Figure 1 for Imaging mode should be interpreted as follows:
The philosophy for patterns 0-25 is that a good X-ray pattern must be compact, with the highest charge at the center, and isolated (all pixels around are below threshold).
Patterns 26-29 are the so-called diagonal patterns, not expected from a genuine X-ray, but which can arise in case of Si fluorescence or of pileup of two monopixel events.
The E and E data in the ODF event lists have nothing to do with the "white" and "crossed" pixels.
Figure 2 for Timing mode should be interpreted in the same way as in Imaging mode, with the difference that the place where maximum charge occurs is ignored. There are only crosses in the rows above and below that of the event. This means that the Timing pattern analysis is purely 1-D (it is insensitive to other rows), because each Timing 'row' is actually the sum of 100 true rows, so the rows are not physically related. Because there is no test for maximum charge, all doubles appear as PATTERN=1, whether leading or trailing.
Patterns 2 and 3 are not bounded. They are mostly not due to true X-rays, but to cosmic-ray tracks. After binning by 100 rows, cosmic-ray tracks appear as an horizontal string of pixels above threshold. Their beginning will be recorded by an event with PATTERN=2 and their end by an event with PATTERN=3. In other words, patterns 2 and 3 normally occur in pairs and define a horizontal string of pixels above threshold.
XMM-Newton SOC -- 2021-11-30