CMEG common mistakes
CMEG · Understand the traps and how to avoid them.
Treating an empty room as proof of no patient artifact
Patient movement, muscle and accessories introduce sources absent from an empty room.
What to do instead
Review patient-related contamination using timing, spatial pattern and auxiliary information.
Improvising a cryogenic repair
Pressure, cold and atmospheric hazards require specialized procedures.
What to do instead
Stop unsafe work and arrange qualified assessment through the laboratory safety process.
Changing an implant without authorization
A technical artifact does not justify an independent clinical device change.
What to do instead
Document the issue and consult the responsible clinical and device team.
Prioritizing a clean recording over an acute event
Patient protection is more important than an uninterrupted dataset.
What to do instead
Follow the clinical safety response and annotate the event when safe.
Assuming averaging removes every artifact
The square-root model does not apply to every coherent or nonstationary contaminant.
What to do instead
Check assumptions, reproducibility and artifact timing rather than rely on the trial count alone.
Trying to recover clipping by rescaling a display
Presentation changes cannot restore information lost during acquisition.
What to do instead
Investigate the acquisition problem and document the limitation while preserving original data.
Interpreting absent task activity without checking performance
Poor engagement, timing or acquisition can weaken a response.
What to do instead
Document actual performance and limitations before a qualified clinical interpretation.
Combining runs with unreviewed head geometry
Changed head position can alter source-to-sensor relationships.
What to do instead
Use documented run-specific geometry and an appropriate validated alignment or compensation procedure.
Equating fit with anatomical truth
An excellent fit can coexist with incorrect coordinates or source assumptions.
What to do instead
Check coregistration, model adequacy and analysis sensitivity before interpreting the location.
Promising surgery outcome from a tight cluster
Source concordance is evidence rather than an outcome guarantee.
What to do instead
Integrate the clinical history and wider multidisciplinary assessment without promising seizure freedom.
Waiting for every check before reporting a serious change
Delay can prevent the responsible team from assessing a time-critical event.
What to do instead
Communicate promptly through the agreed pathway while checks proceed in parallel.
Overwriting the only raw file
The original evidence and processing audit trail can be lost.
What to do instead
Preserve source data and maintain traceable derived outputs and settings.