Method for automatic localization of MR-visible markers using morphological image processing and conventional pulse sequences: Feasibility for image-guided procedures

Automated fiducial marker detection and localization in volumetric computed tomography images: a three-step hybrid approach with deep learning

Purpose: Automating fiducial detection and localization in the patient’s pre-operative images can lead to better registration accuracy, reduced human errors, and shorter intervention time. Most current approaches are optimized for a single marker type, mainly spherical adhesive markers. A fully automated algorithm is proposed and evaluated for screw and spherical titanium fiducials, typically used in high-accurate frameless surgical navigation.

Approach: The algorithm builds on previous approaches with morphological functions and pose estimation algorithms. A 3D convolutional neural network (CNN) is proposed for the fiducial classification task and evaluated for both traditional closed-set and emerging open-set classifiers. A proposed digital ground-truth experiment, with cone-beam computed tomography (CBCT) imaging software, is performed to determine the localization accuracy of the algorithm. The localized fiducial positions in the CBCT images by the presented algorithm were compared to the actual known positions in the virtual phantom models. The difference represents the fiducial localization error (FLE).

Results: A total of 241 screws, 151 spherical fiducials, and 1550 other structures are identified with the best true positive rate 95.9% for screw and 99.3% for spherical fiducials at 8.7% and 3.4% false positive rate, respectively. The best achieved FLE mean and its standard deviation for a screw and spherical marker are 58 (14) and 14 (6)  μm, respectively.

Conclusions: Accurate marker detection and localization were achieved, with spherical fiducials being superior to screws. Large marker volume and smaller voxel size yield significantly smaller FLEs. Attenuating noise by mesh smoothing has a minor effect on FLE. Future work will focus on expanding the CNN for image segmentation.

Techniques for Interventional MRI Guidance in Closed-Bore Systems

Efficient image guidance is the basis for minimally invasive interventions. In comparison with X-ray, computed tomography (CT), or ultrasound imaging, magnetic resonance imaging (MRI) provides the best soft tissue contrast without ionizing radiation and is therefore predestined for procedural control. But MRI is also characterized by spatial constraints, electromagnetic interactions, long imaging times, and resulting workflow issues. Although many technical requirements have been met over the years-most notably magnetic resonance (MR) compatibility of tools, interventional pulse sequences, and powerful processing hardware and software-there is still a large variety of stand-alone devices and systems for specific procedures only.Stereotactic guidance with the table outside the magnet is common and relies on proper registration of the guiding grids or manipulators to the MR images. Instrument tracking, often by optical sensing, can be added to provide the physicians with proper eye-hand coordination during their navigated approach. Only in very short wide-bore systems, needles can be advanced at the extended arm under near real-time imaging. In standard magnets, control and workflow may be improved by remote operation using robotic or manual driving elements.This work highlights a number of devices and techniques for different interventional settings with a focus on percutaneous, interstitial procedures in different organ regions. The goal is to identify technical and procedural elements that might be relevant for interventional guidance in a broader context, independent of the clinical application given here. Key challenges remain the seamless integration into the interventional workflow, safe clinical translation, and proper cost effectiveness.

Navigated MRI-guided liver biopsies in a closed-bore scanner: experience in 52 patients

OBJECTIVES

To evaluate clinical effectiveness and diagnostic efficiency of a navigation device for MR-guided biopsies of focal liver lesions in a closed-bore scanner.

METHODS

In 52 patients, 55 biopsies were performed. An add-on MR navigation system with optical instrument tracking was used for image guidance and biopsy device insertion outside the bore. Fast control imaging allowed visualization of the true needle position at any time. The biopsy workflow and procedure duration were recorded. Histological analysis and clinical course/outcome were used to calculate sensitivity, specificity and diagnostic accuracy.

RESULTS

Fifty-four of 55 liver biopsies were performed successfully with the system. No major and four minor complications occurred. Mean tumour size was 23 ± 14 mm and the skin-to-target length ranged from 22 to 177 mm. In 39 cases, access path was double oblique. Sensitivity, specificity and diagnostic accuracy were 88 %, 100 % and 92 %, respectively. The mean procedure time was 51 ± 12 min, whereas the puncture itself lasted 16 ± 6 min. On average, four control scans were taken.

CONCLUSIONS

Using this navigation device, biopsies of poorly visible and difficult accessible liver lesions could be performed safely and reliably in a closed-bore MRI scanner. The system can be easily implemented in clinical routine workflow.

KEY POINTS

• Targeted liver biopsies could be reliably performed in a closed-bore MRI. • The navigation system allows for image guidance outside of the scanner bore. • Assisted MRI-guided biopsies are helpful for focal lesions with a difficult access. • Successful integration of the method in clinical workflow was shown. • Subsequent system installation in an existing MRI environment is feasible.

Targeting Accuracy, Procedure Times and User Experience of 240 Experimental MRI Biopsies Guided by a Clinical Add-On Navigation System

OBJECTIVES

MRI is of great clinical utility for the guidance of special diagnostic and therapeutic interventions. The majority of such procedures are performed iteratively ("in-and-out") in standard, closed-bore MRI systems with control imaging inside the bore and needle adjustments outside the bore. The fundamental limitations of such an approach have led to the development of various assistance techniques, from simple guidance tools to advanced navigation systems. The purpose of this work was to thoroughly assess the targeting accuracy, workflow and usability of a clinical add-on navigation solution on 240 simulated biopsies by different medical operators.

METHODS

Navigation relied on a virtual 3D MRI scene with real-time overlay of the optically tracked biopsy needle. Smart reference markers on a freely adjustable arm ensured proper registration. Twenty-four operators - attending (AR) and resident radiologists (RR) as well as medical students (MS) - performed well-controlled biopsies of 10 embedded model targets (mean diameter: 8.5 mm, insertion depths: 17-76 mm). Targeting accuracy, procedure times and 13 Likert scores on system performance were determined (strong agreement: 5.0).

RESULTS

Differences in diagnostic success rates (AR: 93%, RR: 88%, MS: 81%) were not significant. In contrast, between-group differences in biopsy times (AR: 4:15, RR: 4:40, MS: 5:06 min:sec) differed significantly (p<0.01). Mean overall rating was 4.2. The average operator would use the system again (4.8) and stated that the outcome justifies the extra effort (4.4). Lowest agreement was reported for the robustness against external perturbations (2.8).

CONCLUSIONS

The described combination of optical tracking technology with an automatic MRI registration appears to be sufficiently accurate for instrument guidance in a standard (closed-bore) MRI environment. High targeting accuracy and usability was demonstrated on a relatively large number of procedures and operators. Between groups with different expertise there were significant differences in experimental procedure times but not in the number of successful biopsies.

Prospective real-time head motion correction using inductively coupled wireless NMR probes

PURPOSE

Head motion continues to be a major source of artifacts and data quality degradation in MRI. The goal of this work was to develop and demonstrate a novel technique for prospective, 6 degrees of freedom (6DOF) rigid body motion estimation and real-time motion correction using inductively coupled wireless nuclear magnetic resonance (NMR) probe markers.

METHODS

Three wireless probes that are inductively coupled with the scanner's RF setup serve as fiducials on the subject's head. A 12-ms linear navigator module is interleaved with the imaging sequence for head position estimation, and scan geometry is updated in real time for motion compensation. Flip angle amplification in the markers allows the use of extremely small navigator flip angles (∼1°). A novel algorithm is presented to identify marker positions in the absence of marker specific receive channels. Motion correction is demonstrated in high resolution 2D and 3D gradient recalled echo experiments in a phantom and humans.

RESULTS

Significant improvement of image quality is demonstrated in phantoms and human volunteers under different motion conditions.

CONCLUSION

A novel real-time 6DOF head motion correction technique based on wireless NMR probes is demonstrated in high resolution imaging at 7 Tesla.

An MRI-powered and controlled actuator technology for tetherless robotic interventions

This paper presents a novel actuation technology for robotically assisted MRI-guided interventional procedures. In the proposed approach, the MRI scanner is used to deliver power, estimate actuator state and perform closed-loop control. The actuators themselves are compact, inexpensive and wireless. Using needle driving as an example application, actuation principles and force production capabilities are examined. Actuator stability and performance are analyzed for the two cases of state estimation at the input versus the output of the actuator transmission. Closed-loop needle position control is achieved by interleaving imaging pulse sequences to estimate needle position (transmission output estimation) and propulsion pulse sequences to drive the actuator. A prototype needle driving robot is used to validate the proposed approach in a clinical MRI scanner.

Catheter tip tracking for MR-guided interventions using discrete Kalman filter and mean shift localization

PURPOSE

This paper presents and evaluates stochastic computer algorithms used to automatically detect and track marked catheter tip during MR-guided catheterization. The algorithms developed employ extraction and matching of regional features of the catheter tip to perform the localization.

METHOD

To perform the tracking, a probability map that indicates the possible locations of the catheter tip in the MR images is first generated. This map is generated from the similarity to a given marker template. The method to assess the similarity between the marker template image and the different positions on each MR frame is based on speeded-up robust features extracted from the gradient image. The probability map is then used in two different stochastic localization frameworks mean shift (MS) localization and Kalman filter (KF) to track the position of the catheter using pairs of orthogonal projection of 2D MR images. The algorithm developed was tested on catheter tip marked with LC resonant circuit (of size 2 mm x 2 cm) tuned to the Larmor frequency of the MRI scanner and for different image resolutions (1, 3, 5 and 7 mm squared pixel).

RESULTS

The tracking performance was very robust for the two algorithms MS and KF with image resolution as low as 3 mm where the localization error was about 1 mm for KF and 0.9 mm for MS. For the 5-mm resolution images, the error was 2.2 mm for both KF and MS, and for the 7-mm resolution images, the error was 3.6 and 3.7 mm for KF and MS, respectively.

CONCLUSION

Both KF and MS gave comparable results when it comes to accuracy for the different image resolutions. The results showed that the two tracking algorithms tracked the catheter tip with high robustness for image resolution of 3 mm and with acceptable reliability for image resolution as poor as 5 mm with the resonant marker configuration used.

Suitability of miniature inductively coupled RF coils as MR-visible markers for clinical purposes

PURPOSE

MR-visible markers have already been used for various purposes such as image registration, motion detection, and device tracking. Inductively coupled RF (ICRF) coils, in particular, provide a high contrast and do not require connecting wires to the scanner, which makes their application highly flexible and safe. This work aims to thoroughly characterize the MR signals of such ICRF markers under various conditions with a special emphasis on fully automatic detection.

METHODS

The small markers consisted of a solenoid coil that was wound around a glass tube containing the MR signal source and tuned to the resonance frequency of a 1.5 T MRI. Marker imaging was performed with a spoiled gradient echo sequence (FLASH) and a balanced steady-state free precession (SSFP) sequence (TrueFISP) in three standard projections. The signal intensities of the markers were recorded for both pulse sequences, three source materials (tap water, distilled water, and contrast agent solution), different flip angles and coil alignments with respect to the B(0) direction as well as for different marker positions in the entire imaging volume (field of view, FOV). Heating of the ICRF coils was measured during 10-min RF expositions to three conventional pulse sequences. Clinical utility of the markers was assessed from their performance in computer-aided detection and in defining double oblique scan planes.

RESULTS

For almost the entire FOV (±215 mm) and an estimated 82% of all possible RF coil alignments with respect to B(0), the ICRF markers generated clearly visible MR signals and could be reliably localized over a large range of flip angles, in particular with the TrueFISP sequence (0.3°-4.0°). Generally, TrueFISP provided a higher marker contrast than FLASH. RF exposition caused a moderate heating (≤5 °C) of the ICRF coils only.

CONCLUSIONS

Small ICRF coils, imaged at low flip angles with a balanced SSFP sequence showed an excellent performance under a variety of experimental conditions and therefore make for a reliable, compact, flexible, and relatively safe marker for clinical use.

Tracking and position control of an MRI-powered needle-insertion robot

The excellent imaging capabilities of MRI technology are standardizing this modality for a variety of interventional procedures. To assist radiologists, MRI compatible robots relying on traditional actuation technologies are being developed. Recently, a robot that is not only MRI compatible but also MRI powered was introduced. This surgical robot is imaged and actuated through interleaved MRI pulses, and can be controlled to perform automated needle insertion. Using the electromagnetic field generated by the MRI scanner, the robot can exercise adequate forces to puncture tissue. Towards the goal of automation, this paper reports results on tracking and control of an MRI-powered robot tagged with a fiducial marker. Tracking is achieved using non-selective RF pulses and balanced gradient readouts. To suppress the signal received from the tissue, spoiler gradients and background suppression are introduced. Their effects on tracking are quantified and are used to optimize the algorithm. Subsequently, a Kalman filter is employed for robustness. The developed algorithm is used to demonstrate position controlled needle insertion ex vivo.

MR-guided liver tumor ablation employing open high-field 1.0T MRI for image-guided brachytherapy

OBJECTIVE

To determine the feasibility and safety of image-guided brachytherapy employing a modified open high-field MR system.

METHODS

This is a follow-up study of a development project enabling technologies for interventional use of 1.0T open MRI. Modifications included coils and in-bore visualization, fluoroscopic sequences and user interfaces. We recruited 104 patients with 224 liver malignancies to receive MR-guided brachytherapy. Interventions were performed >20 min after Gd-EOB-DTPA. We recorded interventional parameters including the intervention time (from acquisition of the first scout until the final sequence for brachytherapy treatment planning). Two reviewers assessed MR-fluoroscopic images in comparison to plain CT as used in CT intervention, applying a rating scale of 1-10. Statistical analysis included Friedman and Kendall's W tests.

RESULTS

We employed freehand puncture with interactive dynamic imaging for navigation. Technical success rate was 218 complete ablations in 224 tumours (97%). The median intervention time was 61 min. We recorded no adverse events related to the use of MRI. No major complications occurred. The rate of minor complications was 4%. Local control at 3 months was 96%. Superiority of MR-fluoroscopic, Gd-EOB-DTPA-enhanced images over plain CT was highly significant (P < 0.001).

CONCLUSION

MR-guided brachytherapy employing open high-field MRI is feasible and safe.