Because the T1 of CSF is long (approximately 3800 ms), a long TR is required for full recovery of longitudinal magnetisation. At a TR of 10000 ms, a TI of 2500 ms almost completely nulls the signal from CSF/simple fluid. A TE of 100 ms produces good T2 contrast.
Using a TR of 10000 ms, a TI of 2500 ms, and a TE of 100ms results in a sequence that is positively weighted for PD and T2, but negatively weighted for T1. Increases in PD and T2 result in INCREASED signal (the filter slopes upwards). Increases in T1 result in DECREASED signal (the filter slopes downwards). FLAIR is not a synergistic sequence, but with these parameters, changes in white matter signal signal are 36% due to PD weighting, 13% due to T1 weighing, and 51% due to T2 weighting -- so this is mostly a T2 weighted sequence.
Increasing the TE increases the positive T2 weighting and decreases the negative T1 weighting, but at the expense of decreasing overall signal. At some point, the decreased signal to noise from increasing the TE offsets the advantages of decreased negative T1 weighting. This threshold is largely dependent on radiologist preference.
Because grey matter sits on the steeper part of the T1 tissue filter, changes in grey matter signal are more T1 weighted (31% vs 13%) and less T2 weighted (35% vs 51%).
So even though FLAIR is the gold standard for looking at changes in white matter, the increased T1 seen in gliosis/edema/demyelnation/etc . . . works to decrease signal while the T2 prolongation increases signal.
Some sites like to use a shorter TR to reduce the acquisition time. At a TR of 4500 a TI of 1600 provides good CSF suppression. However, this results in a sequence where changes in white matter signal are due 24% to negative T1 weighting (compared to 13%) and 45% to positive T2 weighting. This sequence is less sensitive to white matter pathology because there is more negative T1 weighting working against the less positive T2 weighting. Remenbert that most white matter pathology results in both T1 and T2 prolongation.
Likewise, for the sequence with a shorter TR of 4500ms, pathologic changes in grey matter signal are now due 43% to negative T1 weighting and only 29% to positive T2 weighting. For the grey matter this sequence is more T1 weighted than it is T2 wighted!
Return to the FLAIR with TR = 10000 ms, TI = 2500 ms, and TE = 100 ms. Notice how steep the curve is to left of simple fluid/CSF. If there is hemorrhage in the CSF spaces - either subdural or subarachnoid, the T1 value of the CSF/blood mixture will decrease. Because the curve to the left of CSF is very steep this will result in markedly increased signal in the CSF spaces due to the stong negative T1 weighting of the T1 filter at the T1 of CSF. This is why FLAIR is a sensitive sequence for subarachnoid hemorrhage.
The T1-FLAIR sequence uses a short TE to reduce T2 weighting. At a TR of 10000 ms, TI of 2500 ms, and TE of 10 ms the CSF is still nulled. There is negative T1 weighting but for changes in white matter the sequence is still primarily positive PD weighted (67%). Remember above that dropping the TR resulted in increased T1 weighting. Dropping the TR to 2100 ms and the TI to 880 ms still results in nulled CSF, but now changes in white matter signal become 55% T1 weighted and changes in grey matter signal become 66% T1 weighted. The low TR results in a fast acquisition. Since the T1 filter is very steep just to the left of the T1 of fluid, the T1-FLAIR is also very sensitive to subarachnoid and sub-dural hemorrhage.