Journal of the International Commission on Radiation Units and Measurements, Jul 1, 2003
Radiography of the chest is a fascinating topic. Volumes can be written about the application of ... more Radiography of the chest is a fascinating topic. Volumes can be written about the application of the examination to modern clinical medicine (Fraser et al., 1999). The presentations of disease on chest radiographs are myriad and often manifest the subtlety and nuance that many may still associate with radiological practice. Insight gained by practitioners long ago has equal relevance today and remains the basis for image interpretation to the present time. Students of medical radiographic science also find much of interest in the chest examination. No other radiographic examination has been studied as extensively. The application of image science to chest radiography produces many useful and informative results . At the same time, studies of chest radiography often demonstrate the limitations of radiological science, as the interaction between the image and the observer can still confound the best present analyses. This report describes the technical aspects of chest radiography, with emphasis on issues related to image quality. The discussions may therefore be of interest to radiologists and clinicians who diagnose chest disease, physicists, engineers and radiological technologists.
Journal of the International Commission on Radiation Units and Measurements, Jul 1, 2003
Az The area under an ROC curve, often used as an index of detectability or diagnostic accuracy. a... more Az The area under an ROC curve, often used as an index of detectability or diagnostic accuracy. aliasing artifacts Artifacts that may arise when an image is sampled at a spatial frequency that is less than required by the spatial frequency content of the image. The minimum sampling frequency needed to avoid aliasing artifacts is twice the maximum spatial frequency content of the image. auto-correlation function The auto-correlation function is the ensemble average of the joint second moment of an imaging process. The autocorrelation function of radiographic mottle indicates the magnitude and the texture of the noise pattern in a radiograph. bootstrap sensitometry A method to determine the H & D curve of a screen-film system by overlapping separate segments of the curve, where each segment is measured independently. This method was devised to measure the curve by using two parameter combinations such as tube voltage-time, tube currenttime, step wedgetime, step wedge-distance, instead of one parameter in order to cover a large range of X-ray exposure rates incident on the screen-film system. characteristic curve See H & D curve. contrast A broad concept representing the difference or the ratio in image signals, such as the optical density, transmittance or radiant exposure, at two selected locations. convolution In radiographic science, the mathematical procedure that transforms an object (or input) distribution by the line spread function or point spread function into an image (or output) distribution, when the imaging system is linear and shift invariant. crossover exposure The additional radiant exposure to the film emulsion caused by light that is emitted by the screen placed opposite that
The concept of selective exposure radiography encompasses those techniques which spatially modula... more The concept of selective exposure radiography encompasses those techniques which spatially modulate the incident x-ray field to produce a more uniform exit field arising from the patient. The resulting reduction in the dynamic range of the exposure field offers several advantages. In conventional radiography, selective exposure techniques allow all areas of the image to be placed in the linear portion of the film characteristic curve so that local contrast is maximized. With video systems, the reduction in dynamic range minimizes the impact of electronic noise behind the least transmissive regions of the patient. With both electronic and photographic detectors, selective exposure radiography is characterized by uniform quantum statistics and uniform scatter across the image. Several selective exposure techniques currently are being investigated. They include compensating filters placed manually in the x-ray field as well as fan-beam geometries in which the x-ray tube output is modulated with a feedback circuit to maintain constant exposure to the image receptor. At the University of Wisconsin, we have been investigating a digital system which uses an initial low-dose patient image to design an attenuator with transmission complimentary to that of the patient. The attenuator is fabricated for each patient and is positioned automatically in the x-ray beam prior to the acquisition of the final compensated image. The possible applications of this device include chest radiography, coronary angiography, subtraction angiography, and accurate digital videodensitometry.
Digital breast tomosynthesis is a form of limited angle tomography, in which section (slice) imag... more Digital breast tomosynthesis is a form of limited angle tomography, in which section (slice) images are produced from a series of discrete projection images acquired at different angles. Tomosynthesis can be useful in breast imaging by providing potentially better visibility of lesions over conventional mammography, especially in patients with dense breasts. This talk will cover the various physics aspects of DBT, including reconstruction algorithms, the importance of deblurring, and optimizing image acquisition parameters. Remaining important research questions in DBT will be presented and discussed. The presentation will also discuss MQSA Certificate extension process for currently approved digital breast tomosynthesis (DBT) systems. Training requirements, manufacturer required tests for Mammography Equipment Evaluation (MEE) as acceptance tests, and phantom imaging for the purpose of approval of certificate extension will be described. The talk will emphasize the specific tests where special attention must be given and will discuss how the techs should be advised to perform these tests. LEARNING OBJECTIVES 1. To understand the fundamentals of tomosynthesis reconstruction, including deblurring, algorithm choice, and optimization 2. To understand FDA's certificate extension process for DBT 3. To understand the requirements for MEE 4. To understand the required AEC tracking data Research sponsored in part by NIH, Siemens, and GE Healthcare.
The last few years have seen a rapid increase in the number of digital imaging devices produced f... more The last few years have seen a rapid increase in the number of digital imaging devices produced for radiographic applications. With digitized video/image intensifiers, computed radiography (CR), and more recently the advent of flat-panel imagers, an almost bewildering number of choices of digital imaging devices is available. The good news from this rapid progress in digital radiography is that devices are becoming available with image quality superior to that available just a few years ago. The challenge from this proliferation of digital devices is making the difficult decision about which device is most suitable for a particular application. Radiologists will rightly ask, âHow can I know which device I should purchase for my clinical imaging needs?â Imaging physicists, on the other hand, will be concerned with verifying the image quality specifications of the various manufacturers, developing appropriate algorithms to extend the utility of the devices, and developing appropriate clinical imaging protocols to take best advantage of the devices' imaging performance. Physicists and radiologists will need to work collaboratively to determine the appropriate applications for these devices, and the performance that may be expected from them. In order to assess the performance of these devices, it is necessary to consider both physical image-quality parameters as well as the observer's perceptual response. Both of these areas are more difficult with digital devices than with screen film. Because the image has been sampled, quantized, and processed, there are a number of changes necessary in traditional measurements of image-quality. Furthermore, because there is an almost unlimited degree of image processing that can be done to the images, it becomes more difficult to gauge observer acceptance of the resulting images. There are a number of physical and observer-based measurements which can be used to gauge image-quality, and these will be considered in this chapter. This chapter is divided into four sections. First, global measures of image-quality will be addressed, using measurements in Cartesian (or image-intensity) space. Second, measures in spatial frequency space will be considered. Third, methods of assessing image processing will be considered, and last, observer assessment will be addressed. Of course, the final stage in assessment of image-quality is the ability of a particular imaging device and clinical protocol to improve diagnostic accuracy. The issue of diagnostic accuracy, while very important, is an entire field of study in itself, and is therefore beyond the scope of this chapter. The interested reader is referred to other chapters in this volume on measuring diagnostic accuracy.
Digital x-ray tomosynthesis is a technique for producing slice images using conventional x-ray sy... more Digital x-ray tomosynthesis is a technique for producing slice images using conventional x-ray systems. It is a refinement of conventional geometric tomography, which has been known since the 1930s. In conventional geometric tomography, the x-ray tube and image receptor move in synchrony on opposite sides of the patient to produce a plane of structures in sharp focus at the plane containing the fulcrum of the motion; all other structures above and below the fulcrum plane are blurred and thus less visible in the resulting image. Tomosynthesis improves upon conventional geometric tomography in that it allows an arbitrary number of in-focus planes to be generated retrospectively from a sequence of projection radiographs that are acquired during a single motion of the x-ray tube. By shifting and adding these projection radiographs, specific planes may be reconstructed. This topical review describes the various reconstruction algorithms used to produce tomosynthesis images, as well as approaches used to minimize the residual blur from out-of-plane structures. Historical background and mathematical details are given for the various approaches described. Approaches for optimizing the tomosynthesis image are given. Applications of tomosynthesis to various clinical tasks, including angiography, chest imaging, mammography, dental imaging and orthopaedic imaging, are also described.
We have used a prototype digital beam attenuator (DBA) system to generate patient-specific digita... more We have used a prototype digital beam attenuator (DBA) system to generate patient-specific digitally-prepared compensating filters for chest radiography of a human subject. The compensated radiographs demonstrate substantially more information in areas such as the mediastinum and upper abdomen which normally are underpenetrated in conventional chest radiographs. The compensated image was acquired with high contrast, high speed film-screen receptors improving the visibility of pulmonary parenchymal detail while minimizing patient radiation exposure. Currently we are limited by a two-hour preparation time and position the attenuator manually. We are developing a second generation DBA system featuring fast (15 second) fabrication times and automatic positioning of the attenuator. We expect that these features will relieve some of the misregistration errors present in our initial examination.
ABSTRACT Correlated-polarity noise reduction (CPNR) is a novel noise reduction technique that use... more ABSTRACT Correlated-polarity noise reduction (CPNR) is a novel noise reduction technique that uses a statistical approach to reduce noise while maintaining excellent resolution and a “normal” noise appearance. It is applicable to any type of medical imaging, and we introduced it at SPIE 2011 for reducing dose three-fold in radiography while maintaining excellent image quality. In this current work, we demonstrate for the first time its use in reducing the noise in CT images as a means of reducing the dose in CT. Simulated chest CT images were generated using the XCAT phantom and Poisson noise was added to simulate a conventional full-dose CT image and a half-dose CT image. CPNR was applied to the half-dose images in projection image space, and then the images were reconstructed using filtered backprojection with a Feldkamp methodology. The resulting CPNR processed half-dose images showed essentially equivalent relative standard deviation in the central heart region to the full-dose images, and about 0.7 times that in half-dose images that were not processed with CPNR. This noise reduction was consistent with a two-fold reduction in dose that is possible with CPNR in CT. The CPNR images demonstrated virtually identical sharpness of vessels and no apparent artifacts. We conclude that CPNR shows strong promise as a new noise reduction method for dose reduction in CT. CPNR could also be used in combination with model-based iterative reconstruction techniques for yet further dose reduction.
ABSTRACT Correlated-polarity noise reduction (CPNR) is a novel noise reduction technique that use... more ABSTRACT Correlated-polarity noise reduction (CPNR) is a novel noise reduction technique that uses a statistical approach to reduce noise while maintaining excellent resolution and a “normal” noise appearance. It is applicable to any type of medical imaging, and we introduced it at SPIE 2011 for reducing dose three-fold in radiography while maintaining excellent image quality. In this current work, we demonstrate for the first time its use in reducing the noise in CT images as a means of reducing the dose in CT. Simulated chest CT images were generated using the XCAT phantom and Poisson noise was added to simulate a conventional full-dose CT image and a half-dose CT image. CPNR was applied to the half-dose images in projection image space, and then the images were reconstructed using filtered backprojection with a Feldkamp methodology. The resulting CPNR processed half-dose images showed essentially equivalent relative standard deviation in the central heart region to the full-dose images, and about 0.7 times that in half-dose images that were not processed with CPNR. This noise reduction was consistent with a two-fold reduction in dose that is possible with CPNR in CT. The CPNR images demonstrated virtually identical sharpness of vessels and no apparent artifacts. We conclude that CPNR shows strong promise as a new noise reduction method for dose reduction in CT. CPNR could also be used in combination with model-based iterative reconstruction techniques for yet further dose reduction.
A circuit has been constructed to perform spatial frequency filtration on DSA images at real-time... more A circuit has been constructed to perform spatial frequency filtration on DSA images at real-time video rates. The 10-bit device performs low-pass or high-pass filtering, and with external memory can perform bandpass and more sophisticated filtering. Pixels in the convolving kernel are weighted independendently in the x- and y-directions to provide a Gaussian-like convolving function. The kernel width ranges from 3-30 pixels and appropriate weighting yields a FWHM of the Gaussian kernel function as small as 0.85 pixel width for horizontal image widths of ≤256 pixels and as small as 1.25 pixel width for a horizontal image width of 512 pixels. Applications to be investigated include scatter and glare correction for videodensitometry, enhancement of arteries behind large opacified structures such as the ventricle and aorta, noise suppression in low-spatial frequency DSA exams, edge-enhancement of images, and partial-pixel shifting. Peli and Lim have suggested a more sophisticated algorithm which enhances high-pass filtration only in dark regions of an image. This and other techniques may be implemented with the current circuit to enhance small detail in highly opacified regions such as the ventricle, while leaving the rest of the image unaltered.
ABSTRACT Assessment of the resolution properties of nonlinear imaging systems is a useful but cha... more ABSTRACT Assessment of the resolution properties of nonlinear imaging systems is a useful but challenging task. While the modulation transfer function (MTF) fully describes contrast resolution as a function of spatial frequency for linear systems, an equivalent metric does not exist for systems with significant nonlinearity. Therefore, this preliminary investigation attempts to classify and quantify the amount of scaling and distortion imposed on a given image signal as the result of a nonlinear process (nonlinear image processing algorithm). As a proof-of-concept, a median filter is assessed in terms of its principle frequency response (PFR) and distortion response (DR) functions. These metrics are derived in frequency space using a sinusoidal basis function, and it is shown that, for a narrow-band sinusoidal input signal, the scaling and distortion properties of the nonlinear filter are described exactly by PFR and DR, respectively. The use of matched sinusoidal basis and input functions accurately reveals the frequency response to long linear structures of different scale. However, when more complex (multi-band) input signals are considered, PFR and DR fail to adequately characterize the frequency response due to nonlinear interaction effects between different frequency components during processing. Overall, the results reveal the context-dependent nature of nonlinear image processing algorithm performance, and they emphasize the importance of the basis function choice in algorithm assessment. In the future, more complex forms of nonlinear systems analysis may be necessary to fully characterize the frequency response properties of nonlinear algorithms in a context-dependent manner.
Journal of the International Commission on Radiation Units and Measurements, Jul 1, 2003
Radiography of the chest is a fascinating topic. Volumes can be written about the application of ... more Radiography of the chest is a fascinating topic. Volumes can be written about the application of the examination to modern clinical medicine (Fraser et al., 1999). The presentations of disease on chest radiographs are myriad and often manifest the subtlety and nuance that many may still associate with radiological practice. Insight gained by practitioners long ago has equal relevance today and remains the basis for image interpretation to the present time. Students of medical radiographic science also find much of interest in the chest examination. No other radiographic examination has been studied as extensively. The application of image science to chest radiography produces many useful and informative results . At the same time, studies of chest radiography often demonstrate the limitations of radiological science, as the interaction between the image and the observer can still confound the best present analyses. This report describes the technical aspects of chest radiography, with emphasis on issues related to image quality. The discussions may therefore be of interest to radiologists and clinicians who diagnose chest disease, physicists, engineers and radiological technologists.
Journal of the International Commission on Radiation Units and Measurements, Jul 1, 2003
Az The area under an ROC curve, often used as an index of detectability or diagnostic accuracy. a... more Az The area under an ROC curve, often used as an index of detectability or diagnostic accuracy. aliasing artifacts Artifacts that may arise when an image is sampled at a spatial frequency that is less than required by the spatial frequency content of the image. The minimum sampling frequency needed to avoid aliasing artifacts is twice the maximum spatial frequency content of the image. auto-correlation function The auto-correlation function is the ensemble average of the joint second moment of an imaging process. The autocorrelation function of radiographic mottle indicates the magnitude and the texture of the noise pattern in a radiograph. bootstrap sensitometry A method to determine the H & D curve of a screen-film system by overlapping separate segments of the curve, where each segment is measured independently. This method was devised to measure the curve by using two parameter combinations such as tube voltage-time, tube currenttime, step wedgetime, step wedge-distance, instead of one parameter in order to cover a large range of X-ray exposure rates incident on the screen-film system. characteristic curve See H & D curve. contrast A broad concept representing the difference or the ratio in image signals, such as the optical density, transmittance or radiant exposure, at two selected locations. convolution In radiographic science, the mathematical procedure that transforms an object (or input) distribution by the line spread function or point spread function into an image (or output) distribution, when the imaging system is linear and shift invariant. crossover exposure The additional radiant exposure to the film emulsion caused by light that is emitted by the screen placed opposite that
The concept of selective exposure radiography encompasses those techniques which spatially modula... more The concept of selective exposure radiography encompasses those techniques which spatially modulate the incident x-ray field to produce a more uniform exit field arising from the patient. The resulting reduction in the dynamic range of the exposure field offers several advantages. In conventional radiography, selective exposure techniques allow all areas of the image to be placed in the linear portion of the film characteristic curve so that local contrast is maximized. With video systems, the reduction in dynamic range minimizes the impact of electronic noise behind the least transmissive regions of the patient. With both electronic and photographic detectors, selective exposure radiography is characterized by uniform quantum statistics and uniform scatter across the image. Several selective exposure techniques currently are being investigated. They include compensating filters placed manually in the x-ray field as well as fan-beam geometries in which the x-ray tube output is modulated with a feedback circuit to maintain constant exposure to the image receptor. At the University of Wisconsin, we have been investigating a digital system which uses an initial low-dose patient image to design an attenuator with transmission complimentary to that of the patient. The attenuator is fabricated for each patient and is positioned automatically in the x-ray beam prior to the acquisition of the final compensated image. The possible applications of this device include chest radiography, coronary angiography, subtraction angiography, and accurate digital videodensitometry.
Digital breast tomosynthesis is a form of limited angle tomography, in which section (slice) imag... more Digital breast tomosynthesis is a form of limited angle tomography, in which section (slice) images are produced from a series of discrete projection images acquired at different angles. Tomosynthesis can be useful in breast imaging by providing potentially better visibility of lesions over conventional mammography, especially in patients with dense breasts. This talk will cover the various physics aspects of DBT, including reconstruction algorithms, the importance of deblurring, and optimizing image acquisition parameters. Remaining important research questions in DBT will be presented and discussed. The presentation will also discuss MQSA Certificate extension process for currently approved digital breast tomosynthesis (DBT) systems. Training requirements, manufacturer required tests for Mammography Equipment Evaluation (MEE) as acceptance tests, and phantom imaging for the purpose of approval of certificate extension will be described. The talk will emphasize the specific tests where special attention must be given and will discuss how the techs should be advised to perform these tests. LEARNING OBJECTIVES 1. To understand the fundamentals of tomosynthesis reconstruction, including deblurring, algorithm choice, and optimization 2. To understand FDA's certificate extension process for DBT 3. To understand the requirements for MEE 4. To understand the required AEC tracking data Research sponsored in part by NIH, Siemens, and GE Healthcare.
The last few years have seen a rapid increase in the number of digital imaging devices produced f... more The last few years have seen a rapid increase in the number of digital imaging devices produced for radiographic applications. With digitized video/image intensifiers, computed radiography (CR), and more recently the advent of flat-panel imagers, an almost bewildering number of choices of digital imaging devices is available. The good news from this rapid progress in digital radiography is that devices are becoming available with image quality superior to that available just a few years ago. The challenge from this proliferation of digital devices is making the difficult decision about which device is most suitable for a particular application. Radiologists will rightly ask, âHow can I know which device I should purchase for my clinical imaging needs?â Imaging physicists, on the other hand, will be concerned with verifying the image quality specifications of the various manufacturers, developing appropriate algorithms to extend the utility of the devices, and developing appropriate clinical imaging protocols to take best advantage of the devices' imaging performance. Physicists and radiologists will need to work collaboratively to determine the appropriate applications for these devices, and the performance that may be expected from them. In order to assess the performance of these devices, it is necessary to consider both physical image-quality parameters as well as the observer's perceptual response. Both of these areas are more difficult with digital devices than with screen film. Because the image has been sampled, quantized, and processed, there are a number of changes necessary in traditional measurements of image-quality. Furthermore, because there is an almost unlimited degree of image processing that can be done to the images, it becomes more difficult to gauge observer acceptance of the resulting images. There are a number of physical and observer-based measurements which can be used to gauge image-quality, and these will be considered in this chapter. This chapter is divided into four sections. First, global measures of image-quality will be addressed, using measurements in Cartesian (or image-intensity) space. Second, measures in spatial frequency space will be considered. Third, methods of assessing image processing will be considered, and last, observer assessment will be addressed. Of course, the final stage in assessment of image-quality is the ability of a particular imaging device and clinical protocol to improve diagnostic accuracy. The issue of diagnostic accuracy, while very important, is an entire field of study in itself, and is therefore beyond the scope of this chapter. The interested reader is referred to other chapters in this volume on measuring diagnostic accuracy.
Digital x-ray tomosynthesis is a technique for producing slice images using conventional x-ray sy... more Digital x-ray tomosynthesis is a technique for producing slice images using conventional x-ray systems. It is a refinement of conventional geometric tomography, which has been known since the 1930s. In conventional geometric tomography, the x-ray tube and image receptor move in synchrony on opposite sides of the patient to produce a plane of structures in sharp focus at the plane containing the fulcrum of the motion; all other structures above and below the fulcrum plane are blurred and thus less visible in the resulting image. Tomosynthesis improves upon conventional geometric tomography in that it allows an arbitrary number of in-focus planes to be generated retrospectively from a sequence of projection radiographs that are acquired during a single motion of the x-ray tube. By shifting and adding these projection radiographs, specific planes may be reconstructed. This topical review describes the various reconstruction algorithms used to produce tomosynthesis images, as well as approaches used to minimize the residual blur from out-of-plane structures. Historical background and mathematical details are given for the various approaches described. Approaches for optimizing the tomosynthesis image are given. Applications of tomosynthesis to various clinical tasks, including angiography, chest imaging, mammography, dental imaging and orthopaedic imaging, are also described.
We have used a prototype digital beam attenuator (DBA) system to generate patient-specific digita... more We have used a prototype digital beam attenuator (DBA) system to generate patient-specific digitally-prepared compensating filters for chest radiography of a human subject. The compensated radiographs demonstrate substantially more information in areas such as the mediastinum and upper abdomen which normally are underpenetrated in conventional chest radiographs. The compensated image was acquired with high contrast, high speed film-screen receptors improving the visibility of pulmonary parenchymal detail while minimizing patient radiation exposure. Currently we are limited by a two-hour preparation time and position the attenuator manually. We are developing a second generation DBA system featuring fast (15 second) fabrication times and automatic positioning of the attenuator. We expect that these features will relieve some of the misregistration errors present in our initial examination.
ABSTRACT Correlated-polarity noise reduction (CPNR) is a novel noise reduction technique that use... more ABSTRACT Correlated-polarity noise reduction (CPNR) is a novel noise reduction technique that uses a statistical approach to reduce noise while maintaining excellent resolution and a “normal” noise appearance. It is applicable to any type of medical imaging, and we introduced it at SPIE 2011 for reducing dose three-fold in radiography while maintaining excellent image quality. In this current work, we demonstrate for the first time its use in reducing the noise in CT images as a means of reducing the dose in CT. Simulated chest CT images were generated using the XCAT phantom and Poisson noise was added to simulate a conventional full-dose CT image and a half-dose CT image. CPNR was applied to the half-dose images in projection image space, and then the images were reconstructed using filtered backprojection with a Feldkamp methodology. The resulting CPNR processed half-dose images showed essentially equivalent relative standard deviation in the central heart region to the full-dose images, and about 0.7 times that in half-dose images that were not processed with CPNR. This noise reduction was consistent with a two-fold reduction in dose that is possible with CPNR in CT. The CPNR images demonstrated virtually identical sharpness of vessels and no apparent artifacts. We conclude that CPNR shows strong promise as a new noise reduction method for dose reduction in CT. CPNR could also be used in combination with model-based iterative reconstruction techniques for yet further dose reduction.
ABSTRACT Correlated-polarity noise reduction (CPNR) is a novel noise reduction technique that use... more ABSTRACT Correlated-polarity noise reduction (CPNR) is a novel noise reduction technique that uses a statistical approach to reduce noise while maintaining excellent resolution and a “normal” noise appearance. It is applicable to any type of medical imaging, and we introduced it at SPIE 2011 for reducing dose three-fold in radiography while maintaining excellent image quality. In this current work, we demonstrate for the first time its use in reducing the noise in CT images as a means of reducing the dose in CT. Simulated chest CT images were generated using the XCAT phantom and Poisson noise was added to simulate a conventional full-dose CT image and a half-dose CT image. CPNR was applied to the half-dose images in projection image space, and then the images were reconstructed using filtered backprojection with a Feldkamp methodology. The resulting CPNR processed half-dose images showed essentially equivalent relative standard deviation in the central heart region to the full-dose images, and about 0.7 times that in half-dose images that were not processed with CPNR. This noise reduction was consistent with a two-fold reduction in dose that is possible with CPNR in CT. The CPNR images demonstrated virtually identical sharpness of vessels and no apparent artifacts. We conclude that CPNR shows strong promise as a new noise reduction method for dose reduction in CT. CPNR could also be used in combination with model-based iterative reconstruction techniques for yet further dose reduction.
A circuit has been constructed to perform spatial frequency filtration on DSA images at real-time... more A circuit has been constructed to perform spatial frequency filtration on DSA images at real-time video rates. The 10-bit device performs low-pass or high-pass filtering, and with external memory can perform bandpass and more sophisticated filtering. Pixels in the convolving kernel are weighted independendently in the x- and y-directions to provide a Gaussian-like convolving function. The kernel width ranges from 3-30 pixels and appropriate weighting yields a FWHM of the Gaussian kernel function as small as 0.85 pixel width for horizontal image widths of ≤256 pixels and as small as 1.25 pixel width for a horizontal image width of 512 pixels. Applications to be investigated include scatter and glare correction for videodensitometry, enhancement of arteries behind large opacified structures such as the ventricle and aorta, noise suppression in low-spatial frequency DSA exams, edge-enhancement of images, and partial-pixel shifting. Peli and Lim have suggested a more sophisticated algorithm which enhances high-pass filtration only in dark regions of an image. This and other techniques may be implemented with the current circuit to enhance small detail in highly opacified regions such as the ventricle, while leaving the rest of the image unaltered.
ABSTRACT Assessment of the resolution properties of nonlinear imaging systems is a useful but cha... more ABSTRACT Assessment of the resolution properties of nonlinear imaging systems is a useful but challenging task. While the modulation transfer function (MTF) fully describes contrast resolution as a function of spatial frequency for linear systems, an equivalent metric does not exist for systems with significant nonlinearity. Therefore, this preliminary investigation attempts to classify and quantify the amount of scaling and distortion imposed on a given image signal as the result of a nonlinear process (nonlinear image processing algorithm). As a proof-of-concept, a median filter is assessed in terms of its principle frequency response (PFR) and distortion response (DR) functions. These metrics are derived in frequency space using a sinusoidal basis function, and it is shown that, for a narrow-band sinusoidal input signal, the scaling and distortion properties of the nonlinear filter are described exactly by PFR and DR, respectively. The use of matched sinusoidal basis and input functions accurately reveals the frequency response to long linear structures of different scale. However, when more complex (multi-band) input signals are considered, PFR and DR fail to adequately characterize the frequency response due to nonlinear interaction effects between different frequency components during processing. Overall, the results reveal the context-dependent nature of nonlinear image processing algorithm performance, and they emphasize the importance of the basis function choice in algorithm assessment. In the future, more complex forms of nonlinear systems analysis may be necessary to fully characterize the frequency response properties of nonlinear algorithms in a context-dependent manner.
Uploads