Application of deconvolutional networks for feature interpretability in epilepsy detection
Sihao Shao, Yu Zhou, Ruiheng Wu, Aiping Yang, Qiang Li · 2025
Read Application of deconvolutional networks for feature interpretability in epilepsy detection by Sihao Shao, Yu Zhou, Ruiheng Wu, Aiping Yang, Qiang Li (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-11802560.txt.
Description
Sihao Shao, Yu Zhou, Ruiheng Wu, Aiping Yang, and Qiang Li published this study, dated 2024 in its text, on seizure detection that can explain itself. Algorithms already scan EEG recordings for epilepsy, but most cannot say which scalp channels drove a verdict, and clinicians notice. The team built a fully convolutional network with squeeze-and-excitation modules, then swapped modules in and out to compare designs.
On the CHB-MIT benchmark the model balanced sensitivity and specificity at a geometric mean of 82.7 percent, beating the rival module combinations for stability. Because the design is deconvolutional, channel contributions could be quantified: midline and frontotemporal sites, FZ, CZ, PZ, FT9, FT10, and T8, mattered most. The authors pitch interpretability as a clinical feature, not decoration, since a detector that shows its reasons earns review time.
Catalog record
| CREATOR | Sihao Shao, Yu Zhou, Ruiheng Wu, Aiping Yang, Qiang Li |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 33 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |