Table 1. Datasets selected at different stages of the analysis: hit finding selection, PSD-fitting quality filtering, particle-size filtering and single-hit diffraction pattern selection.
The percentage of the chosen dataset to the initial one S0 is given in parentheses.
| Analysis step | Dataset name | Number of diffraction patterns |
|---|---|---|
| Initial dataset | S0 | 1.2 × 107 |
| Hit-finding classification | Shit | 191183 (1.6%) |
| PSD-fitting score filtering | Sfit | 179886 (1.5%) |
| Particle-size filtering | SD | 18213 (0.1%) |
| First EM-based classification |
|
1609 |
| Second EM-based classification |
|
1402 |
| Third EM-based classification |
|
1366 |
| Fourth EM-based classification |
|
1401 |
| Fifth EM-based classification |
|
2119 |
| Final EM-based classification | SEM | 1085 (0.009%) |
| Manual selection | Sman | 1393 (0.01%) |