Back

Direct Generation of Images from EEG using Schrödinger Bridge

Nobe, S.; Sasai, S.

2025-08-22 neuroscience
10.1101/2025.08.18.670451 bioRxiv
Show abstract

Real-world data is often noisy, making it challenging to extract true signals Non-invasively recorded neural activities are among the most difficult data, yet its precise signal reconstruction is highly anticipated by communities developing non-invasive brain-machine interfaces. Several noise sources contribute to this challenge, including unrelated neuronal activity, non-brain bioelectricity, attenuation by the skull and scalp, and environmental noises. Additionally, the accumulation of noise varies significantly across subjects and recording sessions, resulting in widely diverging distributions of degraded observations. In this study, we propose modeling the noise accumulation process as a Schrodinger bridge and decoding the true signal by reversing this process. Compared to conventional guided Diffusion approach, our Schrodinger bridge approach effectively models diverse noise processes within a single framework, exhibiting greater robustness to inter-subject variability. Also, our approach doesnt require pre-aligning brain and image representations, which is an additional compute cost in the conventional approach.

Matching journals

The top 8 journals account for 50% of the predicted probability mass.

1
Journal of Neural Engineering
○ 221 papers in training set
Top 0.3%
11.6%
Similar papers in this journal
2
IEEE Transactions on Medical Imaging
● 21 papers in training set
Top 0.1%
7.1%
Similar papers in this journal
3
PLOS Computational Biology
● 1863 papers in training set
Top 5%
7.1%
Similar papers in this journal
4
NeuroImage
○ 903 papers in training set
Top 2%
6.6%
Similar papers in this journal
5
Imaging Neuroscience
● 282 papers in training set
Top 0.9%
6.6%
Similar papers in this journal
6
IEEE Transactions on Biomedical Engineering
● 40 papers in training set
Top 0.2%
5.4%
7
Human Brain Mapping
○ 329 papers in training set
Top 1%
4.7%
8
PLOS ONE
● 5266 papers in training set
Top 32%
4.2%
50% of probability mass above
9
Frontiers in Neuroscience
○ 256 papers in training set
Top 1%
3.4%
10
Scientific Reports
○ 3612 papers in training set
Top 34%
3.2%
11
Medical Image Analysis
○ 35 papers in training set
Top 0.3%
3.1%
12
IEEE Journal of Biomedical and Health Informatics
● 37 papers in training set
Top 0.3%
3.1%
13
IEEE Transactions on Neural Systems and Rehabilitation Engineering
● 49 papers in training set
Top 0.4%
3.1%
14
Nature Communications
○ 5641 papers in training set
Top 40%
2.4%
15
Frontiers in Computational Neuroscience
○ 60 papers in training set
Top 0.6%
2.3%
16
Neural Computation
● 39 papers in training set
Top 0.4%
2.1%
17
Neurocomputing
○ 13 papers in training set
Top 0.2%
1.5%
18
Frontiers in Psychiatry
○ 87 papers in training set
Top 2%
1.3%
19
Neural Networks
○ 35 papers in training set
Top 0.4%
1.3%
20
Nature Machine Intelligence
○ 70 papers in training set
Top 2%
1.3%
21
Cognitive Neurodynamics
○ 18 papers in training set
Top 0.2%
1.3%
22
eLife
● 5828 papers in training set
Top 60%
1.1%
23
Patterns
○ 78 papers in training set
Top 2%
1.1%
24
IEEE Access
● 35 papers in training set
Top 1%
1.0%
25
Journal of Neuroscience Methods
○ 122 papers in training set
Top 2%
0.8%
26
Network Neuroscience
● 126 papers in training set
Top 2%
0.6%

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.