Late Gadolinium Enhancement (LGE) allows imaging of infarction by visualizing the accumulation of contrast agent within the myocardium using an inversion recovery (IR) prepared sequence. However, myocardial hyperenhancement can sometimes be poorly detected in the vicinity of epicardial fat because both fat and hyperenhanced tissue appear bright, generating the need for fat signal suppression. One previous approach to fat-suppressed LGE was to null fat signal with appropriately timed tip-up/tip-down fat-selective RF pulses [1,2], but this technique can leave residual fat signal in the image due both to sensitivity to off-resonance and timing requirements to achieve fat nulling. The goal of the present work is to achieve complete removal of fat signal from LGE images and separate visualization of fat by the Phase-Sensitive ChEmical Selection (PiSCES) method, combining phase-sensitive (PS) image reconstruction [3] with customized timing of fat-selective RF pulses.

PiSCES is a modification of PS LGE. First, an IR pulse is played with an inversion time (TI) chosen to null normal myocardial tissue. Next, fat-selective tip-up and tip-down pulses are applied, with the tip-down pulse played immediately prior to data acquisition such that fat experiences minimal T₁ relaxation and longitudinal magnetization of fat signal lies below the transverse plane (M_z<0) during acquisition (Fig. 1).

After PS reconstruction, fat is represented by negative pixel values and appears black due to its opposite signal polarity (M_z<0) vs. hyperenhanced tissue (M_z>0). To achieve this, the following condition for TI_fat, the time between the fat tip-down pulse and the center of k-space acquisition, must be met: TI_fat < -T_1,fat*log( 0.5 + exp(-(TI-T_t)/T_1,fat) – exp(-TI/T_1,fat) + 0.5*exp(-2*RR/T_1,fat) ), where T_t: time between IR and fat tip-up pulses, RR: R-R interval duration.

PiSCES was applied in 22 patients using Discovery MR750w 3.0T (GE Healthcare) and compared to conventional fat-suppressed magnitude LGE. Informed consent was obtained.

In 19 out of 22 exams, conventional fat nulling resulted in significantly reduced but not completely removed fat signal (Fig. 2(b,f,j)), whereas with PiSCES fat signal was completely removed (Fig. 2(c,g,k)), as judged by an experienced cardiac radiologist. There were 2 scans with incomplete fat suppression, presumably due to poor shim and 1 scan with enlarged fat region, likely because of partial volume effects of fat and normal myocardium, which is a potential drawback to the method to be further examined.The negative polarity of fat at readout and appropriate windowing of the image led to complete elimination of fat signal with the new PiSCES approach. PiSCES improved the diagnose of pericarditis, as it better differentiated pericardial hyperenhancement from fat (Fig. 2(k)). Additionally the technique is relatively insensitive to timing of the fat-selective RF pulses: fat appears black as long as fat has negative polarity, thus the method has potential to be more robust across a variety of imaging conditions. Fat signal and fluid (due to its long T₁) can be identified by the negative pixel values and overlaid in color (Fig. 2(d,h,l), further improving cardiac tissue differentiation.