In conventional MRI, RF excitation and data acquisition is time-inter­lea­ved, leading to unwanted dead times, which severely limit the de­tec­tion of tissue with ultra-short T₂. Concurrent excitation and acquisition (CEA) allows detecting the MR signal during radiofrequency (RF) excitation without dead times [1]. CEA was shown to be feasible in a clinical MRI system using an active decoupling technique that combines geometrical decoupling with phase/amplitude (PA) decoupling [2, 3]. Active decoupling, however, suffers from instabilities of transmit signal, and alterations in achieved isolation due to motion and changes in loading.

Pick-up coils (PUCs) have been used for monitoring impedance changes due to differences in loading in transmit (Tx) coils [4]. In this work, the active decoupling setup was upgraded by integrating PUCs for simultaneous monitoring of the Tx and the leakage signals. The monitored signals were used to retrospectively eliminate leakage signals, and it was tested in CEA of a human wrist.

CEA experiments were conducted on a clinical 3T MRI system (Tim Trio, Siemens) with an 8-channel parallel transmit unit. Low noise amplifiers HD24388 (gain: 23dB, NF: 0.7dB) and HD29980 (gain: 36dB, NF: 1.0dB) (HD Communications Corp, NY) were connected at 2 Tx channels. The CEA coil setup (Fig. 1) consisted of a custom-made primary coil (Tx1), a decoupling coil (Tx2), and a receive coil (Rx), all of which had an integrated PUC (Inner diameter: 5 mm). PUCs were used for monitoring of Tx and leakage signals during acquisition. In a CEA experiment the acquired MR signal, s(t), can be described as:

$$ s(t) = FID\otimes B_{1}(t)+h(t)\otimes B_{1}(t) $$

where the ideal FID signal is convolved with the RF excitation signal B₁(t), and a frequency-dependent leakage component h(t)*B₁(t), which is the remaining B₁-induced voltage due to imperfect de­coupling [5]. The leakage was measured using the Rx-PUC, and the B₁(t) alone is measured using Tx-PUC. Geometrical decoupling was achieved by orthogonal placement of Tx1 and Rx coil. The remaining B₁-induced current in the Rx coil was cancelled by adjusting the phase and amplitude scale of Tx2 (PA decoupling), which was performed with a phase/amplitude sensitivity of 0.01°/ 3±0.5dB and 0.001 / 3±0.5dB (at a 4 V_peak RF voltage). Because the input power at Tx2 is lower than Tx1 due to geometrical decoupling, the flip angle in the sample is assumed to be mainly determined by the RF transmit field of Tx1.

After PA decoupling using a water phantom, the right wrist of a healthy subject was imaged. 3D CEA MR data were acquired with a radial CEA pulse sequence (Fig. 2) with 80000 spokes during a chirp RF excitation (Df = 64kHz, duration: 4ms) at a TR = 6ms resulting in a total acquisition time of 7:48 min:s. At an input RF power of 120 mW the estimated flip angle amounted to 2°. CEA data were re-gridded onto a 512³ Cartesian grid using Kaiser-Bessel interpolation. For anatomical reference a 3D FLASH data set was acquired with FOV = 224mm, TR = 10ms, TE = 2.85ms, a =2°, 200Hz/px bandwidth, 0.9mm resolution and 1.88mm partition thickness.

In this prototype setup, phase and amplitude adjustment required setup times of 30-45 min. Geometrical and PA decoupling resulted in 24dB and 56dB isolation, respectively. The PA decoupling varied 14dB along the excitation bandwidth of 64 kHz. Transmit noise was measured to be 8dB higher than the receive noise floor (-96dBm). Replacing the water phantom with the loading of the subject’s wrist resulted in only a 12dB decrease in decoupling, which was sufficient to avoid saturation of the receiver. Continuous gradient ramping in CEA sequence decreased acoustic gradient switching noise considerably.

Anatomical details of the wrist bones and cartilage are clearly visible as shown in the coronal slice of the 3D CEA image in Fig. 3. Note, that the slice positions are not exactly identical since the hand was repositioned. The short T₂* components have a higher signal intensity in CEA than in FLASH, however, the CEA contrast is close to a proton density contrast. Employing inversion or saturation pulses other contrasts can be produced with CEA. Simultaneous monitoring provided the leakage signal directly without a need for modeling and eliminates the potential imperfections due to the differences in the digital and analog version of the Tx signal. With a dynamic analog and digital decoupling system automatic decoupling of Tx and Rx coils could be achieved. In general, in vivo CEA MRI was shown to be feasible using an active decoupling setup with PUCs for leakage and transmit signal monitoring.