Highlights:
• Overview of the advantages and challenges of 3D imaging applied to MSK. .
• Discussion of possible solutions for rapid 3D MSK imaging.

Target audience: Clinicians and researchers interested in rapid 3D MSK Imaging.
Objective: To provide an non-exhaustive overview of the challenges associated with rapid 3D MSK imaging and the techniques proposed to solve them.In this talk we will discuss the pros and cons of 3D MSK imaging from a technical prospective. Using select examples, we will explore how the transition from 2D (slice-selective) to 3D (volumetric) imaging influences the contrast, resolution and acquisition time. The presentation will start with the fundamental principles of 3D imaging from which we will buildup to the latest developments, such as compressed sensing and magnetic resonance fingerprinting.

Advantages: Volumetric MR methods facilitate high resolution images with an isotropic voxel size. Such a uniformly spaced 3D data set can be reformatted to visualize any desired plane, thereby eliminating the need to repeat the same scan in different orientations. Moreover, all else being equal, 3D methods have a higher signal to noise ratio [1].

Challenges: In general, it is not practical to switch from a 2D to a 3D image encoding process and keep all other sequence parameters the same. Whereas slice-selective sequences provide the opportunity to interleave multiple slices within a single scan, volumetric sequences must acquire each line in k-space sequentially. When a long repetition time (TR) is desired, the 2D multi-slice approach is much more efficient than a volumetric acquisition because it can sample all N slices within one single TR. In other words, the equivalent 3D protocol would last N times longer. Simply decreasing the TR to compensate for the reduced sampling efficiency is generally not an option, because it will deteriorate the contrast between tissues. Alternatively, one could envision increasing the “turbo factor” in a turbo spin echo sequence (TSE). The turbo factor reflects the number of spin-echoes formed after each excitation and corresponds to the number of k-space lines measured per TR. While a larger turbo factor will increase the data sampling efficiency, it also lengthens the total duration of the spin echo train. During this time the transverse magnetization relaxes (T2), attenuating the signal intensity from one echo to the next. When these measurements are placed together in a single 3D k-space, the different T2 weightings of these samples lead to blurring in the final image.

Contrast weighted imaging: The majority of MSK protocols are based on a TSE sequence with a relatively long TR. Several methods have been proposed to replace multi-slice TSE based sequences with a volumetric alternative. From a technical prospective these solutions can be divided into two complementary categories: (1) sequence design and (2) under-sampling.

(1) Sequence design based strategies aim to improve the rate at which k-space lines are acquired. One notable solution was proposed by Mugler, et al.,[2]. Their optimized 3D-TSE sequence strives to produce a more stable transverse magnetization throughout the spin-echo train by varying the angle of the refocusing pulses. With effective turbo factors as high as a 100, the total acquisition time can be made comparable to traditional 2D-TSE protocols while offering a higher through-plane resolution [3].

(2) Undersampling strategies strive to minimize the number of samples needed to reconstruct the final image. When imaging an extended volume, the region can be covered by a larger number receive coils distributed in multiple directions. With a suitable receiver array-coil, undersampling in orthogonal directions results in a significantly lower g-factor penalty [4, 5, 6], Moreover, compared to a 2D sampling strategy, there are more ways in which the samples can be distributed in 3D k-space. Incoherent sampling combined with compressed sensing can enable previously unprecedented acceleration factors [7], which can help facilitate volumetric MSK imaging within a clinically expectable time.

Quantitative Imaging: Volumetric quantitative imaging methods can be relatively fast. Sequences such as DESPOT [8], use a relatively short TR and can measure T1 and T2 within a clinically acceptable time. Aside from the diagnostic benefits of quantitative measures, a complete set of multi-parametric maps can also be used to synthesize contrast weighted images that would have otherwise taken to much time to measure. However, such quantitative methods are often more susceptible to experimental imperfections such as B0 and B1-field non-uniformities. Magnetic resonance fingerprinting [9,10,11] may be able to overcome these hurdles by quantifying the B0 and B1 distribution and isolate their contribution from the desired tissue specific parameters [9, 12].

High resolution volumetric imaging in MSK allows retrospective reformatting of the data into any desired plane. However, it can be a challenge to obtain such a 3D dataset with the desired contrast within a clinically acceptable time frame.