Binary classification of dead detector elements in flat panel detectors using convolutional neural networks.

Objective.Medical physicists routinely perform quality assurance on digital detection systems, part of which involves the testing of flat panel detectors. Flat panels may degrade over time as an increasing number of individual detector elements begin to malfunction. The pixels that correspond to these elements are corrected for using information elsewhere in the detector system, however these corrected elements still constitute a loss in image quality for the system as a whole. These correction methods, as well as the location and number of dead detector elements, are often only available to the vendor of the digital detection system, but not to the medical physicist responsible for the quality assurance of the system.Approach.We greatly expand upon a previous work by providing a novel technique for classifying dead detector elements at single pixel resolution. We also demonstrate that this technique can be trained on one detector, and then tested and validated on another with moderate success, which demonstrates some ability to generalize to different detectors. The technique requires 3 flat field, or 'noise', images to be taken to predict the dead detector element maps for the system.Main results.Models using only for-processing pixel data were unable to successfully generalize from one detector to the other. Models preprocessed using the standard deviation across three for-processing images were able to classify dead detector element maps with an F1score ranging from 0.4527 to 0.8107 and recall ranging from 0.5420 to 0.9303 with better performance, on average, observed using the low exposure data set.Significance. Many physicists do not have access to the dead detector maps for their diagnostic digital radiography systems. CNNs are capable of predicting the dead detector maps of flat panel detectors with single pixel resolution. Physicists can implement this tool by acquiring three flat field images and then inputting them into the model. Model performance saw a marginal increase when trained on the low exposure set data, as opposed to the high exposure set data, indicating high exposure, low relative noise images may not be necessary for optimal performance. Model performance across detectors manufactured by different vendors requires further investigation.

Dead detector element detection in flat panels using convolutional neural networks.

BACKGROUND: Independent testing of image quality metrics is important to provide an unbiased determination of medical imaging performance. Due to the underreporting by vendors of dead detector elements, which are elements that do not function but may be corrected using information from surrounding pixels, the health of all imaging detector elements is infrequently reported and extraordinarily difficult to independently determine or verify through traditional means without vendor largesse. In many instances, dead detector data, or dead pixel maps, are only available at the discretion of the vendors, which renders these data inaccessible to many medical physicists. PURPOSE: We provide a mechanism to predict which detector elements are dead from a single flat field image and describe a convolutional neural network (CNN) to address this task. This technique provides a vendor-independent and potential correction algorithm-independent means for obtaining this information, which directly relates to the image quality of the diagnostic imaging system. METHODS: A data set of 61 flat field images was obtained from two Varian on-board imaging (OBI) systems. These images were taken with a range of kVp and mAs settings and with differing levels of copper filtration in the beam path. The dead pixel maps for these detectors were also acquired and used to label the ground truth subimages-or small square images from the original, larger image-from the flat field images. This data set was then used to train, test, and validate five unique CNNs on the task of identifying dead detector elements from subimages derived from the flat field images. RESULTS: The CNNs were validated with an average precision of 0.96, an average recall of 0.48, and an average F1 score of 0.55 attributed to the scarcity and imbalanced nature of available data, and the difficulty in training neural networks using flat-field images. However, performance drastically improved when utilizing a subset of the data with a high dynamic range. CONCLUSION: This work shows the feasibility of using a CNN to detect dead detector elements from flat field images taken on a detector. Robust training of this type of algorithm could lead to a generalized model that may provide an independent evaluation of dead detectors for a wide range of vendors and models. Ultimately, this method can be a valuable tool for physicists performing quality assurance procedures on all digital diagnostic imaging systems, including X-ray, mammography, and fluoroscopy.

Orientation of the round window membrane: A normative study of inner ear anatomical orientation using 2D projections of 3D volumes.

INTRODUCTION: Orientation of the Round Window Membrane (RWM) is an important metric to establish if utilized as a potential access for targeted delivery of magnetically guided nanomedicines to the inner ear. Orientation with respect to an internal reference frame (such as the planes defined by the semicircular-canals [SCC]) may provide an internally consistent basis if the basis is orthogonal and consistent (from patient to patient). MATERIALS AND METHODS: Utilizing a micro computed tomography (CT), 20 temporal bones are scanned for anatomical information. The scanned data sets are loaded into an imaging program to provide volumetric reconstruction and segmentation. Volumetric models of the anatomical relationships between the inner ear SCC and the RWM are utilized to get normative projection angle information and are statistically analyzed. RESULTS: Micro-CT shows low to moderate reliability for reproducibility, intraobserver, and interobserver measurements; in addition, it provides mean values (±SD) for the various measured angles. The combined mean angular values for surface orientation of the RWM, with respect to the SCC basis (quasi-orthogonal spherical coordinate system), was 57.0° ± 20.9°as measured from the line defining the posterior SCC plane in the direction of the line defining the superior SCC plane. An angle of 65.2° ± 19.1° was measured for an angle away from the line defining the horizontal SCC plane.

Targeted Ganglionated Plexi Denervation Using Magnetic Nanoparticles Carrying Calcium Chloride Payload.

OBJECTIVES: This study sought to develop a novel targeted delivery therapy to ablate the major atrial ganglionated plexi (GP) using magnetic nanoparticles carrying a CaCl2 payload. BACKGROUND: Prior studies indicated the role of hyperactivity of the cardiac autonomic nervous system in the genesis of atrial fibrillation. METHODS: Twenty-eight male mongrel dogs underwent a bilateral thoracotomy. CaCl2-encapsulated magnetic nanoparticles (Ca-MNP) included magnetite in a sphere of biocompatible, biodegradable poly(lactic-co-glycolic acid). A custom external electromagnet focusing the magnetic field gradient (2,600 G) on the epicardial surface of the targeted GP was used to pull Ca-MNP into and release CaCl2 within the GP. The ventricular rate slowing response to high frequency stimulation (20 Hz, 0.1 ms) of the GP was used to assess the GP function. RESULTS: The minimal effective concentration of CaCl2 to inhibit the GP function was 0.5 mmol/l. Three weeks after CaCl2 (0.5 mmol/l, n = 18 GP) or saline (n = 18 GP) microinjection into GP, the increased GP function, neural activity, and atrial fibrillation inducibility, as well as shortened effective refractory period in response to 6 h of rapid atrial pacing (1,200 beats/min) were suppressed by CaCl2 microinjection. After intracoronary infusion of Ca-MNP, the external electromagnet pulled Ca-MNP to the targeted GP and suppressed the GP function (n = 6 GP) within 15 min. CONCLUSIONS: Ca-MNP can be magnetically targeted to suppress GP function by calcium-mediated neurotoxicity. This novel approach may be used to treat arrhythmias related to hyperactivity of the cardiac autonomic nervous system, such as early stage of atrial fibrillation, with minimal myocardial injury.

Comparison of computed tomography shielding design methods using RadShield.

PURPOSE: To comprehensively compare four computed tomography (CT) scanner shielding design methods using RadShield, a Java-based graphical user interface (GUI). METHODS: RadShield, a floor plan based GUI, was extended to calculate air kerma and barrier thickness using accepted methods from the National Council on Radiation Protection and Measurements (NCRP), the British Institute of Radiology, and a method using isodose maps, for spatially distributed points beyond user defined barriers. For a stationary CT scanner, the overall shielding recommendations found using RadShield were also compared to those found by American Board of Radiology certified diagnostic medical physicists using the conventional NCRP dose length product method and the isodose map method. RESULTS: The results between methods differed significantly for calculation point locations beyond the gantry and to the rear of the gantry. Overall shielding design recommendations across the four methods yielded similar average air kerma and thickness values for the barriers. CONCLUSIONS: RadShield was extended to perform CT shielding design and proved reliable using four methods.

RadShield: semiautomated shielding design using a floor plan driven graphical user interface.

The purpose of this study was to introduce and describe the development of RadShield, a Java-based graphical user interface (GUI), which provides a base design that uniquely performs thorough, spatially distributed calculations at many points and reports the maximum air-kerma rate and barrier thickness for each barrier pursuant to NCRP Report 147 methodology. Semiautomated shielding design calculations are validated by two approaches: a geometry-based approach and a manual approach. A series of geometry-based equations were derived giv-ing the maximum air-kerma rate magnitude and location through a first derivative root finding approach. The second approach consisted of comparing RadShield results with those found by manual shielding design by an American Board of Radiology (ABR)-certified medical physicist for two clinical room situations: two adjacent catheterization labs, and a radiographic and fluoroscopic (R&F) exam room. RadShield's efficacy in finding the maximum air-kerma rate was compared against the geometry-based approach and the overall shielding recommendations by RadShield were compared against the medical physicist's shielding results. Percentage errors between the geometry-based approach and RadShield's approach in finding the magnitude and location of the maximum air-kerma rate was within 0.00124% and 14 mm. RadShield's barrier thickness calculations were found to be within 0.156 mm lead (Pb) and 0.150 mm lead (Pb) for the adjacent catheteriza-tion labs and R&F room examples, respectively. However, within the R&F room example, differences in locating the most sensitive calculation point on the floor plan for one of the barriers was not considered in the medical physicist's calculation and was revealed by the RadShield calculations. RadShield is shown to accurately find the maximum values of air-kerma rate and barrier thickness using NCRP Report 147 methodology. Visual inspection alone of the 2D X-ray exam distribution by a medical physicist may not be sufficient to accurately select the point of maximum air-kerma rate or barrier thickness.

A Two-Magnet System to Push Therapeutic Nanoparticles.

Magnetic fields can be used to direct magnetically susceptible nanoparticles to disease locations: to infections, blood clots, or tumors. Any single magnet always attracts (pulls) ferro- or para-magnetic particles towards it. External magnets have been used to pull therapeutics into tumors near the skin in animals and human clinical trials. Implanting magnetic materials into patients (a feasible approach in some cases) has been envisioned as a means of reaching deeper targets. Yet there are a number of clinical needs, ranging from treatments of the inner ear, to antibiotic-resistant skin infections and cardiac arrhythmias, which would benefit from an ability to magnetically "inject", or push in, nanomedicines. We develop, analyze, and experimentally demonstrate a novel, simple, and effective arrangement of just two permanent magnets that can magnetically push particles. Such a system might treat diseases of the inner ear; diseases which intravenously injected or orally administered treatments cannot reach due to the blood-brain barrier.