Radial k-space filling technique: Beautiful, motion free MRI Scans

One of the major challenges in Magnetic Resonance Imaging (MRI) is its inherent sensitivity to respiratory and physiological motion, which can lead to artifacts and degraded image quality. These types of artifacts are more pronounced on uncooperative or debilitated patients, like the elderly and children, or also on demanding body regions, such as abdomen and head/neck (fig. 1).

Fig. 1 Severe motion artifacts on uncooperative patient (A) and challenging body region (B). Motion artifacts degrade the image quality and may obscure pathology.

Traditional Cartesian acquisition schemes are very sensitive to motion and ghosting artifacts, hence non-Cartesian acquisition schemes, like radial technique, can be used to overcome these limitations (fig. 2).

Fig. 2 Cartesian 2D T2-w FLAIR (A) and non-Cartesian (Radial) 2D T2-w FLAIR (B). Cartesian acquisition suffers from severe motion artifacts, while Radial acquisition easily highlights the cerebellar tumor, offering diagnostic imaging. The image dataset was acquired at 1.5 Tesla.

Technical Aspects

JG Pipe developed the radial k-space filling technique in the late 90s. The basic idea was to sample the k-space in a rotating fashion using a set of radial strips or blades (fig. 3). The center of the k-space is oversampled, providing redundancy of information, which means that the data for each new blade can be compared to the data from previous blades for consistency and motion artifact reduction. Radial reconstruction algorithm needs 8-32 blade lines, which are obtained within a TR.

Fig. 3 Rectilinear (A) and radial (B) k-space filling techniques. In radial technique, blades (or radial stripes) are obtained along the phase and frequency encoding direction in a rotating fashion.


Radial technique offers numerous advantages, making feasible the MR Imaging of uncooperative patients and demanding body regions. Specifically:

  1. It provides robustness to motion (fig. 2).

  2. It continuously updates the center of k-space, which leads to increased SNR and CNR (center of k-space contains the highest signal amplitude and contributes most to image contrast).

  3. It reduces the flow artifacts (fig. 4).

  4. It can help greatly reduce susceptibility artifact (fig. 5).

  5. Many papers have established that Radial acquisition technique has proved to be advantageous in reducing truncation, cross-talk and chemical shift artifacts.

Fig. 4 Cartesian (A) and non-Cartesian T2-w (B) acquisitions on a patient with hydrocephalus. Note that standard rectilinear technique (A) suffers from pulsation artifacts (red arrows), while Radial technique (B) provides motion free brain imaging.

Fig. 5 Non-Cartesian (Radial) T2-w (A), 3D SWI (B) and SS-EPI T2*-W GRE (C) acquisitions. The presence of a metallic object (ferromagnetic or not) causes large distortions in the magnetic field and significant susceptibility artifacts. The range of signal loss depends on the type of metal, on the field strength and on the pulse sequence (spin echo, gradient echo, echo planar imaging, radial). Radial technique can greatly reduce the susceptibility artifacts, offering high-quality imaging.


Unfortunately, Radial technique has some drawbacks, like every other technique in MRI. Specifically:

  1. It increases the acquisition time, if high-quality images without streaking artifacts are needed.

  2. Image contrast is worst, when just 100% blade coverage factor is used (fig. 6). Complete coverage of the k-space (without gaps between the blades) is achieved with a blade coverage factor of 157%.

  3. Current 2D Radial techniques correct only for in-plane motion.

  4. 3D Acquisition is available only from one vendor (see table 1).

  5. 5. Radial undersampling creates streaking artifacts​ (fig 7)

Fig. 6 Cartesian (A) and non-Cartesian FLAIR (B) acquisitions on a patient with brain metastasis (red arrows). Radial technique theoretically provides higher SNR and CNR in comparison with rectilinear (Cartesian) technique. Although, note that the lesion is better highlighted on image (A). That's because just 100% blade coverage factor was used for non-Cartesian acquisition.

Fig. 7 Radial acquisitions with phase oversampling of 0% (A) and 50% (B). Note the streaking artifacts (red arrows) on image (A) due to radial undersampling. Increasing the phase oversampling to 50% results in acquisition without streaking artifacts. Large FOV, high phase oversampling and at least 100% blade coverage factor lead to images with fewer streaking artifacts. Image courtesy of Bac Nguyen.

Finally, because each vendor uses a different name for the Radial technique, Table 1 provides cross-vendor comparison.

Table 1 Vendor-specific acronyms for the available Radial techniques.

#MRIScanning #blade #motion #motionartifact #artifact #kspace #cartesian #noncartesian #radial #radialimaging