TABLE 2.
The methods for classifying epilepsy using deep learning in MEG data.
| References | Moda-lity | Problem that was solved in that study | Database | Data acquisition | Source locali-zation | Fea-tures | Classi-fication | Performance metrics | |||||||||
| Sample size | Age range |
Sex (M:F) |
Source | Total dura-tion | Cha-nnels |
Seg-ment/ epoch length |
Fre-quency samp-ling | Pre-process-ing |
Sensiti-vity (%) |
Specifi-city (%) |
Accu-racy (%) |
||||||
| Aoe et al., 2019 | MEG | EPs vs HCs | 140 EPs, 26 SCIs, 67 HCs | 21–86 | 123:110 | Osaka University hospital | 4 or 5 min | 160 | 800 ms | 1,000 Hz or 2,000 Hz | low-pass filter: 50 Hz, high-pass filter: 1,000 Hz, down-sampled to 1,000 Hz | / | / | MNet | / | / | 63.4 ± 12.7 |
| EPs vs SCIs | / | / | 79.8 ± 11.7 | ||||||||||||||
| EPs vs SCIs vs HCs | / | / | 70.7 ± 10.6 | ||||||||||||||
| Gu et al., 2020 | MEG | SPS vs CPs | 32 EPs | 20–32 | 1:1 | Nanjing Brain Hospital, Nanjing Medical University | 40 min | / | 2 min | 1,200 Hz | band-pass filter: 0.03–300 Hz | / | / | MSAM | 90.8 | 90.7 | 83.6 |
| Fujita et al., 2022 | MEG | EPs vs HCs | 90 Eps, 90 HCs | 7–86 | 93:87 | Osaka University hospital | 4 or 5 min | 160 | 2,400 ms | 1000 Hz or 2,000 Hz | low-pass filter: 500 Hz, high-pass filter: 0.1 Hz, bandstop filter: 60 Hz, down-sampled to 1,000 Hz | / | relative power, functioncal connectivity (FC), phase-amplitude coupling (PAC) | MNet (Aoe et al., 2019) | 90 | 90 | 90 |