TrueFisp versus HASTE sequences in 3T cine MRI: Evaluation of image quality during phonation in patients with velopharyngeal insufficiency

A segmentation-informed deep learning framework to register dynamic two-dimensional magnetic resonance images of the vocal tract during speech

Objective

Dynamic magnetic resonance (MR) imaging enables visualisation of articulators during speech. There is growing interest in quantifying articulator motion in two-dimensional MR images of the vocal tract, to better understand speech production and potentially inform patient management decisions. Image registration is an established way to achieve this quantification. Recently, segmentation-informed deformable registration frameworks have been developed and have achieved state-of-the-art accuracy. This work aims to adapt such a framework and optimise it for estimating displacement fields between dynamic two-dimensional MR images of the vocal tract during speech.

Methods

A deep-learning-based registration framework was developed and compared with current state-of-the-art registration methods and frameworks (two traditional methods and three deep-learning-based frameworks, two of which are segmentation informed). The accuracy of the methods and frameworks was evaluated using the Dice coefficient (DSC), average surface distance (ASD) and a metric based on velopharyngeal closure. The metric evaluated if the fields captured a clinically relevant and quantifiable aspect of articulator motion.

Results

The segmentation-informed frameworks achieved higher DSCs and lower ASDs and captured more velopharyngeal closures than the traditional methods and the framework that was not segmentation informed. All segmentation-informed frameworks achieved similar DSCs and ASDs. However, the proposed framework captured the most velopharyngeal closures.

Conclusions

A framework was successfully developed and found to more accurately estimate articulator motion than five current state-of-the-art methods and frameworks.

Significance

The first deep-learning-based framework specifically for registering dynamic two-dimensional MR images of the vocal tract during speech has been developed and evaluated.

Objective and subjective evaluation of Velopharyngeal Dysfunction (VPD) following surgical repair of the cleft palate using the furlow palatoplasty - A new tool

Velopharyngeal dysfunction (VPD) diagnosis and speech surgery outcomes are currently based solely on subjective evaluation criteria consisting of perceptual speech assessment and functional imaging. This study describes an objective and comparable method in VPD assessment and investigates the concurrence between the objective and subjective evaluations. The present study included 20 paediatric patients presenting with VPD after primary repair (intravelar veloplasty) of cleft palate. Our protocol was based on computerized analysis of voice parameters by means of an objective tool, spectrography integrated with Multi-Dimensional Voice Program (MDVP). The protocol also included perceptual evaluation by speech therapist and phoniatrician (consensus listening), and parents. This is a single surgeon, single centre experience and all patients underwent a secondary Furlow's palatoplasty. Assessments were performed pre- and postoperatively and upon completion of speech therapy. Results were compared using the two-tailed t student test for paired data. Statistical significance was set for p-values <0.05. Data analysis confirmed an improvement in velopharyngeal closure after surgery and speech therapy consistently with the results of perceptual evaluations. The results of the study confirmed the availability and reliability of an objective method for VPD evaluation based on the analysis of voice parameters with investigations that are simple and easily available in a hospital setting.

Deep-learning-based segmentation of the vocal tract and articulators in real-time magnetic resonance images of speech

BACKGROUND AND OBJECTIVE

Magnetic resonance (MR) imaging is increasingly used in studies of speech as it enables non-invasive visualisation of the vocal tract and articulators, thus providing information about their shape, size, motion and position. Extraction of this information for quantitative analysis is achieved using segmentation. Methods have been developed to segment the vocal tract, however, none of these also fully segment any articulators. The objective of this work was to develop a method to fully segment multiple groups of articulators as well as the vocal tract in two-dimensional MR images of speech, thus overcoming the limitations of existing methods.

METHODS

Five speech MR image sets (392 MR images in total), each of a different healthy adult volunteer, were used in this work. A fully convolutional network with an architecture similar to the original U-Net was developed to segment the following six regions in the image sets: the head, soft palate, jaw, tongue, vocal tract and tooth space. A five-fold cross-validation was performed to investigate the segmentation accuracy and generalisability of the network. The segmentation accuracy was assessed using standard overlap-based metrics (Dice coefficient and general Hausdorff distance) and a novel clinically relevant metric based on velopharyngeal closure.

RESULTS

The segmentations created by the method had a median Dice coefficient of 0.92 and a median general Hausdorff distance of 5mm. The method segmented the head most accurately (median Dice coefficient of 0.99), and the soft palate and tooth space least accurately (median Dice coefficients of 0.92 and 0.93 respectively). The segmentations created by the method correctly showed 90% (27 out of 30) of the velopharyngeal closures in the MR image sets.

CONCLUSIONS

An automatic method to fully segment multiple groups of articulators as well as the vocal tract in two-dimensional MR images of speech was successfully developed. The method is intended for use in clinical and non-clinical speech studies which involve quantitative analysis of the shape, size, motion and position of the vocal tract and articulators. In addition, a novel clinically relevant metric for assessing the accuracy of vocal tract and articulator segmentation methods was developed.

Real-time dynamic 3-T MRI assessment of spine kinematics: a feasibility study utilizing three different fast pulse sequences

BACKGROUND

Half-Fourier acquisition single-shot turbo spin-echo (HASTE), continuous radial gradient-echo (GRE), and True FISP allow real-time dynamic assessment of the spine.

PURPOSE

To evaluate the feasibility of adding dynamic sequences to routine spine magnetic resonance imaging (MRI) for assessment of spondylolisthesis.

MATERIAL AND METHODS

Retrospective review was performed of patients referred for dynamic MRI of the cervical or lumbar spine between January 2017 and 2018 who had flexion-extension radiographs within two months of MRI. Exclusion criteria were: incomplete imaging; spinal hardware; and inability to tolerate dynamic examination. Blinded, independent review by two board-certified musculoskeletal radiologists was performed to assess for spondylolisthesis (>3 mm translation); consensus review of dynamic radiographs served as the gold standard. Cervical spinal cord effacement was assessed. Inter-reader agreement and radiographic concordance was calculated for each sequence.

RESULTS

Twenty-one patients were included (8 men, 13 women; mean age 47.9 ± 16.5 years). Five had MRI of the cervical spine and 16 had MRI of the lumbar spine. Mean acquisition time was 18.4 ± 1.7 min with dynamic sequences in the range of 58-77 s. HASTE and True FISP had the highest inter-reader reproducibility (κ = 0.88). Reproducibility was better for the lumbar spine (κ = 0.94) than the cervical spine (κ = 0.28). Sensitivity of sequences for spondylolisthesis was in the range of 68.8%-78.6%. All three sequences had high accuracy levels: ≥90.5% averaged across the cervical and lumbar spine. Cervical cord effacement was observed during dynamic MRI in two cases (100% agreement).

CONCLUSION

Real-time dynamic MRI sequences added to spine MRI protocols provide reliable and accurate assessment of cervical and lumbar spine spondylolisthesis during flexion and extension.

Clinical feasibility of 2D dynamic sagittal HASTE flexion-extension imaging of the cervical spine for the assessment of spondylolisthesis and cervical cord impingement

PURPOSE

To assess the utility of a 2D dynamic HASTE sequence in assessment of cervical spine flexion-extension, specifically (1) comparing dynamic spondylolisthesis to radiographs and (2) assessing dynamic contact upon or deformity of the cord.

METHODS

Patients with a dynamic flexion-extension sagittal 2D HASTE sequence in addition to routine cervical spine sequences were identified. Static and dynamic listhesis was first determined on flexion-extension radiographs reviewed in consensus. Blinded assessment of the dynamic HASTE sequence was independently performed by 2 radiologists for (1) listhesis and translation during flexion-extension and (2) dynamic spinal cord impingement (cord contact or deformity between neutral, flexion and extension).

RESULTS

32 scans in 32 patients (9 males, 23 females) met inclusion criteria acquired on 1.5 T (n = 15) and 3 T (n = 17) scanners. The mean acquisition time was 51.8 s (range 20-95 seconds). Dynamic translation was seen in 14 patients on flexion-extension radiographs compared to 12 (reader 1) and 13 (reader 2) patients on HASTE, with 90.6 % agreement (K = 0.83; p = 0.789). In all cases dynamic listhesis was ≤3 mm translation with one patient showing dynamic listhesis in the range 4-6 mm. Four cases (13 %) demonstrated deformity of the cord between flexion-extension, not present in the neutral position. For cord impingement there was strong inter-reader agreement (K = 0.93) and the paired sample Wilcoxon signed rank test found no significant difference between the impingement scores of the two readers (p = 0.787).

CONCLUSIONS

A sagittal dynamic flexion-extension HASTE sequence provides a rapid addition to standard MRI cervical spine protocols, which may useful for assessment of dynamic spondylolisthesis and cord deformity.

Comparison of contrast-enhanced videofluoroscopy to unenhanced dynamic MRI in minor patients following surgical correction of velopharyngeal dysfunction

OBJECTIVES

To compare dynamic magnetic resonance imaging (MRI) with videofluoroscopy (VFS) regarding image quality and assessment of gap size between soft palate (SP) and posterior pharyngeal wall (PPW) in children and adolescents following surgical correction of velopharyngeal dysfunction (VPD).

METHODS

Twenty-one patients undergoing unenhanced 3-T MRI and contrast-enhanced VFS were included in this IRB-approved prospective study. The MRI scan protocol comprised refocused gradient-echo sequences in transverse and sagittal planes during speech, with TE 1.97 ms, TR 3.95 ms, flip angle 8°, matrix size 128 × 128, and 5-mm slice thickness. Radial k-space sampling and sliding window reconstruction were used to achieve an image acquisition rate of 28 frames per second (fps). VFS with 30 fps was similarly performed in both planes. Closure of the velopharyngeal port during phonation was evaluated by two experienced radiologists.

RESULTS

Eleven (52.4%) patients displayed a complete closure, whereas ten (47.6%) patients showed a post-operative gap during speech. VFS and MRI equally identified the cases with persistent or recurrent VPD. Differences in SP-PPW distance between VFS (3.9 ± 1.6 mm) and MRI (4.1 ± 1.5 mm) were not statistically significant (p = 0.5). The subjective overall image quality of MRI was rated inferior (p < 0.001) compared with VFS, with almost perfect inter-rater agreement (κ = 0.90). The presence of susceptibility artifacts did not limit anatomical measurements.

CONCLUSION

Dynamic MRI is equally reliable as VFS to assess persistent or recurrent inadequate velum closure in patients following surgical treatment of VPD.

KEY POINTS

• Unenhanced 3-T dynamic MRI and contrast-enhanced videofluoroscopy are equally useful for the identification of patients with incomplete velopharyngeal closure during speech. • MRI using refocused gradient-echo acquisition with radial k-space sampling and sliding window reconstruction generates diagnostic images with 28 frames per second. • MRI can offer a radiation-free alternative to currently established videofluoroscopy for young patients.

Evaluation of velopharyngeal function using high-speed cine-magnetic resonance imaging based on T2-weighted sequences: a preliminary study

The objective was to introduce a new technique for visualizing the three-dimensional (3D) movements of velopharyngeal-related muscles using high-speed cine-magnetic resonance imaging (MRI) based on T2-weighted sequences. The evaluation of phonation- and water swallowing-related events was performed in 11 healthy subjects. Specifically, whether cine-MRI could precisely visualize normal velopharyngeal function during these two events was examined. The 3D movements of the soft palate, superior pharyngeal constrictor muscles, and levator veli palatini muscles were visualized in all 11 subjects. A noteworthy finding was that the magnetic resonance signals of the superior constrictor pharyngeal muscles and the levator veli palatini muscles were significantly higher during phonation and during water swallowing than at rest. This initial study suggests that the 3D movements of velopharyngeal-related muscles can be successfully and precisely visualized without side effects. The magnetic resonance signal changes seen in the superior pharyngeal constrictor and levator veli palatini muscles using the technique described here should be useful to develop better methods of evaluation of velopharyngeal function.

Dynamic cerebellar herniation in Chiari patients during the cardiac cycle evaluated by dynamic magnetic resonance imaging

PURPOSE

Cerebellar herniation in Chiari patients can be dynamic, following the cerebrospinal fluid pulsatility during the cardiac cycle. We present a voxel intensity distribution method (VIDM) to automatically extract the pulsatility-dependent herniation in time-resolved MRI (CINE MRI) and compare it to the simple linear measurements. The degree of herniation is furthermore compared on CINE and static sequences, and the cerebellar movement is correlated to the presence of hydrocephalus and syringomyelia.

METHODS

The cerebellar movement in 27 Chiari patients is analyzed with VIDM and the results were compared to linear measurements on an image viewer (visual inspection, VI) using a paired t test. Second, an ANOVA test is applied to compare the degree of herniation on static 3D MRI and CINE. Finally, the Pearson's correlation coefficient is calculated for the correlation between cerebellar movement and the presence of hydrocephalus and syringomyelia.

RESULTS

VIDM showed significant movement in 85% of our patients. Assuming that movement < 1 mm cannot be detected reliably on an image viewer, VI identified movement in 29.6% of the patients (p = 0.002). The herniation was greater on static sequences than on CINE in most cases, but this was not statistically significant. The cerebellar movement was not correlated with hydrocephalus or syringomyelia (Pearson's coefficient < 0.3).

CONCLUSIONS

VIDM is a sensitive method to detect tissue movement on CINE MRI and could be used for Chiari patients, but also for the evaluation of cyst membranes, ventriculostomies, etc. The cerebellar movement appears not to correlate with hydrocephalus and syringomyelia in Chiari patients.

Comparison of Cartesian and Non-Cartesian Real-Time MRI Sequences at 1.5T to Assess Velar Motion and Velopharyngeal Closure during Speech

Dynamic imaging of the vocal tract using real-time MRI has been an active and growing area of research, having demonstrated great potential to become routinely performed in the clinical evaluation of speech and swallowing disorders. Although many technical advances have been made in regards to acquisition and reconstruction methodologies, there is still no consensus in best practice protocols. This study aims to compare Cartesian and non-Cartesian real-time MRI sequences, regarding image quality and temporal resolution trade-off, for dynamic speech imaging. Five subjects were imaged at 1.5T, while performing normal phonation, in order to assess velar motion and velopharyngeal closure. Data was acquired using both Cartesian and non-Cartesian (spiral and radial) real-time sequences at five different spatial-temporal resolution sets, between 10 fps (1.7×1.7×10 mm3) and 25 fps (1.5×1.5×10 mm3). Only standard scanning resources provided by the MRI scanner manufacturer were used to ensure easy applicability to clinical evaluation and reproducibility. Data sets were evaluated by comparing measurements of the velar structure, dynamic contrast-to-noise ratio and image quality visual scoring. Results showed that for all proposed sequences, FLASH spiral acquisitions provided higher contrast-to-noise ratio, up to a 170.34% increase at 20 fps, than equivalent bSSFP Cartesian acquisitions for the same spatial-temporal resolution. At higher frame rates (22 and 25 fps), spiral protocols were optimal and provided higher CNR and visual scoring than equivalent radial protocols. Comparison of dynamic imaging at 10 and 22 fps for radial and spiral acquisitions revealed no significant difference in CNR performance, thus indicating that temporal resolution can be doubled without compromising spatial resolution (1.9×1.9 mm2) or CNR. In summary, this study suggests that the use of FLASH spiral protocols should be preferred over bSSFP Cartesian for the dynamic imaging of velopharyngeal closure, as it allows for an improvement in CNR and overall image quality without compromising spatial-temporal resolution.