Regionally resolved cardiac metabolism using a dipole-loop array coil for 7 T 31P-MRSI

Abstract

Purpose: We introduce a novel commercial phosphorus-31 (³¹P) dipole-loop array coil, describing the coil hardware and testing its performance on phantoms. We used this coil to assess cardiac metabolism per region in healthy volunteers.

Methods: B₁ ⁺ field maps were simulated and compared to maps measured with a set of CSI sequences with varying voltages. Seventeen volunteers were scanned with 7 T phosphorus-31 magnetic resonance spectroscopic imaging (³¹P-MRSI). Reproducibility was assessed in nine of these volunteers. Strain was measured for six of these volunteers at 3 T.

Results: Blood- and saturation-corrected Phosphocreatine/γ-adenosine triphosphate (PCr/ATP) ratios were measured for four regions of the left ventricle: 1.86 in septum, 2.25 in anterior wall, 1.41 in inferior wall, and 1.53 in lateral wall, respectively. These are in the expected range compared to previous studies. B₁ ⁺ maps show good signal uniformity around the position of the heart (0.13 ± 0.06 μT/sqrt(W)). Intrasession and intersession coefficients of reproducibility were 0.22-0.88 and 0.29-0.79, respectively. Linear modeling shows that regional PCr/γATP correlates with circumferential strain but not radial strain. This requires corroboration by a larger study including patients with impaired function and energetics.

Conclusion: Dipole-loop array coils present a promising new approach for human cardiac ³¹P-MRSI at 7 T. Their favorable B₁ ⁺ uniformity at depth and specific absorption rate over loop arrays and improved SNR when combined with loops for reception could be beneficial for further clinical studies measuring energetics by ³¹P-MRSI at 7 T. The new capability to assess PCr/γATP ratios across the whole left ventricle could enable clinical studies to investigate regional changes in cardiac energetics for the first time.

Keywords: 31P‐MRSI; array coils; heart energetics; heart failure; metabolic imaging; spectroscopic imaging.

FIGURE 1

(A) Photograph of the coil showing the posterior and anterior elements. (B) Interface box connecting the coil to the scanner, which includes Tx/Rx switches and preamplifiers. (C) Position of the 8× ¹H and 8× ³¹P Tx/Rx dipoles on the body phantom (right side). The ¹H dipoles are stacked outside the ³¹P dipoles. Schematic of the coil interface electronics (left side). The scanner ³¹P output from TIM socket 3 is split four ways and fed to the central four dipoles for Tx (A2, A3, P2, and P3). All 8× ³¹P dipoles and 16× ³¹P loops are used for Rx. The scanner ¹H output from TIM socket 1 is fed to 1× posterior ¹H dipole for Tx; all 8× dipoles are used for Rx. This is due to limitations in the 7 T Terra software (Siemens Healthcare, Erlangen, Germany) that allow either parallel Tx or non‐proton scans in a single session. Cables with different lengths were connected from the dipoles (A2, A3, P2, and P3) to the back of the interface box for RF shimming. ¹H, hydrogen‐1; ³¹P, phosphorus‐31; TIM, total imaging matrix; Tx/Rx, transmit/receive.

FIGURE 2

(A) Anterior element of the body array coil. (B) Zoom photograph of the A2 element of the coil. In this element, coil position marker (cod liver oil capsules, referred to as ¹ H fiducials) were placed in the housing of the A2 element. A phosphorus calibration fiducial (see section 2.4.3 for content) was also installed. (C) Zoom illustration from within a coil element. Each housing from A1‐4 and P1‐4 contains one ¹H dipole, one ³¹P dipole, and two ³¹P Rx loops. (D, E) The ³¹P and ¹H PCBs are shown, respectively. PCB, printed circuit boards.

FIGURE 3

Comparison of the experimental B₁ ⁺ map with the simulations. (A) A transverse slice of the Duke model (IT'IS Foundation, Zurich, Switzerland) used for simulations experiments. The RF shim target for the simulated B₁ ⁺ map is outlined with a red cross. (B) ³¹P simulation results for transmitting only with the dipole A2. (C) The simulated map in μT/sqrt(W). (D) Peak SAR per average power input per dipole. (E) An illustration of the phantom used for the experiments. (F) The experimental B₁ ⁺ map in μT/sqrt(W). SAR, specific absorption rate.

FIGURE 4

Chronogram of our custom MRSI sequence. (A) RF Tx/Rx (ADC) events. (B) Schematic gradient waveforms.

FIGURE 5

(A) B₁ ⁺ value measured experimentally with a series of FID sequences of varying voltages and comparison with simulation results and (B) distances from the dipole to the center of the shaft. Only the dipoles highlighted in red were used for Tx in B₁ ⁺ map and in vivo experiments.

FIGURE 6

(A) Mid‐short axis GRE localizer acquired with breath hold with CSI matrix overlaid and PSF on the top‐left corner of the image. The BISTRO saturation band is shown in yellow. (B, J) Real part of spectra from the corresponding voxels (black) and corresponding AMARES fits (red). AMARES, advanced method for accurate, robust, and efficient spectral fitting; BISTRO, B₁‐insensitive train to obliterate signal; GRE, gradient echo; PST, point‐spread‐function.

FIGURE 7

(A) 3 T short‐axis cine image from a volunteer showing the four cardiac segments. (B) Illustration of different types of strain measurement. (C, D) Results of stepwise linear modeling: (C) Plot of CS versus PCr/γATP) with all other parameters kept fixed, and (D) with the other parameters regressed out to show only the effects attributed specifically to CS and PCr/γATP. Plots (C, D) were made with the “plotSlice.m” and “plot.m” MatLab R2022b (MathWorks, Natick, MA) functions for linear modeling. γATP, adenosine triphosphate; CS, circumferential strain.

FIGURE 8

(A) Intersubject PCr/γATP bar plot in the septal, inferior, lateral, and anterior walls of the heart. For participants included in the reproducibility study, the PCr/γATP ratio from the first scan in session 1 has been used. (B) Intrasession difference of the PCr/γATP value between the first and second scan in session 1 for all subjects that participated in the reproducibility study and for each segment. (C) Intersession difference of the PCr/γATP value between the first scan of sessions 1 and 2. (D) Intersession variability defined by the difference of the PCr/γATP value between the first scan of sessions 1 and 2 and compared with coefficients of reproducibility from previous studies.,