Those interested in new MRI contrast for functional molecular imaging of brain activity.Multiple studies suggest that neuronal activity leads to defined pH changes [1,2,3]. This suggests that a new type of functional molecular imaging of brain activity is possible through pH modulations. It was concluded that T1ρ-fMRI contrast during visual stimulus comes in part from pH changes [3]. In this work, we used an order of magnitude more pH sensitive method, Chemical Exchange Saturation Transfer (CEST) at 7T, to test the concept of pH fMRI.All experiments were done on a 7T Philips MR system. A total of 3 healthy volunteers participated in the breathing experiments and fMRI sessions. During the breathing experiments, the volunteers were scanned throughout a hypercapnic breathing challenge (6 min) in which end-tidal CO2 (PetCO2) was targeted at 10mmHg about individual subject baseline levels (Respiract, Thornhill Reasearch Inc., Canada). PetCO2 was regulated to change brain pH, followed by attempts to measure the changes by amide-CEST MRI, associated with endogenous proteins and peptides, and 31P MRS. A low-power 3D steady-state CEST sequence was used [4]. Saturation prepulse (a single 1.5µT-amplitude RF-spoiled 8ms sinc-gauss pulse followed by a 50mT/m spoiler of 5ms) interleaved with a readout (sagittal acquisition, segmented EPI, EPI factor 7 with a binomial pulse for water only excitation, FOV 150x225x190mm³, 2mm resolution, SENSE factor 2, TR/TE/FA=25ms/4.2ms/12°, k-space center-weighted acquisition); temporal resolution 12.6s. A T1-weighted anatomical scan was used to create masks of white matter (WM) and gray matter (GM) to calculate whole-brain (masked by a B0 inhomogeneity of ±0.1ppm) averaged CEST signal. 3D ³¹P chemical shift imaging (CSI, 25x25x25mm³, 5min, 2 averages) used Pi as a probe for pH measurements. Functional imaging was based on the aforementioned amide-CEST sequence. The visual challenge paradigm consisted of 10blocks (Fig.3A). Four even cycles of flashing checkerboard and rest were presented with a 126s period, followed by two rests. A total of 100 3D volumes were acquired at two alternating off-resonance frequency offsets (3.5ppm–pH sensitive amide-CEST [5] and 300ppm normalization points with RF=0 as a control) in 21min. fMRI data were analyzed by using standard preprocessing steps, including motion correction and spatial smoothing. A general linear model was used to generate individual statistical maps and calculate signal change. Co-registration and segmentation were done in FSL.Intracellular pH (pHi) sensitivity of both T1ρ [6] and amide-CEST MRI [5] were studied before. Amide-CEST is almost 10 times more pH sensitive compared to T1ρ. Breathing experiments were performed to find the limit of pHi detection by amide-CEST. An increase of PCO2 by 10mmHg is expected to decrease pHi by ca. 0.05 [7]. CO2-induced T2* effect (Fig. 1B) was corrected by point-by-point normalization of amide-CEST points by the control points (Fig. 1C). No pHi effect after the correction, as demonstrated by undistinguishable from the baseline amide-CEST signals (Fig. 1C), was noticeable. Nor did ³¹P CSI measure any pHi changes even during two hypercapnic breathing challenges (separated by a rest of 10min) (Fig. 2). The inability of 31P to measure such small changes is not surprising, since it has a pH precision of ±0.1 [8]. However, the limit of amide-CEST pH sensitivity has never been measured. A surprisingly big pHi change of 0.15 was measured by 31P during a visual fMRI study [9], which is well within the pH sensitivity of amide-CEST [10]. Both BOLD and pH-sensitive amide-CEST activation maps are shown in Fig. 3 (similar maps were obtained for the other two subjects). A BOLD effect of ca. 1% was found in all three subjects, whereas no statistically significant contrast due to pHi changes was found. In contrary to the results in [9], where T1ρ fMRI was used to detect activity-induced pH changes, we did not see any activation due to pH in visual fMRI by using an order of magnitude more pH sensitive technique – amide-CEST. To test the concept of pH-fMRI, fMRI data, corrupted with a Gaussian noise, were simulated by using Bloch-McConnell equations [11] with the same experimental parameters. The visual stimulus evoked response was simulated by changing an exchange rate of amide-CEST pool, and assuming a BOLD and activity-evoked pH effects of 2% and 0.1, respectively. In the simulated data, pH effect could be separated from that of BOLD by using three regressors.This is the first proof-of-concept study showing that pH-fMRI is feasible. In vivo, however, better acquisition strategies with higher SNR are to be developed to be able to take advantage of the new way of looking at brain activity.