The current MRI-based neuro-imaging methods can be separated in two independent categories. A) Anatomical T₁-weighted imaging methods to investigate structural features, e.g., investigating pathology or plasticity [1]; and B) Fast T₂^*-weighted functional imaging methods to investigate transient brain activity changes. Here we sought to develop, apply, and evaluate a new method that combines the advantages of both imaging approaches: We combine the quantifiability, the tissue type specificity, and the high spatial resolution of anatomical MRI with the high temporal resolution of EPI-based fMRI. This new approach will enable research into the underlying mechanisms associated with GM density change and T₁ changes as seen in studies about plasticity, sleep, learning etc., and furthermore to investigate the structural changes during increased brain activity as in conventional fMRI experiments. The new imaging method for simultaneous acquisition of T₁ and T₂^*-weighted signal is based on a multi-contrast inversion-recovery (IR) simultaneous multi-slice (SMS) EPI with blipped CAIPI [2] and a new MRI signal processing pipeline similar to the one of MP2RAGE [1].

Experiments were performed on a 7T Siemens scanner with a 32-channel NOVA head-coil. Four volunteers participated in six experiments. The sequence parameters were: TI1/TI2/TR=0.9/2.4/3.0 s (as indicated in Fig. 1), SMS-factor=3 with a FoV/3 blipped-CAIPI EPI [3], GRAPPA-2 with FLEET [4]. Image reconstruction was performed online using split-Slice GRAPPA [5] as implemented in MGH C2P (http://www.nmr.mgh.harvard.edu/software/c2p/sms), #slices=30, nominal resolution=1×1×1.3mm³ and 0.7mm inter-slice gap. Slices were positioned as depicted in Fig. 1F. T₁-weighting was achieved by an optimized TR-FOCI inversion. To determine T₁, a Bloch-equation model of the sequence was fitted to the multi-TI data (similar to MP2RAGE [1]) by numerically inverting Eq. 1.

$$$\frac{S(\color{red}{TI_1})}{S(\color{green}{TI_2})}= \overbrace{\frac{sin(\alpha)}{sin(\alpha)}}^{=1}\times \overbrace{ \frac{M_0}{M_0} }^{=1} \times \overbrace{ \frac{e^{-\frac{TE}{T_2^\star}}}{e^{-\frac{TE}{T_2^\star}}} }^{=1} \times \frac{ \frac{1-(1+\chi) e^{-\frac{\color{red}{TI_1}}{T_1}} +\chi e^{-\frac{TR}{2T_1}} - cos (\alpha) e^{-\frac{TR}{2T_1}}+ cos (\alpha) e^{-\frac{TR}{T_1}} }{1+cos^2 (\alpha)e^{-\frac{TR}{T_1}}}}{1-e^{-\frac{2TR}{T_1}} \left(1- \left(\frac{1-(1+\chi) e^{-\frac{\color{green}{TI_2}}{T_1}} +\chi e^{-\frac{TR}{2T_1}} - cos (\alpha) e^{-\frac{TR}{2T_1}}+ cos (\alpha) e^{-\frac{TR}{T_1}} }{1+cos^2 (\alpha)e^{-\frac{TR}{T_1}}}\right)\right)}$$$

The pulse-specific B₁⁺-independent inversion-efficiency $$$\chi$$$ had previously been determined to be 0.8, the flip angle $$$\alpha$$$ is taken from the acquired B₁⁺-map with MAFI [6], and TI is the effective slice-specific inversion time. Segmentation and cortical layering algorithms were implemented in C++ and applied on the EPI-based T₁-maps using the equi-volume approach [7]. A 12-min finger-tapping paradigm was used here to investigate the performance and feasibility of the proposed technique.

Raw EPI images for the two TIs are shown in Fig. 1A-B for a single TR. A representative single-TR T₁-map is shown in Fig. 2A. Voxels with T₁ increases of more than 2.5% are shown in red overplayed to a T₁-map averaged over all 240 time steps (Fig. 2B). The estimated T₁ values increase by up to 100 ms during the task condition. The temporal evolution of T₁ and simultaneous T₂^*-weighted signal can be seen in Fig. 2C averaged across voxels within M1. Single-TR T₁-maps have enough SNR to apply tissue segmentation and layering (Fig. 2D) to provide measures of cortical thickness and cortical profiles of T₁. T₁-increases within GM are dominated from upper cortical layers (Fig. 2E). Cortical thickness seems to increase up to 4% during neural activity (Fig. 2F), where the GM band seems to expand towards WM and not so much towards CSF [8].The results presented here demonstrate that advanced EPI-based imaging methods are robust enough to provide anatomical information that is conventionally achievable only with lengthy (and therefore more motion-sensitive) so-called “anatomical” sequences. Both, the T₁-increase and the “swelling” of the GM might be interpreted as a volume increase of blood, which has a longer T₁ than GM (a.k.a. the VASO contrast). A CBF and inflow contribution to the contrast seems unlikely, as two modules of QUIPPS [9] saturation pulses are played out below the imaging slab right prior imaging (yellow Fig. 1E). The quantitative nature of the method may be particularly appealing when comparing conditions across a long time (longer than scanner drift), across days, and across, scanners and platforms. The fact of having only two TI for T₁-estimation makes the method very fast and minimizes MT-effects, however, it cannot be used for fitting to multi-compartment models. More research is needed to validate the quantitative T₁ values with more established ‘anatomical’ sequences. We developed a multi-contrast inversion recovery SMS CAIPI EPI MR sequence for fast dynamic high-resolution whole brain quantitative T₁-mapping. With this new method, we could observe functional changes of GM morphology during neural activation, such as T₁-increases of up to 100 ms and GM “swelling” of up to 4%. This method may prove important in quantitative, non-invasive studies investigating morphological GM changes in neural activity, plasticity, sleep and pathology.