Machine learning for automated 3-dimensional segmentation of the spine and suggested placement of pedicle screws based on intraoperative cone-beam computer tomography.

OBJECTIVE: The goal of this study was to develop and validate a system for automatic segmentation of the spine, pedicle identification, and screw path suggestion for use with an intraoperative 3D surgical navigation system. METHODS: Cone-beam CT (CBCT) images of the spines of 21 cadavers were obtained. An automated model-based approach was used for segmentation. Using machine learning methodology, the algorithm was trained and validated on the image data sets. For measuring accuracy, surface area errors of the automatic segmentation were compared to the manually outlined reference surface on CBCT. To further test both technical and clinical accuracy, the algorithm was applied to a set of 20 clinical cases. The authors evaluated the system's accuracy in pedicle identification by measuring the distance between the user-defined midpoint of each pedicle and the automatically segmented midpoint. Finally, 2 independent surgeons performed a qualitative evaluation of the segmentation to judge whether it was adequate to guide surgical navigation and whether it would have resulted in a clinically acceptable pedicle screw placement. RESULTS: The clinically relevant pedicle identification and automatic pedicle screw planning accuracy was 86.1%. By excluding patients with severe spinal deformities (i.e., Cobb angle > 75° and severe spinal degeneration) and previous surgeries, a success rate of 95.4% was achieved. The mean time (± SD) for automatic segmentation and screw planning in 5 vertebrae was 11 ± 4 seconds. CONCLUSIONS: The technology investigated has the potential to aid surgeons in navigational planning and improve surgical navigation workflow while maintaining patient safety.

Automatic bone removal for 3D TACE planning with C-arm CBCT: Evaluation of technical feasibility.

PURPOSE: To evaluate the technical feasibility of automatically removing the ribs and spine from C-arm cone-beam computed tomography (CBCT) images acquired during transcatheter arterial chemoembolization (TACE). MATERIAL AND METHODS: Fifty-eight patients (45.8 ± 5.0 years) with unresectable hepatocellular carcinoma (HCC) underwent transcatheter arterial chemoembolization and had intraprocedural CBCT imaging. Automatic bone removal was performed using model-based segmentation of the ventral cavity. Two interventional radiologists independently evaluated the performance of bone removal, remaining soft tissue retention, and general usability (where both the bone is appropriately removed while retaining soft tissue) for 3D TACE planning on a four-level (complete/excellent, adequate/good, incomplete/questionable, insufficient/bad) score. The proportion of inter-reader agreement was calculated. RESULTS: For ribs and spine removal, 98.3-100% and 100% of cases showed complete or adequate performance, respectively. In 96.6% of the cases, soft tissue was at least adequately retained. 91.3-93.1% of the cases demonstrated good or excellent general usability for TACE planning. Satisfactory inter-reader agreement proportion was achieved in ribs (93.1%) and spine removal (89.7%), soft tissue retention (84.5%), and general usability for TACE planning (72.4%). CONCLUSION: Intraprocedural automatic bone removal on CBCT images is technically feasible and offers good removal of ribs and spine while preserving soft tissue. Its clinical value needs further assessment.

Multimodality Imaging of Ethiodized Oil-loaded Radiopaque Microspheres during Transarterial Embolization of Rabbits with VX2 Liver Tumors.

Purpose To assess the visibility of radiopaque microspheres during transarterial embolization (TAE) in the VX2 rabbit liver tumor model by using multimodality imaging, including single-snapshot radiography, cone-beam computed tomography (CT), multidetector CT, and micro-CT. Materials and Methods The study was approved by the institutional animal care and use committee. Fifteen VX2-tumor-bearing rabbits were assigned to three groups depending on the type of embolic agent injected: 70-150-µm radiopaque microspheres in saline (radiopaque microsphere group), 70-150-µm radiopaque microspheres in contrast material (radiopaque microsphere plus contrast material group), and 70-150-µm radiolucent microspheres in contrast material (nonradiopaque microsphere plus contrast material group). Rabbits were imaged with single-snapshot radiography, cone-beam CT, and multidetector CT. Three to 5 weeks after sacrifice, excised livers were imaged with micro-CT and histologic analysis was performed. The visibility of the embolic agent was assessed with all modalities before and after embolization by using a qualitative three-point scale score reading study and a quantitative assessment of the signal-to-noise ratio (SNR) change in various regions of interest, including the tumor and its feeding arteries. The Kruskal-Wallis test was used to compare the rabbit characteristics across groups, and the Wilcoxon signed rank test was used to compare SNR measurements before and after embolization. Results Radiopaque microspheres were qualitatively visualized within tumor feeding arteries and targeted tissue with all imaging modalities (P < .05), and their presence was confirmed with histologic examination. SNRs of radiopaque microsphere deposition increased after TAE on multidetector CT, cone-beam CT, and micro-CT images (P < .05). Similar results were obtained when contrast material was added to radiopaque microspheres, except for additional image attenuation due to tumor enhancement. For the group with nonradiopaque microspheres and contrast material, retained tumoral contrast remained qualitatively visible with all modalities except for micro-CT, which demonstrated soluble contrast material washout over time. Conclusion Radiopaque microspheres were visible with all imaging modalities and helped increase conspicuity of the tumor as well as its feeding arteries after TAE in a rabbit VX2 liver tumor model. (©) RSNA, 2015.

C-arm cone-beam computed tomography in interventional oncology: technical aspects and clinical applications.

C-arm cone-beam computed tomography (CBCT) is a new imaging technology integrated in modern angiographic systems. Due to its ability to obtain cross-sectional imaging and the possibility to use dedicated planning and navigation software, it provides an informed platform for interventional oncology procedures. In this paper, we highlight the technical aspects and clinical applications of CBCT imaging and navigation in the most common loco-regional oncological treatments.

Image quality improvements in C-Arm CT (CACT) for liver oncology applications: preliminary study in rabbits.

INTRODUCTION: C-Arm CT (CACT) is a new imaging modality in liver oncology therapy that allows for the acquisition of 3D images intra-procedurally. CACT has been used to enhance intra-arterial therapies for the liver by improving lesion detection, avoiding non-target embolization, and allowing for more selective delivery of agents. However, one of the limitations of this technology is image artifacts created by respiratory motion. PURPOSE: To determine in this preliminary study improvements in image acquisition, motion compensation, and high resolution 3D reconstruction that can improve CACT image quality (IQ). MATERIAL AND METHODS: Three adult male New Zealand white rabbits were used for this study. First, a control rabbit was used to select the best x-ray acquisition imaging protocol and then two rabbits were implanted with liver tumor to further develop 3D image reconstruction and motion compensation algorithms. RESULTS: The best IQ was obtained using the low 80 kVp protocol with motion compensated reconstruction with high resolution and fast acquisition speed (60 fps, 5 s/scan, and 312 images). CONCLUSION: IQ improved by: (1) decreasing acquisition time, (2) applying motion-compensated reconstruction, and (3) high resolution 3D reconstruction. The findings of this study can be applied to future animal studies and eventually could be translated into the clinical environment.

High-quality 3-D coronary artery imaging on an interventional C-arm x-ray system.

PURPOSE: Three-dimensional (3-D) reconstruction of the coronary arteries during a cardiac catheter-based intervention can be performed from a C-arm based rotational x-ray angiography sequence. It can support the diagnosis of coronary artery disease, treatment planning, and intervention guidance. 3-D reconstruction also enables quantitative vessel analysis, including vessel dynamics from a time-series of reconstructions. METHODS: The strong angular undersampling and motion effects present in gated cardiac reconstruction necessitate the development of special reconstruction methods. This contribution presents a fully automatic method for creating high-quality coronary artery reconstructions. It employs a sparseness-prior based iterative reconstruction technique in combination with projection-based motion compensation. RESULTS: The method is tested on a dynamic software phantom, assessing reconstruction accuracy with respect to vessel radii and attenuation coefficients. Reconstructions from clinical cases are presented, displaying high contrast, sharpness, and level of detail. CONCLUSIONS: The presented method enables high-quality 3-D coronary artery imaging on an interventional C-arm system.

Clinical feasibility of a fully automated 3D reconstruction of rotational coronary X-ray angiograms.

BACKGROUND: Although fixed view x-ray angiography remains the primary technique for anatomic imaging of coronary artery disease, the known shortcomings of 2D projection imaging may limit accurate 3D vessel and lesion definition and characterization. A recently developed method to create 3D images of the coronary arteries uses x-ray projection images acquired during a 180 degrees C-arm rotation and continuous contrast injection followed by ECG-gated iterative reconstruction. This method shows promise for providing high-quality 3D reconstructions of the coronary arteries with no user interaction but requires clinical evaluation. METHODS AND RESULTS: The reconstruction strategy was evaluated by comparing the reconstructed 3D volumetric images with the 2D angiographic projection images from the same 23 patients to ascertain overall image quality, lesion visibility, and a comparison of 3D quantitative coronary analysis with 2D quantitative coronary analysis. The majority of the resulting 3D volume images were rated as having high image quality (66%) and provided the physician with additional clinical information such as complete visualization of bifurcations and unobtainable views of the coronary tree. True-positive lesion detection rates were high (90 to 100%), whereas false-positive detection rates were low (0 to 8.1%). Finally, 3D quantitative coronary analysis showed significant similarity with 2D quantitative coronary analysis in terms of lumen diameters and provided vessel segment length free from the errors of foreshortening. CONCLUSIONS: Fully automated reconstruction of rotational coronary x-ray angiograms is feasible, produces 3D volumetric images that overcome some of the limitations of standard 2D angiography, and is ready for further implementation and study in the clinical environment.

Optimization of acquisition trajectories for 3D rotational coronary venography.

OBJECTIVE: Rotational coronary X-ray imaging on C-arm systems provides a multitude of diagnostic projections from the vascular tree with a single contrast agent bolus. The acquisition trajectory is typically limited to a circular arc with a fixed caudo-cranial angulation. This may cause sub- optimal projection directions for specific vessel segments for all acquired views, e.g., those segments orthogonal to the axis of rotation. In this paper, a method is presented to calculate a patient-independent acquisition trajectory with respect to vessel foreshortening and overlap for multiple vessel segments of the coronary tree. This method can be applied to artery as well as vein anatomy. METHODS: Rotational coronary venograms of 14 patients have been used to generate three-dimensional mesh representations with a semi-automatic two view modeling algorithm. The venous tree is divided into seven different vessel segments. Foreshortening and overlap of every segment are calculated and combined for all patients in a measure called obstruction value. The weighted obstruction values of all vessel segments define a cost function for the entire two-dimensional angular range of the C-arm system. Viterbi's algorithm is used to calculate an optimal trajectory with respect to this cost function. The method is validated by leave-one-out cross-validation on the 14 rotational venography data sets and on simulated venograms of a segmented computed tomography (CT) data set. Projection images with a foreshortening value below 10% and overlap below 20% are rated 'optimal'. RESULTS: In 12 (85.7%) data sets, 43% more optimal images were acquired using the presented method compared to the standard circular arc trajectory. As well, in 13 (92.8%) data sets 38% more vessel segments can be optimally visualized in the acquired images. The test on the CT data set showed that the resulting average root-mean-square error of the extracted centerline points of the segmented CT data set compared to the error based on the views from the circular arc was reduced from 2.52 to 1.55 mm. CONCLUSION: In a first test, the method proved to deliver improved image quality by reducing foreshortening and overlap of vessel segments and may therefore also improve the centerline extraction accuracy of the semi-automatic two view modeling method.

Improvement of cardiac CT reconstruction using local motion vector fields.

The motion of the heart is a major challenge for cardiac imaging using CT. A novel approach to decrease motion blur and to improve the signal to noise ratio is motion compensated reconstruction which takes motion vector fields into account in order to correct motion. The presented work deals with the determination of local motion vector fields from high contrast objects and their utilization within motion compensated filtered back projection reconstruction. Image registration is applied during the quiescent cardiac phases. Temporal interpolation in parameter space is used in order to estimate motion during strong motion phases. The resulting motion vector fields are during image reconstruction. The method is assessed using a software phantom and several clinical cases for calcium scoring. As a criterion for reconstruction quality, calcium volume scores were derived from both, gated cardiac reconstruction and motion compensated reconstruction throughout the cardiac phases using low pitch helical cone beam CT acquisitions. The presented technique is a robust method to determine and utilize local motion vector fields. Motion compensated reconstruction using the derived motion vector fields leads to superior image quality compared to gated reconstruction. As a result, the gating window can be enlarged significantly, resulting in increased SNR, while reliable Hounsfield units are achieved due to the reduced level of motion artefacts. The enlargement of the gating window can be translated into reduced dose requirements.

Adsorption and co-adsorption of ethylene and carbon monoxide on silica-supported monodisperse Pt nanoparticles: volumetric adsorption and infrared spectroscopy studies.

The adsorption of carbon monoxide and ethylene, and their sequential adsorption, was studied over a series of Pt/SBA-15 catalysts with monodisperse particle sizes ranging from 1.7 to 7.1 nm by diffuse-reflectance infrared spectroscopy and chemisorption. Gas adsorption was dependent on the Pt particle size, temperature, and sequence of gas exposure. Adsorption of CO at room temperature on Pt/SBA-15 gives rise to a spectroscopic feature assigned to the C-O stretch: nu(CO) = 2075 cm-1 (1.9 nm); 2079 cm-1 (2.9 nm); 2082 cm-1 (3.6 nm); and 2090 cm-1 (7.1 nm). The intensity of the signal decreased in a sigmoidal fashion with increasing temperature, thereby providing semiquantitative surface coverage information. Adsorption of ethylene on Pt/SBA-15 gave rise to spectroscopic features at approximately 1340, approximately 1420, and approximately 1500 cm-1 assigned to ethylidyne, di-sigma-bonded ethylene, and pi-bonded ethylene, respectively. The ratio of these surface species is highly dependent on the Pt particle size. At room temperature, Pt particles stabilize ethylidyne as well as di-sigma- and pi-bonded ethylene; however, ethylidyne predominated on the surfaces of larger particles. Ethylidyne was the only identifiable species at 403 K, with its formation being more facile on larger particles. Co-adsorption experiments reveal that the composition of the surface layer is dependent on the order of exposure to gases. Exposure of a C2H4-covered Pt surface to CO resulted in an approximately 50% decrease in chemisorbed CO compared to a fresh Pt surface. The nu(CO) appeared at 2050 cm-1 on Pt/SBA-15 pretreated with C2H4 at room temperature. The di-sigma-bonded and pi-bonded species are the most susceptible to displacement from the surface by CO. The formation of ethylidyne appeared to be less sensitive to the presence of adsorbed carbon monoxide, especially on larger particles. Upon exposure of C2H4 to a CO-covered Pt surface, little irreversible uptake occurred due to nearly 100% site blocking. These results demonstrate that carbon monoxide competes directly with ethylene for surface sites, which will have direct implications on the poisoning of the heterogeneously catalyzed conversion of hydrocarbons.

Evaluation of the influence of acquisition and reconstruction parameters for 16-row multidetector CT on coronary calcium scoring using a stationary and dynamic cardiac phantom.

A calcium-scoring phantom with hydroxyapatite-filled cylindrical holes (0.5 to 4 mm) was used. High-resolution scans were performed for an accuracy baseline. The phantom was mounted to a moving heart phantom. Non-moving data with the implementation of an ECG-signal were acquired for different pitches (0.2/0.3), heart rates (60/80/95 bpm) and collimations (16 x 0.75/16 x 1.5 mm). Images were reconstructed with a cone-beam multi-cycle algorithm at a standard thickness/increment of 3 mm/1.5 mm and the thinnest possible thickness (0.8/0.4 and 2/1). Subsequently, ECG-gated moving calcium-scoring phantom data were acquired. The calcium volume and Agatston score were measured. The temporal resolution and reconstruction cycles were calculated. High-resolution scans determine the calcium volume with a high accuracy (mean overestimation, 0.8%). In the non-moving measurements, the volume underestimation ranged from about 6% (16 x 0.75 mm; 0.8/0.4 mm) to nearly 25% (16 x 1.5 mm; 3/1.5 mm). Moving scans showed increased measurement errors depending on the reconstructed RR interval, collimation, pitch, heart rate and gantry rotation time. Also, a correlation with the temporal resolution could be found. The reliability of calcium-scoring results can be improved with the use of a narrower collimation, a lower pitch and the reconstruction of thinner images, resulting in higher patient doses. The choice of the correct cardiac phase within the RR interval is essential to minimize measurement errors.

Real-time adaptive filtering for projection reconstruction MR fluoroscopy.

Magnetic resonance (MR) imaging has faced a dramatic increase in real-time capabilities over the last years. However, the application of fast pulse sequences still suffers from low signal-to-noise ratios (SNRs), which can be the limiting factor for the actual acquisition speed. In MR fluoroscopy, filtering along the time and/or spatial domain can be applied to increase the image quality. In this paper, a projection-based noise filter is presented that significantly enhances the SNR in projection reconstruction (PR) fluoroscopy without apparent loss of resolution in the reconstructed images. In contrast to an imaged-based approach, this method allows a very efficient computational implementation. The filter algorithm was implemented on a digital signal processor and was applied to real-time processing during PR fluoroscopy. A quantitative analysis of the improvement in SNR and results for different fluoroscopic MR applications are given. Apart from MR fluoroscopy, the proposed technique has the potential to be applied to low dose computed tomography fluoroscopy.