Applied Radiology

This is the second issue in a series of newsletters that provides input from experts on the opportunities they have created by overcoming adversity in establishing successful PACS entities.

We anticipate that the careful analysis of alternatives and the focus on solutions demonstrated by these experts will be useful to everyone exploring PACS and related technologies. ­­Ronald B. Schilling, PhD

Computed and Digital Radiography

Katherine P. Andriole, PhD

Computed and digital radiography (CR and DR) are modalities that acquire projection radiographs digitally using detectors other than film. Computed radiography refers to projection X-ray imaging with photostimulable or storage phosphors. In this modality, X rays incident on a photostimulable phosphor-based image sensor produce a latent image that is stored in the imaging plate until stimulated to luminesce by laser light. This released light energy can be captured and converted to a digital electronic signal for transmission of images to display and archival devices.

A CR system consists of a screen or plate of a photostimulable phosphor material that is usually contained in a cassette and is exposed in a manner similar to that used for traditional screen-film cassettes. The photostimulable phosphor in the imaging plate absorbs X rays that have passed through the patient, thus "recording" the X-ray image. Like the conventional intensifying screen, CR plates produce light in response to X rays at the time of exposure. Storage phosphor plates, however, have the additional property of being able to store some of the absorbed X-ray energy as a latent image. Irradiation of the imaging plate at some time after the X-ray exposure with red or near-infrared laser light stimulates the phosphor to release some of its stored energy in the form of green, blue, or ultraviolet light. This is the phenomenon of photostimulable luminescence. The intensity of light emitted is linearly proportional to the amount of X rays absorbed by the storage phosphor. ¹

The readout process uses a precision laser spot scanning mechanism in which the laser beam traverses the imaging plate surface in a raster pattern. The stimulated light emitted from the imaging plate is collected and converted into an electrical signal, with optics coupled to a photomultiplier tube. The electrical signal is then amplified, sampled to produce discrete pixels of the digital image, and sent through an analog-to-digital converter (ADC) to quantize the value of each pixel (i.e., a value between 0 and 1023 for a 10-bit ADC or between 0 and 4095 for a 12-bit ADC). Not all of the stored energy in the imaging plate is released during the readout process. Thus, to prepare the imaging plate for reuse, it is briefly flooded with high-intensity (typically fluorescent) light. This erasure step ensures removal of any residual latent image.

Digital radiography refers to devices for direct digital acquisition of projection radiographs in which the digitization of the X-ray signal takes place within the detector itself, providing an immediate full-fidelity image on a soft-copy display monitor. There are two types of DR devices (also called flat-panel detectors): indirect conversion devices, in which light is first generated by using a scintillator or phosphor and then detected by a charge-coupled device or a thin-film-transistor (TFT) array in conjunction with photodiodes; and direct DR devices, which consist of a top electrode, dielectric layer, selenium X-ray photoconductor, and thin-film pixel array. ¹ Direct DR devices offer direct energy conversion of X rays for immediate readout without the intermediate light conversion step and loss of signal.

The basis of DR devices is the large-area TFT active matrix array, or flat panel, in which each pixel consists of a signal collection area or charge collection electrode, a storage capacitor, and an amorphous silicon field-effect transistor switch that allows the active readout of the charge stored in the capacitor. ¹ Arrays of individual detector areas are addressed by orthogonally arranged gate switches and data lines to read the signal generated by the absorption of X rays in the detector. The TFT arrays are used in conjunction with a direct X-ray photoconductor layer or with an indirect X-ray-sensitive, phosphor-coated, light-sensitive detector/photodiode array.

Unlike conventional screen-film radiography, in which film functions as the imaging sensor or recording medium, as well as the display and storage medium, CR and DR eliminate film from the image-recording step, resulting in the separation of image capture from image display and image storage. This separation of functions allows for optimization of each of these steps individually. In addition, both CR and DR can capitalize on features common to digital or filmless imaging; namely the ability to acquire, transmit, display, manipulate, and archive data electronically, potentially overcoming some of the limitations of conventional screen-film radiography. Digital imaging benefits include remote access to images and clinical information by multiple users simultaneously, permanent storage and subsequent retrieval of image data, expedient information delivery to those who need it, and efficient, cost-effective workflow with elimination of film from the equation.

Imaging modality conformance with the Digital Imaging and Communications in Medicine (DICOM) standard is critical. Equally essential is interfacing the Radiology Information System (RIS)­ Hospital Information System (HIS) with the PACS. This greatly facilitates input of patient demographics (name, date, time, medical record number, accession number, exam type), imaging parameters, etc., and enables automatic PACS data verification, correlation, and error correction with the data recorded in the RIS-HIS.

Both CR and DR have extremely wide latitude response detectors, potentially reducing retakes resulting from poor exposure in the screen­film-based environment. The spatial resolution capabilities of CR and DR are comparable to each other but are less than film. However, in many cases, the superior contrast resolution of the digital modalities can compensate for the lesser spatial resolution relative to screen-film radiography. ¹ DR has a higher efficiency detector and immediate image readout. Some current DR problem areas include ease-of-use and portability issues, the requirement for a single device per radiographic room ("one-room-at-a-time" technology), and its high production cost. CR has superior procedural flexibility, can accommodate portable bedside examinations, and is a proven clinical modality. Several timing and productivity studies have concluded that both CR and DR can improve workflow and productivity over analog screen-film, but that DR requires high volume to be more cost effective than CR. ²

Future advances in CR technologies are aimed at improving signal collection and acquisition speed to make CR competitive with DR. Introduced 2 years ago, but not yet commercially available, the "Scan Head" technology (Agfa Healthcare) uses a line scanning and laser stimulation mechanism as opposed to the point-by-point scanning used in current CRs. It is anticipated that this new design will improve CR performance in a number of ways. Scan time will be reduced; tests show a 75% to 80% reduction in scanning time. Image collection is improved and it is anticipated that the product will have a much smaller footprint than today's scanners (desktop size) and be sold at a reduced price. A second innovation on the horizon is the use of a needle versus powder phosphor in the CR detector. Tests show that this advance improves image quality by increasing signal collection and spatial resolution.

Dual-sided reading is another improvement aimed at increasing signal through stimulation and collection from both sides of the imaging plate. In addition, a cassetteless scanner design has been introduced to rival DR workflow and speed. CR detectors have been incorporated into chest, table, and wall buckys with positive results.

While CR research and development continues to produce improvements in the modality, DR advancements are on going as well. Future advances for DR address uses for other examinations in addition to chest, portability issues, and dynamic studies. Initially, DR units were only available for upright chest applications. Now DR detectors have been incorporated into the table bucky. Portable devices are starting to appear for military applications and may become practical for civilian uses. Image processing for both CR and DR continues to evolve. The cost of DR systems is still high for low-volume imaging sites.

Recent technologic advances in CR and DR have begun to make digital projection radiography more prevalent in the clinical arena, with CR currently having the greater clinical installed base. Hardware and software improvements in detector devices, image-reading scanning devices, image-processing algorithms, and the cost and utility of image display devices have contributed to the increased acceptance of these digital counterparts to conventional screen-film radiography. It is envisioned that these devices will continue to get physically smaller, operationally faster, and less expensive, with higher quality output through maturation of image processing algorithms and value-added image analysis and display applications. It is likely that CR and DR will continue to coexist as digital radiographic devices for some time. ¹ n

REFERENCES

1. Andriole KP. Computed and digital radiography. In: Hangiandreou NJ, Young JWR, Morin RL, eds. Electronic Radiology Practice . Chicago: RSNA Publishers; 1999:35-41.

2. Andriole KP, Luth DM, Gould RG. Workflow assessment of DR versus CR and screen-film in the outpatient environment. J Digital Imaging . 2002;15:124-126.

Practical Issues

Digital Detector Systems Technology Part 2: Comparisons

J. Anthony Seibert, PhD

In the first issue of this newsletter, Part 1 of this two-part article addressed limitations of screen-film radiography, and how a "digital" X-ray detector system could potentially overcome these limitations and serve as an input to a picture archiving and communications system (PACS). This article compares the various digital radiography detector systems currently available for clinical implementation and their generic attributes. A basic understanding of the relative merits of these digital detectors is important so that the purchaser/implementer can make an informed decision on what type of digital detector optimally meets the clinical need in terms of cost, throughput, practicality, flexibility, and ease of use, among many attributes.

DIGITAL DETECTOR TYPES

Image Intensifier/TV systems­­ The earliest digital X-ray detectors (circa 1975) were based upon an II/TV system, which adapted the output video signal to an analog to digital converter (ADC). State-of-the-art image intensifiers have a high gain, low noise, and moderate spatial resolution capability, largely due to the structured cesium iodide (CsI) input phosphor material; when coupled with a light converter system (e.g., analog TV or CCD camera), the system produces a low-noise, high-quality signal. Limitations arise from the geometric distortion of the II, large bulky size of the tube and housing, and limited dynamic range due to the saturation characteristics of the TV camera. A light-limiting aperture (iris) in the optical coupling provides the capability to adjust the incident exposure by adjusting the amount of light that is output by the II. Primary uses of these devices are for fluoroscopy, fluorography, and angiography dynamic sequences. Image matrix sizes contain as many as 2000 ¥ 2000 pixels.

Computed radiography (CR)­­ CR is the generic term for the digital imaging detector system using a photostimulable phosphor and a mechanical-optical reader system. This technology was first introduced in the early 1980s, but high system costs, large size, and image quality issues blunted its widespread clinical appearance until the early 1990s, with the introduction of more reliable systems having a smaller footprint and significantly lower costs. In function, CR emulates the screen-film paradigm very closely. The storage phosphor imaging plate (IP) is housed in a cassette that resembles a screen-film cassette with multiple sizes. Once exposed to X-rays, electrons in the phosphor material are elevated to higher energy traps called F-centers in numbers that are proportional to the number of X-rays incident, forming a "latent-image." Subsequently, the exposed imaging plate is electronically "processed" by a scanning laser beam of red light (a HeNe or diode laser with a spot size of ~100 µm), which induces the electrons to absorb energy great enough to exceed the threshold energy of the trap, and release higher energy light photons. A light guide positioned close to the surface of the phosphor collects the photostimulated luminescence photons, converts the light energy to an electronic signal with the use of a sensitive, high gain photomultiplier tube (PMT), and produces a digital signal. Reflecting the laser beam off a rotating polygonal mirror or oscillating galvanometer mirror with electronic timing and sampling of the PMT output signal captures the information along the "scan" direction of the IP. Simultaneous mechanical translation of the IP (the "sub-scan" direction) in the optical stage allows a full scan, with a readout time of typically 45 to 135 seconds, depending on the size of the detector and the throughput of the reader. Once fully acquired, the image data is processed in three steps: image data ranging/scaling, contrast enhancement, and frequency enhancement. Anatomy-specific algorithms tune the output image to radiologists viewing preference, and further image manipulation, when necessary, is easily applied at a computer workstation.

CR is a mature technology, and is the most widespread and flexible system in use for the acquisition of projection radiographs in a PACS environment, but is also the most labor-intensive in terms of after-acquisition handling. Image matrix sizes depend on the detector size, with a 100 to 200 µm pixel size most often employed, producing image sizes of 8 to 10 MB. A digital mammography prototype system with 50 µm sampling pitch produces uncompressed image sizes of 50 MB per image. One CR reader can service multiple rooms; however, extra handling and time is required by the technologist to "process" the images can reduce the efficiency of patient throughput, particularly if films are still being printed.

CCD/CMOS Cameras­­ Digital camera-based detectors represent a very cost-effective alternative to other large-field detector systems; however, the small size of the active camera area requires a significant demagnification of the image. In this geometry, the light emission from the scintillator occurs in all directions (a Lambertian emitter) and the lens only captures a small fraction. Optical lens efficiency is poor, and image statistics become dominated by the number of light photons detected by the camera rather than the number of X-ray photons absorbed in the scintillator. This creates a "secondary quantum sink" where a significant noise penalty impacts the final rendered image. (Note: the II/TV system mentioned earlier does not suffer this loss, as the light image is highly amplified by electronic minification and acceleration gain producing non-limiting brightness.) For incident exposures typical of a 400-speed screen-film detector, the CCD system DQE can be low and not as dose efficient as screen-film detectors or other large-field detectors. With higher input exposures, the DQE increases as light amplification increases. It is important to realize that the DQE of the CCD camera to light photons is often quoted in sales brochures, rather than the more important system DQE for a given incident X-ray exposure. The use of higher kVp is often recommended to help lower patient dose (less attenuation, more transmission through the body) and to increase the intensification factor (more light photons per absorbed X-ray). Despite drawbacks, optically coupled detector systems can be a very cost-effective alternative to CR and flat-panel detectors, and should be considered with the proviso that the radiation dose for a given image quality will be higher. While not as dose efficient, good quality images are obtainable. In certain designs, the digital cameras can be tiled, producing a large-area light-sensitive array that does not suffer the same losses that directly coupled systems do (e.g., tiled CMOS large area detector systems).

TFT Flat-panel Detectors­­ Thin-film-transistor (TFT) arrays for X-ray detector applications are a result of the intense research and production of flat-panel displays for laptop computers. These devices are hewn from silicon semiconductor sheets, upon which discrete pixel areas are formed. There are two distinct categories of flat panel arrays: the indirect and the direct X-ray converters. Both types of detectors are comprised of a charge collection device (the storage capacitor), and electronic switches (the TFT), to electronically "read" the charge pattern that corresponds to the X-ray interaction events after the exposure. Where they differ is the method by which the charge is generated in the indirect device, a scintillator (e.g., structured CsI phosphor) layered on the TFT pixel matrix converts the incident X-rays to proportional light intensities, which in turn are captured by a "photodiode" device on each pixel that converts the light to a proportional number of electrons that are stored in the pixel capacitor. In the direct device, a photoconductive material, such as amorphous selenium (a-Se), is layered between two electrodes and placed under a relatively high voltage. As X-rays interact in the material, electron/hole pairs are generated in proportional numbers, and the charges are stored directly on the pixel storage capacitor. Reading the stored charges in the TFT array is accomplished by turning on the switches (the transistor gates) typically one row at a time, which allows the charges in each pixel (the storage capacitor) to be collected along the columns to an individual charge amplifier where the output signal is digitized and located in the image array. This is repeated for each row until the pixel matrix is fully discharged and the corresponding image stored in digital form. In some systems, multiple groups of charge amplifiers allow readout in parallel from segmented parts of the array to speed up the charge collection process.

A question often arises as to which method (indirect/direct) of readout is superior. There is no clear-cut answer, as both methods are effective in producing images of extremely high quality with high-detection efficiency. In general, the indirect detector (e.g., CsI-TFT array) has a faster readout time, and is more readily adaptable to dynamic fluoroscopic image sequences, while the a-Se photoconductor has a much higher intrinsic spatial resolution. The indirect detector can suffer from inefficient "fill-factor" effects, whereby the electronic components (e.g., the TFT, storage capacitor, and readout lines) represent dead-space that reduce the overall X-ray detection efficiency. This penalty is more severe for smaller pixel areas, and therefore has set a limit on the "reasonable" resolution that can be achieved, which is likely not to be much less than 70 to 80 µm. Direct acquisition devices are prone to frequency aliasing, a situation resulting from the extremely high intrinsic spatial resolution of the photoconductor that causes the higher spatial resolution signals that exist beyond the "Nyquist frequency" to fold back into the lower frequency spectrum, causing additive noise to be superimposed on the image content. Charge trapping and charge recombination are also hurdles in direct acquisition detectors that are less than ideal.

A significant research effort is underway to improve all flat-panel devices and to implement others, with the overall goal to improve detection efficiency, provide necessary spatial resolution, and achieve excellent image quality at a cost that can be justified in the clinical setting. There is definitely some ground to cover before these goals are realized.

COMPARISON OF DIGITAL DETECTORS

Most digital X-ray detectors have the ability to separate the three major attributes of an X-ray detector-namely acquisition, display, and archiving of the image data. As a result, many similar characteristics are common to digital X-ray detectors. This includes wide exposure dynamic range and compensation for variations in incident exposure, which is not necessarily a good thing, as complacency or misuse can lead to instances of under- or overexposure of the patient and potentially poor image quality. Certainly, image processing and contrast/spatial frequency enhancement are critical for the high-quality output of any digital imaging system. There are many strategies that researchers and vendors are currently investigating to seek out the most cost-effective and efficient digital detectors for diagnostic radiology.

Table 1 is based on "generalities" of system capabilities and performance. Certainly there are wide ranges of system capabilities within each of the major types of digital detector systems. As an example, some CR systems have built-in phosphor readers that do not require the technologist to have direct hands-on involvement with the processing step. Be aware that each system must be evaluated on its own merit, capabilities, and unique attributes.

ACCEPTANCE TESTING AND QUALITY CONTROL

Any imaging system is only as good as its weakest link. Maintaining optimal quality requires a robust quality-control program, which starts with the acceptance testing of these systems to determine and verify the "system" capabilities and to benchmark the performance, in terms of spatial resolution, signal to noise ratio, image uniformity, and other system-specific tests suggested by the manufacturer. Digital detectors provide an excellent opportunity to use the computer system to quantitatively measure detector performance, and objectively indicate with a "yes" or "no" whether to use the system for patient imaging. Unfortunately, there still is a lack of tools and testing phantoms that are available from the manufacturers; however, it is important to express the need and demand such phantoms/capabilities when writing specifications and/or negotiating the final purchase.

THE FINAL WORD

There are several key system attributes of digital radiography detectors that must be considered when choosing a particular device for an application. Certainly, the detector operational characteristics, capabilities, throughput, ease of use, cost, longevity, PACS integration, radiation dose, maintenance requirements, and service support are all necessary for a logical and justifiable decision for purchase and clinical implementation. A range of digital systems are well suited to a variety of needs and budgets. In addition to matching a given digital system to the clinical requirements driving the purchase, be aware that an overall system is only as good as the weakest link in the imaging chain. These weak links often include issues such as PACS interfacing, modality worklist functionality, and image display quality, these are not covered in this brief article. Finally, any digital radiographic imaging system that operates as a "stand-alone film-printing machine" that simply replaces an analog film with a digital laser-generated film is too costly to justify. In this situation, it would be better to improve the current analog system until such time that a filmless, all-digital acquisition, display, and archiving capability can be fully implemented. n

Economic Issues

Integration in Radiology

Janice C. Honeyman-Buck, PhD

In the beginning, there was no integration. Until radiology went digital, the concept of integration as we know it today, was not relevant. Then, in the 1970s as computed tomography and digital nuclear medicine were introduced, computer systems were developed to handle these new digital images and data; but each system acted independently and required a unique computer to acquire, analyze, manipulate, store, and display the output of the systems. Magnetic resonance imaging was developed and some institutions realized they were viewing each type of image on a different computer, usually the console of the acquisition modality or an independent console. Each manufacturer had a proprietary image and storage format and custom software for displaying the images. Sometimes different modalities from the same manufacturer had different image formats and required separate display workstations.

In the 1980s it became apparent that users wanted to use one workstation to view and manipulate images from different modalities and different vendors. In 1983 the American College of Radiology and the National Electrical Manufacturer's Association (ACR-NEMA) created a committee to investigate the creation of a standard image and communication definition so images from different manufacturers could be stored, viewed, and manipulated on computers that met the same standard. In 1985, the first standard, ACR-NEMA, version 1 was introduced. This version of the standard included only a hardware interface with a set of commands that enabled point-to-point communication of images and a data dictionary to describe the content of a message containing an image. This early interface was awkward to implement and use, however, it was a very large step forward. In 1988, version 2 was released, which allowed network communications, instead of the hardware point-to-point, and defined the communication standards along with the data definition. In 1992, the term DICOM (Digital Image Communications in Medicine) was coined and the standard as we know it today was introduced as DICOM 3. This evolving standard is an object-oriented model that includes support for the International Organization for Standardization (ISO) communication standards.

A parallel effort to develop standards for the interchange of clinical, financial, and administrative data among independent healthcare oriented computer systems was underway. Health Layer 7 (HL7) was a group founded in 1987 to work on the standard that is now used by virtually all Hospital Information and Radiology Information Systems (HIS/RIS) and most smaller computer driven systems (clinical laboratory, pharmacy, etc).

Radiology's early demands for integration were modest. We wanted to view images from all our modalities on a single workstation. Then our expectations grew. Now we want to integrate the workflow for the entire electronic radiology practice so there is a seamless transfer of data among systems from the time a study is ordered until its reported; then we want the reports and relevant images to be displayed in a meaningful and correct manner to anyone with a legitimate need to use them anywhere in the world, while maintaining patient privacy by keeping data safe from non-authorized use. We want the systems to reduce or eliminate errors and to improve productivity while maintaining the correct information to assure maximum reimbursement. ^1,2 All this in a cost effective manner. Is that too much to ask? Given the history of successes with DICOM and HL7, the answer is no.

CURRENT STATE OF INTEGRATION

Although some of the required integration has been accomplished and is widely used, an overall seamless integration of systems used in radiology and HIS is still under development on several fronts. DICOM has allowed images from multiple vendors and modalities to be archived and viewed in a picture archiving and communication system (PACS). The interface between order information from the RIS/HIS and radiology modalities has been defined in the DICOM Modality Worklist (http://medical.nema.org) and has been implemented by most vendors with varying degrees of success. This allows the technologist to choose an ordered study for a patient from a list provided by the HIS/RIS instead of duplicating data entry. If the HIS/RIS does not support an interface to DICOM Modality worklist, the user will be required to purchase an interface system, frequently called a "broker," which receives HL7 order messages and translates them into DICOM. Modalities query the broker for studies and the technologist picks from the list with a point and click mechanism or a bar-code reader scanning from the request form. The use and satisfactory actions of DICOM Modality Worklist varies on the manufacturer.

When Modality Worklist is not available on the modality, a different interface maybe used. In this case, the technologists enter patient and study information on the console and when the images are sent to the PACS, they are intercepted by a "relay" station that verifies the information with feeds from HIS/RIS. ³ This relay station can verify correct spelling and complete information before the images are sent to an archive or viewing station. If no match can be found, the study will have to be corrected manually by a PACS administrator or quality control technologist.

Some institutions have integrated HIS information with PACS images, although these implementations are still far from common. The most ambitious of these is the Composite Healthcare Computer System (CHCS), the Medical Diagnostic Imaging System (MDIS), and the Digital Imaging System (DIM) initiatives by the Department of Defense. These systems are in use in more than 20 military facilities throughout the world. ^4,5

FUTURE INTEGRATION EFFORTS

Integrating the Healthcare Enterprise (IHE) is an initiative of the Radiological Society of North America (RSNA) and the Healthcare Information and Management Systems Society (HIMSS). IHE provides an environment that promotes open discussion among medical information systems vendors, healthcare providers, administrators, standards groups, and professional societies on integrating heterogeneous information systems. IHE is not a standard, instead it promotes the use of existing standards (DICOM, HL7, ISO) to achieve interoperability. ^6-10

The goal is to provide an environment in which all information systems work together seamlessly to increase efficiency, reduce errors, and, ultimately, improve patient care. A typical scenario would include placing the order (using correct codes and appropriateness criteria), and scheduling the examination. The exam would then be performed and images sent to the PACS and RIS/HIS or other locations for viewing. Proper viewing would be guaranteed through tools that maintain the consistency of presentation in terms of grayscale, user annotations, available image manipulation functionality, etc. If a single acquisition was performed that needed to be sent to different locations for interpretation (as in the case of the chest, abdomen, pelvis CT study), the study would be separated so each radiologist would get the correct slices. Access to the patient's clinical information (such as lab reports, radiology reports, etc.) would be available when viewing images. Key images would be identified so clinicians reviewing the report would be able to see annotated images illustrating the findings. Reporting could be done using voice recognition and a structured report would be produced. This would all be done without the use of paper because worklists would be available for technologists operating equipment and radiologists using diagnostic workstations and voice recognition. If there was a discrepancy in patient information, it could be corrected after the fact and reconciled on all information systems..

IHE defines seven integration profiles that describe categories of interoperability and provide a framework for discussion among vendors of heterogeneous information systems. The description of the seven integration profiles is shown the Table 1. Each of the profiles describes a portion of the total study workflow. Interoperability implemented using the IHE format encourages a tight integration of information systems so data is efficiently kept uniform throughout the HIS. In addition, IHE encourages vendors to work together creating global solutions instead of each institution building a custom integration.

Integration in radiology has come a long way in the nearly 20 years since ACR-NEMA first set out to specify an integration standard. Many institutions have either a limited or complete PACS and most of them have some degree of integration with the HIS/RIS. We expect to have total integration available in many institutions during the first decade of the 21st century. n