EEG-Based Da ase Explici ly Ta ge s he
T ansi ions be ween Si ing and S anding
o Explo ing Neu al Ac i a ion Pa e ns in
Mo o Image y and Execu ion
Benjaka n Uengsawapak, Supa i Kongwudhikunako n,
Suk ipol Kia ha eephong, Wipamas Polpakdee, Ra anaphon Chaisaen,
Po ama e Manoonpong, Chani sada Chuenchi , Gun Bhakdisongkh am and
Thee awi Wilaip asi po n
Abs ac
This s udy p esen s he i s publicly accessible elec oencephalog aphy (EEG)
da ase explici ly a ge ing si - o-s and and s and- o-si ansi ions du ing bo h mo o
execu ion (ME) and mo o image y (MI) asks. Twen y- wo heal hy pa icipan s pe o med
si ing and s anding ansi ions unde well-con olled expe imen al condi ions while
60-channel EEG, elec ooculog aphy (EOG), and elec omyog aphy (EMG) signals we e
synch onously eco ded. The da ase enables he explo a ion o neu al ac i a ion pa e ns
associa ed wi h lowe -limb mo emen s and suppo s he de elopmen o EEG-based
b ain–compu e in e ace (BCI) algo i hms o mobili y assis ance and ehabili a ion. To
alida e he da ase , a benchma k classi ica ion was conduc ed using EEGNe , a compac
con olu ional neu al ne wo k. Resul s demons a ed consis en decoding pe o mance wi h
mean accu acies o app oxima ely 80 % o ME and 70 % o MI, indica ing he eliabili y and
usabili y o he da ase . Addi ionally, analyses o mo emen - ela ed co ical po en ials
(MRCPs) and e en - ela ed desynch oniza ion/synch oniza ion (ERD/ERS) pa e ns
e ealed dis inc neu al signa u es ac oss he ansi ion phases. This da ase p o ides a
comp ehensi e ounda ion o s udying lowe -limb mo o con ol, neu al dynamics, and he
ad ancemen o MI-based BCIs o ehabili a ion and assis i e echnologies.
Da ase Desc ip ion
Pa icipan s
Twen y- h ee heal hy pa icipan s (aged 22–28 yea s; i een males) wi h no known
neu ophysiological abno mali ies we e ec ui ed o his s udy. One pa icipan (S05) was
excluded due o poo signal quali y, esul ing in a inal coho o wen y- wo pa icipan s. P io
o he expe imen , esea ch s a p o ided bo h e bal and w i en explana ions o he s udy
objec i es, p o ocol, ques ionnai e, and expe imen al se up o ensu e pa icipan s'
comp ehension.
1
The demog aphics o he subjec s a e in he able below.
Subjec Code
Age
Gende
Dominan side
EEG Expe ience
Rema k
S01
26
M
igh
yes
S02
23
F
igh
yes
S03
26
M
igh
yes
S04
24
M
igh
no
S05
26
M
igh
no
Excluded
S06
25
M
igh
no
S07
25
M
igh
no
S08
26
F
igh
no
S09
23
F
igh
no
S10
23
F
igh
yes
S11
18
F
igh
no
S12
30
M
igh
no
S13
22
F
igh
no
S14
23
M
igh
no
S15
26
M
bo h
yes
S16
25
M
igh
no
S17
23
M
igh
no
S18
26
M
igh
no
S19
25
M
le
no
S20
27
M
igh
no
S21
28
F
igh
no
S22
28
M
igh
yes
S23
24
M
le
no
2
Da a Collec ion
Physiological signals, including 60 EEG, 2 EOG, and 6 EMG, we e ob ained om he
pa icipan s. EEG and EOG elec odes a e eco ded wi h 1200 Hz sampling a e and hei
placemen s a e illus a ed in he igu es below. The hEOG is placed a he igh emple, while
he EOG is placed a he igh in a o bi al. EMG da a we e eco ded wi h a sampling a e o
2000 Hz. Each o 3 senso s was a ached o he le and igh legs, a ge ing he ollowing
muscles: Soleus (SL), Tibialis An e io (TA), and Rec us Femo is (RF). The EMG senso s’
placemen and sequence a e in o med in he igu es and ables below.
Elec ode placemen posi ions o eco ding EEG and EOG signals.
Channel
Index
Channel
Name
Channel
Index
Channel
Name
Channel
Index
Channel
Name
Channel
Index
Channel
Name
1
p1
17
5
33
cpz
49
p7
2
p2
18
6
34
pz
50
p8
3
a 7
19
cz
35
cp1
51
poz
4
a 8
20
cz
36
cp2
52
oz
5
7
21
c1
37
cp3
53
po3
6
8
22
c2
38
cp4
54
po4
7
7
23
c3
39
cp5
55
po7
8
8
24
c4
40
cp6
56
po8
9
a 3
25
c5
41
p7
57
po9
10
a 4
26
c6
42
p8
58
po10
11
a z
27
c1
43
p1
59
o1
12
z
28
c2
44
p2
60
o2
13
1
29
c3
45
p3
61
hEOG
14
2
30
c4
46
p4
62
EOG
15
3
31
c5
47
p5
63
igge
16
4
32
c6
48
p6
EEG and EOG channel indices and channel names desc ip ion.
3
Su ace EMG senso placemen posi ions o eco ding he EMG signal.
Channel Index
Channel Name
Loca ion
EMG-1
sl_l
le Soleus muscle
EMG-2
sl_
igh Soleus muscle
EMG-3
a_l
le Tibialis An e io muscle
EMG-4
a_
igh Tibialis An e io muscle
EMG-5
_l
le Rec us Femo is muscle
EMG-6
_
igh Rec us Femo is muscle
EMG channel indices, channel names, and placemen loca ions.
Da a/Files Fo ma
Two e sions o he da ase a e p o ided. Raw da a, in .ma o ma , allows lexible
modi ica ion and u he esea ch explo a ion; while, p ep ocessed da a, in . i o ma ,
unde goes da a p ep ocessing s eps, including bandpass il e ing and downsampling o
enhancing ep oducibili y, imp o ing da a quali y, and educing compu a ional load. De ails
a e as ollows.
Raw Da ase
The aw EEG and EOG da a o each pa icipan a e s uc u ed as a 2-D ma ix o
dimension n_channels ╳ n_ imepoin s. An ex a channel ( he 63 d channel) is alloca ed o
e en igge s, which anno a e ask- ela ed e en s, in which he de ails a e p esen ed below.
4
E en T igge Numbe
Desc ip ion
1
Eyes closed in he es ing s a e.
2
Eyes opened in he es ing s a e.
10
S a o ials in ME ac i i y.
11
S a o ME_SIT_STD in ME ac i i y.
12
S a o ME_STD_SIT in ME ac i i y.
13
S a o es ing condi ion (ME_R) in ME ac i i y.
20
S a o ials in MI du ing si .
21
S a o MI_SIT_STD in MI du ing si .
22
S a o MI_SIT_SIT ask in MI du ing si .
23
S a o es ing condi ion in MI du ing si (MI_R_SIT, es while si ing).
30
S a o ials in MI du ing s and.
31
S a o MI_STD_STD ask in MI du ing s and.
32
S a o MI_STD_SIT in MI du ing s and.
33
S a o es ing condi ion in MI du ing s and (MI_R_STD, es while s anding)
No e: T igge 0 indica es ha no e en occu ed. The pa icipan is pe o ming he co esponding ask.
E en igge numbe desc ip ion able. The e en igge is s o ed a EEG channel numbe #63 in he .ma ile
and embedded in o MNE Epochs om he . i ile.
Dic iona y o A ibu e De ini ions
A ibu e de ini ions a e in he able below.
A ibu e
Type
Desc ip ion
Uni s
eeg
loa a ay
EEG da a
μV
emg
loa a ay
EMG da a
V
eeg_ s
loa a ay
EEG imes amps
Seconds
emg_ s
loa a ay
EMG imes amps
Seconds
eeg_channels
s ing a ay
Lis o EEG channels, EOG channels,
and igge
-
emg_channels
s ing a ay
Lis o EMG channels
-
eeg_ s
in a ay
EEG sampling a e
Hz
emg_ s
in a ay
EMG sampling a e
Hz
5
Mini-Tu o ial S ep-By-S ep o he Raw Da ase
impo scipy.io as sio
# Load .ma ile
da a = sio.loadma (' aw_da ase .ma ')
# Lis a iables in he loaded ile
p in (da a.keys())
# Example: check EEG da a
eeg = da a['eeg']
p in (eeg.shape, eeg.d ype)
# Check imes amps
eeg_ s = da a['eeg_ s']
p in (eeg_ s.shape, eeg_ s.d ype)
# EEG i s 5 samples
p in (eeg[:, :5])
# EEG imes amps i s 5 samples
p in (eeg_ s[:5])
P ep ocessed Da ase
The p ep ocessed EEG da a is o ganized as a single h ee-dimensional ma ix wi h
he dimensions n_ ials × n_channels × n_ imepoin s. This ma ix con ains he comple e
da ase , encompassing all mo emen asks om all eco ding sessions. Each ial ep esen s
an ac i i y segmen aligned wi h a igge ma ke . The da a o speci ic mo emen asks can
be accessed using hei ask names (i.e. me_si _s d, me_ _si , me_s d_si ,
me_ _s d, mi_si _s d, mi_ _si , mi_s d_si , mi_ _s d), and he ials om
a desi ed session can be accessed using he index ange speci ied in he able below.
6
Ac i i y
Session
Round
Index Range
(s a _idx:end_idx)
ME
1
1
[0:20, :, :]
2
1
[20:40, :, :]
MI
1
1
[0:10, :, :]
2
[10:20, :, :]
2
1
[20:30, :, :]
2
[30:40, :, :]
Mini-Tu o ial S ep-By-S ep o he P ep ocessed Da ase
impo mne
# Load epoched . i ile
epochs = mne. ead_epochs('MI_S01. i ', p eload=T ue)
#inspec epoch in o ma ion
p in (epochs) # o e iew o epochs
#access da a
da a = epochs.ge _da a() # shape: (n_epochs, n_channels, n_ imes)
p in (da a.shape)
#selec speci ic e en
## Selec only 'mi_si _s d' ask
epochs_mi_si _s d = epochs['mi_si _s d']
p in (epochs_mi_si _s d)
#selec speci ic session
## selec all ials om session 1
session1_mi_si _s d_epochs = epochs_mi_si _s d[0:20,:,:]
Folde S uc u e
Da ase /
│
├── eadme.pd # Da a desc ip ion documen
│
└── 1_ aw_S01.zip # O iginal/ aw da a o subjec 1 (each subjec and session)
│ ├── S01_S1.ma # subjec 1 session 1
│ └── S01_S2.ma # subjec 1 session 2
│
└── 1_ aw_S02.zip
7
│ ├── S02_S1.ma
│ └── S02_S2.ma
└── …
│
└── 1_ aw_S05.zip # S05 Excluded om da ase
└── …
│
└── 1_ aw_S23.zip
│ ├── S23_S1.ma
│ └── S23_S2.ma
│
└── 2_p ocessed_all.zip # P ep ocessed/cleaned da a
├── ME/ # Mo o Execu ion, S01–S23 (22 subjec s, exclude S05)
│ ├── S01. i
│ ├── S02. i
│ └── …
│ └── S23. i
└── MI/ # Mo o Image y, S01–S23 (22 subjec s, exclude S05)
├── S01. i
├── S02. i
└── …
└── S23. i
File Types: .ma ( aw da a) and . i (p ep ocessed da a)
File Naming Con en ion:
1. Raw da ase : S<ID>_S<session num>.ma
2. P ep ocessed da ase : S<ID>. i
Va iable De ini ions:
ID = Subjec ID
session num = Session numbe
8
File size:
The da ase is dis ibu ed as comp essed .zip iles in wo ca ego ies:
1. aw da a: P o ided as sepa a e .zip iles o each subjec , including wo sessions.
Example: 1_ aw_S01.zip con ains aw da a o subjec 1 (~ 2 GB pe subjec ).
Fo all 22 subjec s (S01–S23, excluding S05): app oxima ely 44.12 GB
2. p ep ocessed da a: P o ided as a single comp essed .zip ile, 2_p ocessed_all.zip
(~3.77 GB), including ME and MI da a om all 22 subjec s (S01–S23, excluding
S05).
Da a A ailabili y
The aw and p ep ocessed da a a e a ailable ia he ollowing DOIs in he open-access
online eposi o y, Zenodo (h ps://zenodo.o g).
● aw da a: DOI:10.5281/zenodo.17561969
● p ep ocessed da a: DOI:10.5281/zenodo.17629950
9
