APT^1-4 is an emerging MRI technique for sensitive tissue characterization via an assessment of the local concentration of mobile proteins in cells. Clinical applications are particularly foreseen in oncology, e.g. for differential diagnosis of tumors (grading⁴) and for therapy follow-up³. High duty-cycle and long RF saturation (T_sat) is essential for APT sensitivity, but limited for clinical MRI systems (typically duty-cycle ≤50%; T_sat<1s). A technique based on multi-channel RF transmission was demonstrated previously^4,5, which allows quasi-continuous RF pulses with T_sat≤5s via amplifier alternation. Recently, a number of clinical studies^4,6,8 have been published related to APT in neuro oncology applications, where APT is analyzed based on magnetization transfer asymmetry (MTR_asym; Δω=±3.5ppm). MTR_asym can be measured robustly and efficiently in clinical settings, including 3D acquisitions with large coverage in less than 5 minutes^6,7. For MTR_asym analysis, precise information on the field inhomogeneity (ΔB₀) is required. A fast-spin-echo(FSE)-Dixon acquisition technique allows efficient and simultaneous acquisition of APT weighted (APTw) and ΔB₀ information⁷. An improved FSE-Dixon APTw acquisition protocol with intrinsic ΔB₀ correction and online color display of APTw images was implemented on a clinical MRI scanner using multiple averages with saturation at the amide chemical shift (Δω=+3.5ppm). Contrast homogeneity was evaluated in a volunteer study and is presented together with initial clinical results on brain tumor patients.A volunteer study (N=9, informed consent obtained) was performed at 3.0T (Ingenia, Philips Healthcare). Example patient data of various pathologies (informed consent obtained) were acquired on clinical MRI systems (AchievaTX/Ingenia, Philips) according to a protocol approved by the institutional review board. APTw imaging with intrinsic ΔB₀ correction was performed using 2-channel body coil transmission, 8 (Achieva) or 15-channel (Ingenia) head coil reception and the following parameters in a 3D FSE sequence: RF saturation with T_sat = 2s, B_1,rms=2.0μT, 40 sinc-Gaussian pulses (50ms), 100% duty-cycle; initial 7-point Z-spectral protocol7 (7p): Δω=±3.1ppm; ±3.5ppm; ±3.9ppm and -1560ppm (S₀); improved 9-point Z-spectrum (9p), see Figure 1: adding two further acquisitions at +3.5ppm (averaged for signal intensity), centric profile ordering, 9 slices, FOV 212×184×40mm³, voxel size 1.8×1.8×4.4mm³, TE/TR=6.2ms/~5s (TR depending on B₁ calibration), total acquisition time ~5 min. Acquisition windows and readout gradients were shifted (echo-shift ES) for (7p) ±3.9ppm: ES=+0.4ms; ±3.1ppm: ES=-0.4ms as well as for 9p (ii) +3.5 ppm_#4: ES=-0.4ms, +3.5 ppm_#5:ES=0ms, +3.5 ppm_#6: ES=+0.4ms (see Figure 1) for Dixon-type B₀-mapping. B₀ maps and APTw colormaps of the MTR asymmetry MTR_asym=(S_[−3.5ppm]−S_[+3.5ppm])/S₀ were calculated with a point-by-point B₀ correction within a modified reconstruction SW on the scanner. Corrected Z-spectral images, B₀ maps and APTw colormaps were stored in the image database (DICOM). Areas outside the brain were masked using radial tracing of the brain edge based the S₀ image and a 90% intensity threshold based on the radial line average. A homogeneity analysis was performed measuring the standard deviation sd(vol) of MTR_asym of all voxels inside the brain and sd(slice) comparing average values over slices in FH direction for the 7p and 9p protocols.APTw contrast homogeneity over the imaging volume for the two Z-spectral protocols (7p/9p) was found to be sd(vol,7p)=0.65±0.09% and sd(vol,9p)=0.55±0.09%, respectively. The standard deviation of the mean MTR_asym over the different slices was sd(slice,9p)=0.15±0.09% and sd(slice,9p)=0.07±0.04%, respectively. Thus, the contrast homogeneity for normal brain tissue (grey/white matter, CSF) was significantly improved using the 9p protocol. Apparently, signal averages with different echo shift at Δω=+3.5ppm contribute to robust detection of the APTw signal as well as to more stable B₀-mapping results for inhomogeneity correction, as compared with B₀-mapping over multiple different saturation frequencies (7p). A selected 3D APT result (Figure 2a) from a tumor patient (brain metastasis from cervical cancer; 9p protocol), ±5% of MTR_asym, is displayed in a rainbow color scale. The metastasis is clearly distinguished on a flat background (sd<0.7%) associated with normal appearing grey and white matter. The anatomical GRE images (b) reveal that the APTw hyperintensity stems from the tumor core, but not from apparently edematous areas.