ "units": "microvolts",
+ "notes": "RAW µV, pre-CAR, unfiltered"
+}

data/mi/cap32_20260725_163251_hands-feet-math.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ada910f1f6f3c1a0014a1cebc067eb907b30539869640ebeba9e45ed11d2a907
+size 14444714

data/mi/cap32_20260725_163251_hands-feet-math_raw.fif

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb53aff81563ccfb42ecbff78cbe321b0f1ed5df413efdbb632599ded7ad050f
+size 21620779

docs/crosstalk_report.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35f2f32998a6be07cc2609fb35fac2b1664b780dec4d9f0bfd0e5c37d1544f39
+size 928596

docs/crosstalk_report_zh.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c1cbeb19e862e59582581c4c9e7ed6e0d0160dbc525d42eb35b7422a585d98a
+size 1154873

docs/hardware_acceptance.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c7e95c8d687abe7eacea4a5e9e5b6b52328956e060fd22458ff15a4071adaa6
+size 881721

docs/mi_pilot_report.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f60db53b28fc4bce09da30003c195eb5cccbf5e114c8d9c2d9e5a347bf3625ee
+size 577680

load_cap32.py

@@ -0,0 +1,131 @@
+#!/usr/bin/env python
+"""Standalone loader for the cap32 recordings — numpy only, no project code needed.
+
+Everything you need is in the .npz: raw µV, the per-sample label track, the trial table,
+and the gap track that says which samples were reconstructed after a UDP drop. The paired
+_raw.fif is the same signal in MNE's format with the task labels as annotations, for people
+who would rather start from `mne.io.read_raw_fif`.
+
+  python load_cap32.py                                   # summarise every session
+  python load_cap32.py data/mi/cap32_20260725_163251_hands-feet-math.npz --epochs
+"""
+from __future__ import annotations
+
+import argparse
+import json
+from pathlib import Path
+
+import numpy as np
+
+HERE = Path(__file__).resolve().parent
+
+# The paradigm writes these small integers into the `marker` track at imagery onset.
+CODE_TO_LABEL = {1: "rest", 2: "left", 3: "right", 4: "feet", 5: "tongue", 6: "hands",
+                 10: "math", 11: "words", 12: "song", 13: "navigate", 14: "rotation",
+                 15: "face"}
+
+
+def load(path):
+    """-> dict with data (n_ch, N) µV, fs, ch_names, marker/trigger/gap tracks, trials, meta."""
+    d = np.load(path, allow_pickle=True)
+    out = {k: d[k] for k in d.files if k != "meta_json"}
+    out["fs"] = float(d["fs"])
+    out["ch_names"] = [str(c) for c in d["ch_names"]]
+    out["meta"] = json.loads(str(d["meta_json"])) if "meta_json" in d.files else {}
+    if "trial_name" in d.files:
+        out["trials"] = [
+            dict(i=int(d["trial_index"][k]), name=str(d["trial_name"][k]),
+                 code=int(d["trial_code"][k]), onset=int(d["trial_onset"][k]),
+                 cue_onset=int(d["trial_cue_onset"][k]), end=int(d["trial_end"][k]))
+            for k in range(len(d["trial_name"]))
+        ]
+    return out
+
+
+def valid_trials(rec):
+    """Trials whose imagery window actually exists in the data.
+
+    The 2026-07-25 hands-rest session hit a receiver stall: the acquisition thread blocked
+    on a socket with no timeout, so trials 20-50 were all logged at the same frozen sample
+    index (the end of the file). They are kept in the table for provenance but carry no
+    signal — filter with this, never with len(trials)."""
+    n = rec["data"].shape[1]
+    ts = rec.get("trials", [])
+    return [t for i, t in enumerate(ts)
+            if t["onset"] < n and (i == 0 or t["onset"] > ts[i - 1]["onset"])]
+
+
+def epochs(rec, tmin=-1.0, tmax=4.0, drop_filled=True):
+    """-> X (n_trials, n_ch, n_times) µV, y (labels), times. Cut around imagery onset.
+
+    `drop_filled` removes trials that overlap samples reconstructed across a UDP drop —
+    those samples are linear interpolation, not EEG, and they are flagged in rec['gap']."""
+    fs, X, y = rec["fs"], [], []
+    lo, hi = int(round(tmin * fs)), int(round(tmax * fs))
+    gap = rec.get("gap")
+    for t in valid_trials(rec):
+        a, b = t["onset"] + lo, t["onset"] + hi
+        if a < 0 or b > rec["data"].shape[1]:
+            continue
+        if drop_filled and gap is not None and gap[a:b].any():
+            continue
+        X.append(rec["data"][:, a:b])
+        y.append(t["name"])
+    return np.asarray(X), np.asarray(y), np.arange(lo, hi) / fs
+
+
+def to_mne(rec):
+    """-> mne.io.RawArray with the standard_1020 montage and task annotations."""
+    import mne
+    info = mne.create_info(rec["ch_names"], rec["fs"], "eeg")
+    raw = mne.io.RawArray(rec["data"] * 1e-6, info, verbose="ERROR")   # MNE wants volts
+    raw.set_montage(mne.channels.make_standard_montage("standard_1020"),
+                    match_case=False, on_missing="ignore")
+    ts = valid_trials(rec)
+    if ts:
+        raw.set_annotations(mne.Annotations(
+            onset=[t["onset"] / rec["fs"] for t in ts],
+            duration=[(t["end"] - t["onset"]) / rec["fs"] for t in ts],
+            description=[t["name"] for t in ts]))
+    return raw
+
+
+def summarise(path):
+    rec = load(path)
+    n = rec["data"].shape[1]
+    tr, vt = rec.get("trials", []), valid_trials(rec)
+    gap = rec.get("gap")
+    print(f"\n{path.name}")
+    print(f"  {rec['data'].shape[0]} ch × {n} samp = {n/rec['fs']:.0f} s @ {rec['fs']:.0f} Hz, µV")
+    if tr:
+        cnt = {}
+        for t in vt:
+            cnt[t["name"]] = cnt.get(t["name"], 0) + 1
+        print(f"  trials: {len(vt)} usable of {len(tr)} logged   {cnt}")
+        if len(vt) < len(tr):
+            print(f"  ⚠ {len(tr)-len(vt)} trial(s) logged after the stream stalled — no signal")
+    if gap is not None and gap.any():
+        print(f"  gap-filled: {int(gap.sum())} samp ({100*gap.mean():.3f} %) — interpolated, not EEG")
+    link = rec["meta"].get("link")
+    if link:
+        print(f"  link: {link.get('loss_pct', 0):.3f} % frame loss")
+
+
+def main():
+    ap = argparse.ArgumentParser(description=__doc__,
+                                 formatter_class=argparse.RawDescriptionHelpFormatter)
+    ap.add_argument("path", nargs="?", help="a .npz (default: summarise all)")
+    ap.add_argument("--epochs", action="store_true", help="also cut and report epochs")
+    a = ap.parse_args()
+    paths = [Path(a.path)] if a.path else sorted((HERE / "data" / "mi").glob("*.npz"))
+    for p in paths:
+        summarise(p)
+        if a.epochs:
+            X, y, t = epochs(load(p))
+            if len(X):
+                print(f"  epochs: X {X.shape}  y {dict(zip(*np.unique(y, return_counts=True)))}"
+                      f"  t [{t[0]:.1f}, {t[-1]:.1f}] s")
+
+
+if __name__ == "__main__":
+    main()

sessions.csv

@@ -0,0 +1,6 @@
+file,duration_s,n_samples,sfreq,n_channels,tasks,trials_logged,trials_usable,per_class,gap_samples,frame_loss_pct,format_version
+cap32_20260721_224212,0.8,192,250,32,,0,0,,0,,1
+cap32_20260722_135310,31.0,7757,250,32,,0,0,,0,,1
+cap32_20260725_135441_mi,269.8,67459,250,32,left+right,30,30,left=15;right=15,0,0.108,2
+cap32_20260725_143756_hands-rest,170.1,42520,250,32,hands+rest,50,19,hands=10;rest=9,84,0.308,2
+cap32_20260725_163251_hands-feet-math,675.1,168776,250,32,hands+feet+math,75,75,feet=25;hands=25;math=25,0,0.045,2