Introduction
Magnetic resonance imaging (MRI) of the brain is a state-of-the-art technique for the visualization of a wide variety of neurological and oncological diseases. Using MRI enables precise anatomical delineation and the differentiation of solid components from cystic areas and necrosis, making it superior to other imaging modalities for the diagnostics and monitoring of treatment response in patients of all ages with central nervous system tumors [
1‐
3]; however, optimal image quality is difficult to ensure in children, who have smaller anatomical structures and show more subtle pathological changes that require the use of high spatial resolution. At the same time, the duration of pediatric imaging examination must be as short as possible to minimize motion artifacts and sedation time [
4,
5]. Innovative imaging techniques have been integrated swiftly into pediatric imaging protocols to address these challenges [
6,
7].
Recently introduced as an acceleration technology, compressed sensing is poised to gain a foothold in clinical routines [
8,
9]. It can be combined with parallel imaging techniques, such as sensitivity encoding (SENSE), and is based on variable density sampling and iterative reconstruction to enable higher spatial resolution and shorter scan duration [
6,
10‐
12]. The balancing of these two aims depends on several factors, including acceleration and regularization factors, as well as coil sensitivity [
10,
13,
14]. Several studies have yielded promising results of the combined application of compressed sensing and sensitivity encoding (C-SENSE) in adult populations [
8,
9,
15‐
18], and a few studies have evaluated the application of similar techniques to children, with a primary focus on effects on breathing-dependent scans [
6,
19‐
22].
The purpose of this study was to assess the performance of C‑SENSE as part of a dedicated brain tumor MRI protocol for children. The image quality, examination time, and radiofrequency (RF) energy deposit were assessed.
Discussion
In this study, we applied C‑SENSE to a dedicated pediatric brain tumor MRI protocol and compared image quality, examination times, and energy deposit to those of standard examinations. The results suggest that C‑SENSE helps to provide superior image quality while reducing procedure times and total energy deposit compared with the conventional method.
As a fundamental part of brain tumor MRI, pre-contrast and post-contrast T1-weighted sequences provide information about the general anatomy and tumor-related blood-brain barrier breakdown via contrast enhancement [
23,
24]. A 3D isotropic resolution enables the acquisition of a volumetric dataset and representing all three diagnostically relevant planes with a single scan. The improvement of 3D T1-TFE image quality with C‑SENSE was characterized by the increased sharpness of small structures (i.e., the cerebellum, dura, and intracranial nerves) with no apparent loss of signal or tissue contrast. This property facilitated visual inspection, particularly in contrast-enhancing areas. With the aim of maximizing image quality, implementation of the undersampling and reconstruction algorithm during C‑SENSE 3D T1-TFE sequences reduced the acquisition and reconstruction voxel sizes without loss of the signal-to-noise ratio. This reduction likely contributed to the increased spatial resolution of the sequence.
T2-weighted images help to distinguish between hemorrhage and calcifications, cysts, and solid masses in brain tumor imaging. At the same spatial resolution and with the preservation of contrast, the signal of grey and white matter was slightly more homogeneous on C‑SENSE than on standard T2-TSE images due to the intrinsic denoising capability of C‑SENSE [
6,
10‐
12]. Thus, the readers often preferred the C‑SENSE to the standard T2-TSE images, although image appearance did not differ significantly.
Via CSF suppression, FLAIR images typically aid the detection of vasogenic and cytotoxic edema, gliosis, and gliomatous tumor components. In this study, C‑SENSE had a larger acceleration factor (4.5) than does conventional SENSE (1.8 × 1.3), which led to a slightly noisier image appearance on visual inspection; however, this difference was not deemed impactful for image interpretation. The pseudo-random sampling pattern in the k space of C‑SENSE, in combination with the optimized inversion time, helped to reduce flow-related effects (i.e., pulsation artifacts), and readers preferred C‑SENSE over standard FLAIR images.
Relative to standard examinations, C‑SENSE examinations had reduced diagnostic and total scan times (by 17.6% and 21.1%, respectively), attributable directly to the accelerated performance of the four major scan sequences in the imaging protocol. Although not related directly to scan techniques, the shorter total and between-scan idle times in the C‑SENSE examinations could reflect a shorter duration of imaging volume planning, particularly due to the lack of a coronal T2-TSE sequence in the C‑SENSE examination and could be influenced by differences in operator experience. All of these reductions contributed to the significantly shorter procedural duration of the C‑SENSE examinations, as reflected by the reductions in the total examination and table times (by 15.0% and 12.9%, respectively, relative to standard examination).
The significant decrease in the total energy deposit obtained with C‑SENSE relative to standard examination (by 55.2%) in this study can be attributed to reduced sequence acquisition times and thus the lower sequence-specific SED. The SED reduction also could have been affected by the greater undersampling or scan acceleration achieved with C‑SENSE, leading to an assumed decrease of SAR due to less RF excitation and fewer refocusing pulses or shorter echo trains.
Pediatric brain tumor MRI examinations are often challenging because of poor patient cooperation, the need for additional procedures such as sedation, and patients’ smaller anatomical structures. In general, our findings were consistent with previous brain and abdominal imaging studies conducted with adults [
9,
15,
16] and abdominal imaging studies conducted with children [
6,
19,
20], in which compressed sensing-based technologies were applied to reduce scan times or improve image quality. Regarding the parameter settings of the C‑SENSE protocol, optimization of sequences was conducted during a pilot phase prior to the study, based on our routinely used pediatric brain tumor imaging protocol and existing experience in C‑SENSE applications from literature reports [
8,
9,
15‐
22] as well as other centers. Although this phase was relatively short in order to keep clinical service and patient examinations least disrupted, it represents a typical way for clinical adoption of a new technique in the practice.
The reduction of RF-induced energy in our study is especially advantageous for the examination of sedated or unsedated pediatric patients. To our knowledge, no previous study has examined the amount of RF energy released during pediatric brain tumor MRI examinations. Our findings may help to address concerns about pediatric brain MRI by demonstrating the potential shortening of anesthesia time which could be achieved with C‑SENSE examinations, and which reduces the risks of sedation-related adverse events, airway-related complications, and delayed complications, such as neurotoxicity, particularly in children with severe diseases or disabilities [
31‐
33]. This potential also applies to young patients with brain tumors, who tend to undergo repeated MRI examinations due to the nature of their diseases and surveillance or treatment schemes. The energy deposit reduction may provide a substantial safety benefit for smaller children and newborns, as their limited thermoregulation ability requires careful observation of the RF energy applied during each MRI examination [
34,
35]. In addition, shorter examination and procedure times may improve the cooperation of unsedated children and reduce the number of motion-related artifacts [
7], as well as enabling the economization of the patient care workflow.
This study has several limitations. First, it was performed at a single institution with a relatively small number of patients, which precluded detailed subgroup analysis according to age, body size, or patient cooperation. In addition, the inclusion of patients with limited types of pathology potentially led to selection bias. Second, image analysis was based on expert consensus and thus did not involve total blinding. As our implementation of C‑SENSE in the pediatric brain tumor protocol was performed with the aim of maximizing clinical utility, the spatial resolution and contrast differed from the standards, and experienced readers could easily identify such differences. Third, the image analysis did not include all images from the brain tumor protocol due to the incompatibility of the C‑SENSE software with EPI-based DWI at the time that this study was conducted. Fourth, the limited sample size and the applied scales were not deemed statistically viable for an interrater analysis, as this method is generally applied in large study cohorts. Consensus reading, however, is considered a solid instrument to clinically assess image quality during protocol amendments for smaller patient collectives. Studies including larger cohorts are desirable to further evaluate the full scope of image quality changes through Compressed SENSE. Fifth, the performance of C‑SENSE examinations months after the standard examinations might have led to the introduction of effects due to patient-related or therapy-related changes. Sixth, differences in operator experience are a factor that was not measured in the current study and is difficult to control in clinical practice.
In conclusion, C‑SENSE implementation in this study not only improved image quality and shortened scan times for pediatric brain tumor MRI, but also contributed to a considerable decrease in energy release, thereby addressing a fundamental concern about pediatric MRI; however, further studies are needed to carefully investigate the clinical impacts of acceleration technologies such as C‑SENSE on energy deposit in children.