Objective Image Quality Measures Research Articles

Objective Image Quality Measurements and the EC Quality Criteria for Lumbar Spine Radiographs

A series of AP lumbar spine radiographs taken in four different hospitals were examined and compared with EC Quality Criteria for radiographic technique and film density. Two of the hospitals assessed had modern X ray equipment installed within the last five years and two hospitals had much older X ray equipment. Individual patient doses were measured using thermoluminescence dosemeters calibrated following the UK National Dose Protocol. All the radiographs were reported as being clinically acceptable. The radiographic techniques used mostly complied with the EC quality criteria. Mean doses were less than 50% of the recommended levels. The films were digitised using a Lumisys film digitiser. The mean density of the AP radiographs was 1.24 which compared well with the density measured using a 20 cm water phantom and was within the target range 1.0 to 1.4 recommended in the Quality Criteria. The mean density of the lateral films was 1.4. The spread of mean density was much narrower in the two hospitals with AEC devices. Densities of four anatomical regions were measured and differences between adjacent regions were calculated and found to be small.

Noise characterization of block-iterative reconstruction algorithms: I. Theory.

Researchers have shown increasing interest in block-iterative image reconstruction algorithms due to the computational and modeling advantages they provide. Although their convergence properties have been well documented, little is known about how they behave in the presence of noise. In this work, we fully characterize the ensemble statistical properties of the rescaled block-iterative expectation-maximization (RBI-EM) reconstruction algorithm and the rescaled block-iterative simultaneous multiplicative algebraic reconstruction technique (RBI-SMART). Also included in the analysis are the special cases of RBI-EM, maximum-likelihood EM (ML-EM) and ordered-subset EM (OS-EM), and the special case of RBI-SMART, SMART. A theoretical formulation strategy similar to that previously outlined for ML-EM is followed for the RBI methods. The theoretical formulations in this paper rely on one approximation, namely, that the noise in the reconstructed image is small compared to the mean image. In a second paper, the approximation will be justified through Monte Carlo simulations covering a range of noise levels, iteration points, and subset orderings. The ensemble statistical parameters could then be used to evaluate objective measures of image quality.

Scaling Subjective Impressions of Quality

To assure that the 600 dpi Heidelberg Digimaster 9110 Network Imaging System would deliver best-in-class image quality, extensive image measurement and assessment—including customer surveys and benchmarking—was conducted throughout the engineering and design phases. This process continues for every machine as it rolls through the final assembly area.A significant contributor to the Digimaster 9110's superior image quality was the development of meaningful image quality specifications and requirements. To develop these specifications, well-designed scaling surveys were of paramount importance. These surveys are the keystones to demonstrating the veracity of virtually all objective image quality measures. Unless an objective measure of quality can be carefully correlated with observers' subjective impressions of quality, the metric's usefulness is limited. That is, without establishing a correlation to visual quality, how is one to know if a new toner formulation, for example, produces a meaningful improvement in line sharpness? This paper discusses the relative utility of the four basic types of scales: nominal, ordinal, interval, and ratio. Methods for determining these scales are given along with fundamentals for designing single- and multiplestimulus scaling experiments. Specific examples are included showing how correlation with the subjective impression of quality was established for several metrics of black text image quality.

Role of phase information and eye pursuit in the detection of moving objects in noise.

As part of an ongoing study that uses objective image quality measures to optimize medical imaging x-ray fluoroscopy, we investigated two basic features of the detection of moving cylinders that mimic arteries, catheters, and guide wires. First, we compared detection with and without a phase cue consisting of a nearby alternating light and dark square. Depending on object size and velocity, phase cuing improved detection from 1% to 15% and gave an average of 6%, an effect much smaller than the 38% predicted from a Monte Carlo simulation of the ideal observer. Evidently, humans were limited in their ability to incorporate knowledge of the phase cue. Second, we evaluated the effect of eye pursuit of a fixation point that moved with the target. In general, motion at the highest velocity degraded (74%) and enhanced (68%) detection of small and large objects, respectively. With eye pursuit, both effects were substantially reduced in a manner consistent with a reduced retinal velocity. Our data compared favorably with a human observer model that included a spatiotemporal contrast sensitivity response and smooth-pursuit eye movements with a gain of 0.8. These mechanisms of perception are thought to be present in coronary artery x-ray fluoroscopy imaging, where phase information is available from the moving heart and where motion markers are available from x-ray opaque markers incorporated in thin catheters and guide wires.

A Case Study into the Effect of Radiographic Factors on Image Quality and Dose for a Selenium Based Digital Chest Radiography System

Thoravision, the first commercially available digital radiography system utilising amorphous selenium, has been evaluated from both clinical and physical perspectives. The advent of digital radiography systems has made the optimisation of both image quality and dose an important problem. Initial studies, using the manufacturer's recommended factors, demonstrated that the image quality, measured both objectively and subjectively, was significantly superior to conventional screen-film images. However, some questions still remained regarding the system's performance. A second study using alternative factors showed that the Thoravision always produced acceptable diagnostic images, regardless of the factors, and there are advantages to using lower kVp, objective measurements of image quality confirm this. It is clear that no single set of radiographic factors and image processing algorithms will produce optimised images for every patient and further elucidation of the problem is required. The difference between acceptable and a good clinical image is also highlighted and the implications discussed.

Objective image quality measures and their applications in segmentation evaluation

Image segmentation is a critical step of image analysis. Segmentation evaluation is an effective procedure for studying the performance of segmentation techniques, in which quality measure plays an important role. This paper presents a group of new objective quality measures for segmentation evaluation and compares their performances. In addition, to verify the effectiveness of these new measures, an appropriate classification of segmentation is proposed. According to this classification, several representative algorithms from different categories are selected for comparison testing. Some valuable results are obtained and presented.

Objective image quality measure for block-based DCT coding

Block-based discrete cosine transform (DCT) coding has been widely adopted in image/video compression standards, such as JPEG and MPEG. Such coding will produce a noticeable artifact known as the blocking effect at very low bit rate applications. In the literature, there have been various postprocessing methods proposed to reduce the blocking effect. However, there is no suitable objective criterion evaluating the effectiveness of these various methods. This paper presents a new objective measure evaluating the postprocessing methods. The proposed measure also provides an analytical result on the underlying sources of artifacts.

Optical Modulation Transfer and Contrast Sensitivity with Decentered Small Pupils in the Human Eye

Human observers experience a large decrement in visual acuity when a small artificial pupil is displaced from the center to the edge of the dilated natural pupil. This decrement in visual resolution, called the Campbell effect, has been attributed to the retina, the ocular optics, or a combination of the two. Given the uncertainty about the relative magnitudes of these two components over the range of spatial frequencies used in normal vision, we have obtained objective measurements of the retinal image quality and psychophysical measurements of visual performance, with decentered pupils. The contributions of monochromatic aberrations were determined by using double pass measurements of the modulus of the optical transfer function (MTF). For all of the observers, there was a substantial decrement in the MTF with decentering, showing that even when using a 1.5 mm pupil and appropriate spherical/cylindrical refractive corrections, there is a considerable contribution of monochromatic aberrations to the effect. We have compared these optical MTFs with the psychophysical contrast sensitivity functions (CSFs) measured under exactly the same conditions using green gratings generated on the screen of a color monitor. At the low and intermediate spatial frequencies considered (2–16 c/deg), we find the fall in the CSF is much greater than the fall in the monochromatic MTF, with the difference becoming greater as the spatial frequency increases. We show that this discrepancy can be mostly attributed to the effect of transverse chromatic aberration due to the bandwidth of the green stimulus used for the CSF measurements. In conclusion, the combination of the ocular transverse chromatic aberration and monochromatic aberrations accounts for the loss in visual sensitivity found with a decentered small pupil at low and intermediate spatial frequencies. Copyright © 1996 Elsevier Science Ltd.

Visual instrument image quality metrics and the effects of coma and astigmatism

We examine subjective contrast and edge sharpness discrimination in the presence of different amounts of coma, astigmatism, and combinations of these two aberrations. We also examine the correlation between subjective results and objective image quality measures. These include the integral of the modulation transfer function (MTF) across the frequency range of interest (MTFa), the Strehl ratio, the variance of the wave-front aberration, and the radius of 84% encircled energy of the point-spread function (R84). For the target orientation that is most affected by the aberrations, we find that the MTFa and R84 give good correlation, while the Strehl ratio and the variance do not. In addition, we find that the MTFa correlates with subjective results for all target orientations. We discuss the implications of the results on the automated assessment of the image quality of visual instruments.

Objective image quality measure derived from digital image power spectra

An objective image quality measure based on the digital image power spectrum of normally acquired arbitrary scenes is developed. This image quality measure, which does not require imaging either designed targets or a constant scene, utilizes the previously known invariance property for the power spectra of arbitrary scenes. The measure incorporates a representation of the human visual system, a novel approach to account for directional differences in perspective (scale) for obliquely acquired scenes, and a filter developed to account for imaging system noise as specifically evidenced in the image power spectra. The primary application is to assess the quality of digital images relevant to the image task of detection, recognition, and identification of man-made objects from softcopy displayed versions of visible spectral region digital aerial images. Experimental verification is presented demonstrating very good correlation (r=0.9) of this objective quality measure with visual quality assessments.

Imagery Interpreter Performance in the Comparison of Subjective Estimates of Photographic Image Quality

State-of-the-art research in image quality assessment has been oriented toward objective measures of image quality, requiring microdensitometers and computers. Cost and time constraints in the operational situation, however, place emphasis on man-dependent methods. This paper describes the performance of the image interpreter in a study comparing two of the more widely accepted Air Force subjective measures of image quality: tribar target resolution reading and visual edge matching. These techniques are described, interpreter certification is discussed, data derived from the application of each technique to a common imagery set are presented, and a comparison of the two methods reported.