MR imaging of cortical bone with positive contrast is feasible by several methods, including by incorporating preparation pulses [1,2] into 'ultra-short-TE' (UTE) imaging sequences or by rescaling excitation pulses of such sequences [3]. The latter method is a simple sequence modification and has been demonstrated with a three-dimensional (3D) UTE sequence. By re-rating hard RF excitation pulses to match the excitation rate to $$$T_2$$$ relaxation and imaging at echo times paired with the pulse scaling, it creates 'MR-simulated-CT' images with CT-like contrast [4].

Here, the excitation-rescaling principle is applied to half-pulse excitations [5] to allow two-dimensional (2D) radial encoding for MR-sim-CT imaging of a single slice. Accessing MR-sim-CT images with 2D sequences can be enabling for MRI contexts incompatible with hard pulse excitation and 3D encoding—e.g., interventional procedures requiring body-coil imaging or needing temporal resolution better than a few minutes.

Many slice-selective-excitation methods are incompatible with imaging short-$$$T_2$$$ signals because the time elapsing between the peak and the end of the excitation pulse (or rephasing gradient) is longer than the $$$T_2$$$. One slice-selection method that is compatible is application of half-pulses—splitting excitation into two parts and alternating the slice-select gradient to create the intended profile by complex sum [5]. This permits rapid initation of image encoding following excitation.

The acquisition method for creating MR-sim-CT images alternates between 'fast' (high-$$$\lvert{B_1}\rvert$$$) RF pulses exciting a broad range of $$$T_2$$$ signals and 'slow' (low-$$$\lvert{B_1}\rvert$$$) pulses eliciting only longer-$$$T_2$$$ signals (Fig.1). This principle can be adapted for 2D sequences with half-pulse excitations by scaling RF and gradient waveforms to lower the maximum $$$B_1$$$ but maintain the same spatial excitation profile.

In the simplest conception, one scaling factor $$$s$$$ can be applied to all excitation pulse amplitudes and durations (Fig.1). For given RF and gradient waveforms $$$B_1(t)$$$ and $$$G_z(t)$$$, the modified waveforms are $$\tilde{B_1}(t)={1/s}B_1(t/s)\quad\;\textrm{and}\;\quad\tilde{G_1}(t)={1/s}G(t/s)\textrm{.}$$ De-rating reduces peak-$$$B_1$$$ by a factor of $$$s$$$ but also lowers the effective excitation bandwidth. Scaling the gradient permits maintenance of the slice profile. Assuming these scalings do not affect other sequence timings, no additional modifications are necessary.

Under this modification, equivalent slice profiles can be excited by both full-scale and down-scaled pulses—e.g., $$$s=1$$$ and $$$s=6$$$—so only the $$$T_2$$$-sensitivity of the excitation is modulated (Fig.2). Differences in $$$T_2$$$-sensitivity can be used to highlight short-$$$T_2$$$ signals. The $$$T_2$$$ difference profiles rendered by rescaled half-pulses and hard pulses are similar.

A 2D-radial gradient-spoiled UTE sequence using TBW=4 half-pulses with facility for scaling excitation pulses was implemented for a 1.5T commercially available clinical scanner. The half-pulse was calculated to continue transmit during slice-select gradient-ramp-down time. A maximum nominal-$$$B_1$$$ of 11$$$\mu$$$T was used for half-pulses to avoid amplifier saturation or other nonlinear effects. Tip angle and repetition time (TR) of 15$$$^{\circ}$$$ and 30ms respectively were chosen in analogy to parameters for previously-reported acquisitions for MR-sim-CT imaging. A 5mm slice with 30cm field-of-view (FOV) and 1.2mm resolution was encoded in all 2D acquisitions.

Phantom images were acquired using a transmit/receive head coil to demonstrate slice profile maintenance with down-scaled pulses. Distilled water (bottle 1; $$$T_2$$$=1810ms) and manganese chloride (bottle 2; $$$T_2$$$=2ms) were imaged. Normal image encoding was performed; additionally, the slice profile was encoded by transferring waveforms from one readout axis to the slice-select axis. For comparison between different $$$s$$$ scalings, the slice cross-section was 8$$$\times$$$ sinc-interpolated.

In vivo imaging of a healthy volunteer's lower leg was performed with a transmit/receive extremity coil. As reference, a 3D-radial UTE sequence using hard pulses, also with facility for rescaling excitation pulses and implemented for the same scanner, was used to acquire a volume of the same leg on a different day. The 3D-radial sequence applied maximum nominal-$$$B_1$$$ of 24$$$\mu$$$T because hard-pulses are not severely degraded by amplifier nonlinearity. Isotropic 30cm FOV with 1.2mm resolution was encoded using 12$$$^{\circ}$$$ tip and 8ms TR.

Phantom images acquired under four different scalings of slice-selective pulses depict similar excitation and no apparent profile degradation (Fig.3). The modulated $$$T_2$$$-sensitivity effect is apparent by comparison between the short-$$$T_2$$$ and long-$$$T_2$$$ bottles' profiles. In vivo images depict similar contrast from the 2D acquisition and from the 3D sequence (Fig.4). The MR-sim-CT image derived from slice-selective imaging shows hyperintense cortical bone, similar to non-selective-excitation imaging results.Scaling amplitude and duration of half-pulse excitations for 2D UTE imaging modulates excitation $$$T_2$$$ sensitivity without changing slice profile, creating contrast appropriate for MR-sim-CT images. The 2D images acquired under this adaptation show contrast similar to that achieved by relaxation-rate-matching hard pulses with 3D UTE acquisition. This facilitates MR-sim-CT imaging for MRI contexts in which 3D imaging is not practicable.