Quantitative analysis of an enlarged area Solid State X-ray Image Intensifier (SSXII) detector based on Electron Multiplying Charge Coupled Device (EMCCD) technology.

Present day treatment for neurovascular pathological conditions involves the use of devices with very small features such as stents, coils, and balloons; hence, these interventional procedures demand high resolution x-ray imaging under fluoroscopic conditions to provide the capability to guide the deployment of these fine endovascular devices. To address this issue, a high resolution x-ray detector based on EMCCD technology is being developed. The EMCCD field-of-view is enlarged using a fiber-optic taper so that the detector features an effective pixel size of 37 µm giving it a Nyquist frequency of 13.5 lp/mm, which is significantly higher than that of the state of the art Flat Panel Detectors (FPD). Quantitative analysis of the detector, including gain calibration, instrumentation noise equivalent exposure (INEE) and modulation transfer function (MTF) determination, are presented in this work. The gain of the detector is a function of the detector temperature; with the detector cooled to 5° C, the highest relative gain that could be achieved was calculated to be 116 times. At this gain setting, the lowest INEE was measured to be 0.6 µR/frame. The MTF, measured using the edge method, was over 2% up to 7 cycles/ mm. To evaluate the performance of the detector under clinical conditions, an aneurysm model was placed over an anthropomorphic head phantom and a coil was guided into the aneurysm under fluoroscopic guidance using the detector. Image sequences from the procedure are presented demonstrating the high resolution of this SSXII.

Image acquisition, geometric correction and display of images from a 2×2 x-ray detector array based on Electron Multiplying Charge Coupled Device (EMCCD) technology.

A high resolution (up to 11.2 lp/mm) x-ray detector with larger field of view (8.5 cm × 8.5 cm) has been developed. The detector is a 2 × 2 array of individual imaging modules based on EMCCD technology. Each module outputs a frame of size 1088 × 1037 pixels, each 12 bits. The frames from the 4 modules are acquired into the processing computer using one of two techniques. The first uses 2 CameraLink communication channels with each carrying information from two modules, the second uses a application specific custom integrated circuits, the Multiple Module Multiplexer Integrated Circuit (MMMIC), 3 of which are used to multiplex the data from 4 modules into one CameraLink channel. Once the data is acquired using either of the above mentioned techniques, it is decoded in the graphics processing unit (GPU) to form one single frame of size 2176 × 2074 pixels each 16 bits. Each imaging module uses a fiber optic taper coupled to the EMCCD sensor. To correct for mechanical misalignment between the sensors and the fiber optic tapers and produce a single seamless image, the images in each module may be rotated and translated slightly in the x-y plane with respect to each other. To evaluate the detector acquisition and correction techniques, an aneurysm model was placed over an anthropomorphic head phantom and a coil was guided into the aneurysm under fluoroscopic guidance using the detector array. Image sequences before and after correction are presented which show near-seamless boundary matching and are well suited for fluoroscopic imaging.

Dose Reduction in Fluoroscopic Interventions Using a Combination of a Region of Interest (ROI) X-Ray Attenuator and Spatially-Different, Temporally-Variable Temporal Filtering.

A novel dose reduction technique for fluoroscopic interventions involving a combination of a material x-ray region of interest (ROI) attenuator and spatially different, temporally variable ROI temporal recursive filter, was used to guide the catheter to the ROI in three live animal studies, two involving rabbits and one involving a sheep. In the two rabbit studies presented, a catheter was guided to the entrance of the carotid artery. With the added ROI attenuator the image under the high attenuation region is very noisy. By using temporal filtering with a filter weight of 0.6 on previous frames, the noise is reduced. In the sheep study the catheter was guided to the descending aorta of the animal. The sheep offered a relatively higher attenuation to the incident x-rays and thus a higher temporal filter weight of 0.8 on previous frames was used during the procedure to reduce the noise to levels acceptable by the interventionalist. The image sequences from both studies show that significant dose reduction of 5-6 times can be achieved with acceptable image quality outside the ROI by using the above mentioned technique. Even though the temporal filter weighting outside the ROI is higher, the consequent lag does not prevent perception of catheter movement.

Two dimensional extensible array configuration for EMCCD-based solid state x-ray detectors.

We have designed and developed from the discrete component level a high resolution dynamic x- ray detector to be used for fluoroscopic and angiographic medical imaging. The heart of the detector is a 1024 × 1024 pixel electron multiplying charge coupled device (EMCCD) with a pixel size of 13 × 13 µm(2) (Model CCD201-20, e2v Technologies, Inc.), bonded to a fiber optic plate (FOP), and optically coupled to a 350 µm thick micro-columnar CsI(TI) scintillator via a fiber optic taper (FOT). Our aim is to design an array of these detectors that could be extended to any arbitrary X × Y size in two dimensions to provide a larger field of view (FOV). A physical configuration for a 3×3 array is presented that includes two major sub-systems. First is an optical front end that includes (i) a phosphor to convert the x-ray photons into light photons, and (ii) a fused array of FOTs that focuses light photons from the phosphor onto an array of EMCCD's optically coupled using FOPs. Second is an electronic front end that includes (i) an FPGA board used for generating clocks and for data acquisition (ii) driver boards to drive and digitize the analog output from the EMCCDs, (iii) a power board, and (iv) headboards to hold the EMCCD's while they are connected to their respective driver board using flex cables. This configuration provides a larger FOV as well as region-of- interest (ROI) high-resolution imaging as required by modern neurovascular procedures.

Dose Reduction Technique Using a Combination of a Region of Interest (ROI) Material X-Ray Attenuator and Spatially Different Temporal Filtering for Fluoroscopic Interventions.

We demonstrate a novel approach for achieving patient dose savings during image-guided neurovascular interventions, involving a combination of a material x-ray region of interest (ROI) attenuator and a spatially different ROI temporal filtering technique. The part of the image under the attenuator is reduced in dose but noisy and less bright due to fewer x-ray quanta reaching the detector, as compared to the non-attenuating (or less attenuating) region. First the brightness is equalized throughout the image by post processing and then a temporal filter with higher weights is applied to the high attenuating region to reduce the noise, at the cost of increased lag; however, in the regions where less attenuation is present, a lower temporal weight is needed and is applied to preserve temporal resolution. A simulation of the technique is first presented on an actual image sequence obtained from an endovascular image guided interventional (EIGI) procedure. Then the actual implementation of the technique with a physical ROI attenuator is presented. Quantitative analysis including noise analysis and integral dose calculations are presented to validate the proposed technique.

Graphics Processing Unit (GPU) implementation of image processing algorithms to improve system performance of the Control, Acquisition, Processing, and Image Display System (CAPIDS) of the Micro-Angiographic Fluoroscope (MAF).

We present the image processing upgrades implemented on a Graphics Processing Unit (GPU) in the Control, Acquisition, Processing, and Image Display System (CAPIDS) for the custom Micro-Angiographic Fluoroscope (MAF) detector. Most of the image processing currently implemented in the CAPIDS system is pixel independent; that is, the operation on each pixel is the same and the operation on one does not depend upon the result from the operation on the other, allowing the entire image to be processed in parallel. GPU hardware was developed for this kind of massive parallel processing implementation. Thus for an algorithm which has a high amount of parallelism, a GPU implementation is much faster than a CPU implementation. The image processing algorithm upgrades implemented on the CAPIDS system include flat field correction, temporal filtering, image subtraction, roadmap mask generation and display window and leveling. A comparison between the previous and the upgraded version of CAPIDS has been presented, to demonstrate how the improvement is achieved. By performing the image processing on a GPU, significant improvements (with respect to timing or frame rate) have been achieved, including stable operation of the system at 30 fps during a fluoroscopy run, a DSA run, a roadmap procedure and automatic image windowing and leveling during each frame.

Implementation of digital multiplexing for high resolution X-ray detector arrays.

We describe and demonstrate for the first time the use of the novel Multiple Module Multiplexer (MMMIC) for a 2×2 array of new electron multiplying charge coupled device (EMCCD) based x-ray detectors. It is highly desirable for x-ray imaging systems to have larger fields of view (FOV) extensible in two directions yet to still be capable of doing high resolution imaging over regions-of-interest (ROI). The MMMIC achieves these goals by acquiring and multiplexing data from an array of imaging modules thereby enabling a larger FOV, and at the same time allowing high resolution ROI imaging through selection of a subset of modules in the array. MMMIC also supports different binning modes. This paper describes how a specific two stage configuration connecting three identical MMMICs is used to acquire and multiplex data from a 2×2 array of EMCCD based detectors. The first stage contains two MMMICs wherein each MMMIC is getting data from two EMCCD detectors. The multiplexed data from these MMMICs is then forwarded to the second stage MMMIC in the similar fashion. The second stage that has only one MMMIC gives the final 12 bit multiplexed data from four modules. This data is then sent over a high speed Camera Link interface to the image processing computer. X-ray images taken through the 2×2 array of EMCCD based detectors using this two stage configuration of MMMICs are shown successfully demonstrating the concept.

A 2×2 array of EMCCD-based solid state x-ray detectors.

We have designed and developed a new solid-state x-ray imaging system that consists of a 2×2 array of electron multiplying charge coupled devices (EMCCDs). This system is intended for fluoroscopic and angiographic medical imaging. The key components are the four 1024 × 1024 pixel EMCCDs with a pixel size of 13 × 13 µm(2). Each EMCCD is bonded to a fiber optic plate (FOP), and optically coupled to a 350 µm thick micro-columnar CsI(TI) scintillator via a 3.22∶1 fiber optic taper (FOT). The detector provides x-ray images of 9 line pairs/mm resolution at 15 frames/sec and real-time live video at 30 frames/sec with binning at a lower resolution, independent of the electronic gain applied to the EMCCD. The total field of view (FOV) of the array is 8.45 cm × 8.45 cm. The system is designed to also provide the ability to do region-of- interest imaging (ROI) by selectively enabling individual modules of the array.

Spatially different, real-time temporal filtering and dose reduction for dynamic image guidance during neurovascular interventions.

Fluoroscopic systems have excellent temporal resolution, but are relatively noisy. In this paper we present a recursive temporal filter with different weights (lag) for different user selected regions of interest (ROI) to assist the neurointerventionalist during an image guided catheter procedure. The filter has been implemented on a Graphics Processor (GPU), enabling its usage for fast frame rates such as during fluoroscopy. We first demonstrate the use of this GPU-implemented rapid temporal filtering technique during an endovascular image guided intervention with normal fluoroscopy. Next we demonstrate its use in combination with ROI fluoroscopy where the exposure is substantially reduced in the peripheral region outside the ROI, which is then software-matched in brightness and filtered using the differential temporal filter. This enables patient dose savings along with improved image quality.

EMCCD-based high resolution dynamic x-ray detector for neurovascular interventions.

We have designed and developed from the discrete component level a high resolution dynamic detector for neurovascular interventions. The heart of the detector is a 1024 × 1024 pixel electron multiplying charge coupled device (EMCCD) with a pixel size of 13 × 13 µm(2), bonded to a fiber optic plate (FOP), and optically coupled to a 350 µm micro-columnar CsI(TI) scintillator via a 3.3:1 fiber optic taper (FOT). The detector provides x-ray images of 9 cycles/mm resolution at 15 frames/sec and real time live video at 30 frames/sec with binning at a lower resolution, both independent of gain applied to EMCCD, as needed for region-of-interest (ROI) image guidance during neurovascular interventions.

Novel Multiplexer to Enable Multiple-Module Imaging with Adjustable High Spatial Resolution and Predetermined Display Bandwidth for Array Medical Imaging Systems.

We describe a custom multiple-module multiplexer integrated circuit (MMMIC) that enables the combination of discrete Electron multiplying charge coupled devices (EMCCD) based imaging modules to improve medical imaging systems. It is highly desirable to have flexible imaging systems that provide high spatial resolution over a specific region of interest (ROI) and a field of view (FOV) large enough to encompass areas of clinical interest. Also, such systems should be dynamic, i.e. should be able to maintain a specified acquisition bandwidth irrespective of the size of the imaged FOV. The MMMIC achieves these goals by 1) multiplexing the outputs of an array of imaging modules to enable a larger FOV, 2) enabling a number of binning modes for adjustable high spatial resolution, and 3) enabling selection of a subset of modules in the array to achieve ROI imaging at a predetermined display bandwidth. The MMMIC design also allows multiple MMMICs to be connected to control larger arrays. The prototype MMMIC was designed and fabricated in the ON-SEMI 0.5µm CMOS process through MOSIS (www.mosis.org). It has three 12-bit inputs, a single 12-bit output, three input enable bits, and one output enable, so that one MMMIC can control the output from three discrete imager arrays. The modular design of the MMMIC enables four identical chips, connected in a two-stage sequential arrangement, to readout a 3×3 collection of individual imaging modules. The first stage comprises three MMMICs (each connected to three of the individual imaging module), and the second stage is a single MMMIC whose 12-bit output is then sent via a CameraLink interface to the system computer. The prototype MMMIC was successfully tested using digital outputs from two EMCCD-based detectors to be used in an x-ray imaging array detector system.Finally, we show how the MMMIC can be used to extend an imaging system to include any arbitrary (M×N) array of imaging modules enabling a large FOV along with ROI imaging and adjustable high spatial resolution.

A new wide range Euclidean distance circuit for neural network hardware implementations.

In this paper, we describe an analog very large-scale integration (VLSI) implementation of a wide range Euclidean distance computation circuit - the key element of many synapse circuits. This circuit is essentially a wide-range absolute value circuit that is designed to be as small as possible (80 /spl times/ 76 /spl mu/m) in order to achieve maximum synapse density while maintaining a wide range of operation (0.5 to 4.5 V) and low power consumption (less than 200 /spl mu/W). The circuit has been fabricated in 1.5-/spl mu/m technology through MOSIS. We present simulated and experimental results of the circuit, and compare these results. Ultimately, this circuit is intended for use as part of a high-density hardware implementation of a self-organizing map (SOM). We describe how this circuit can be used as part of the SOM and how the SOM is going to be used as part of a larger bio-inspired vision system based on the octopus visual system.