Table 2:

Comparison of frames retained with FD<0.2 or fFD<0.1 criteria in different datasets. Across most datasets, adopting fFD<0.1 for frame censoring retained significantly more frames that an FD<0.2 criteria. The exception is the Petersen 120+ dataset.

	percent frames retained
Dataset	FD<0.2	fFD<0.1	difference
PIB-PD	63.3%	79.1%	15.7%***
Barch-SZ	57.4%	76.2%	18.9%***
Petersen 120+	79.7%	77.7%	−2.0%***
Iowa Lesion	47.6%	59.7%	12.0%***
Dallas Lifespan	69.0%	84.2%	15.3%***
UIUC Lifespan	66.6%	82.8%	16.2%***
MSC	82.5%	86.0%	3.4%**

^***p<0.001,

^**p<0.01