Digital radiography of subtle pulmonary abnormalities: an ROC study of the effect of pixel size on observer performance

Forty conventional radiographs with examples of mild interstitial infiltrates and subtle pneumothoraces and 40 normal studies of the chest were selected and digitized, with pixel sizes of 1.0, 0.5, 0.2, and 0.1 mm. Observer performance tests were carried out using receiver operating characteristic analysis. Conventional radiographs and digitized images were compared. The results indicate that, in such cases, diagnostic accuracy increases significantly as the pixel size is reduced, at least to the 0.1-mm level. We conclude that, for digital systems using screen-film or similar image receptors, use of a pixel size substantially larger than 0.1 mm may result in some loss of diagnostic accuracy.     

Digitization of conventional x-ray films]

The diagnostic value of a digitization system for analogue films based on a charge-coupled-device (CCD) scanner with adjustable resolution of 2.5 or 5 lp/mm was assessed. Some 110 skeletal radiographs, 50 contrast studies, including 25 of patients with Crohn's disease, and 70 abdominal plain films before and after successful lithotripsy for renal stones were digitized. Receiver operating characteristic (ROC) studies showed improved detection of cortical and trabecular defects with contrast-optimized digitized films. Edge enhancement algorithms yielded no additional information. Inflammatory lesions of Crohn's disease were detected equally well by conventional films and digitized images. A statistically significant improvement (p less than 0.05) in the recognition of disintegration after lithotripsy was found for the processed digitized images. On the basis of this initial investigation, the digitization system tested appears to be useful for diagnostic purposes, especially when contrast resolution plays an important part, as with stone fragmentation or cortical defects.     

Spatial resolution requirements for digital chest radiographs: an ROC study of observer performance in selected cases

Thirty-eight selected clinical radiographs were digitized and displayed on a 1,024-line monitor at pixel sizes of 1.6, 0.8, 0.4, and 0.2 mm. Eighteen experienced radiologists assessed the radiographs and digital images, which included 12 examples of abnormal solitary nodular density, ten examples of septal lines, and 16 controls, six of which showed diffuse lung abnormalities. For each level of spatial resolution and for film reading, observers gave their decision confidence on a sliding scale of probability. Receiver operating characteristic curves were generated from these data. It was found that while spatial resolution requirements for solitary nodules were not critical for pixel sizes at or below 0.8 mm, the requirement for septal lines was likely to be 0.4 mm (1.25 line pairs/mm).     

Subperiosteal resorption: effect of full-frame image compression of hand radiographs on diagnostic accuracy.

Image compression is essential to handle a large volume of digital images, including computed tomographic, magnetic resonance, computed radiographic, and digitized images in a digital radiology operation. Developed during the past few years, full-frame bit allocation performed with the cosine transform technique has been proved to be an excellent irreversible image compression method. This article describes the effect, on the accuracy of diagnosis of subperiosteal resorption, of using the hardware compression module to produce hand radiographs. Receiver operating characteristic analysis of the interpretation of 71 radiographs by five observers demonstrated that there is no statistically significant difference in diagnostic accuracy between the original radiographs and compressed and reconstructed images obtained with a compression ratio as high as 20:1.     

Comparative studies on physical characteristics and clinical efficacy of conventional and digitized chest radiographs

Comparative studies were performed between digitized and conventional radiographs of the chest in terms of their physical characteristics and diagnostic efficacy. The purpose of these studies is to confirm the diagnostic capability of digitized image whether it can use for primary diagnosis in routine works. The results of two studies show good correlation each other. It is strongly suggested that the conventional chest radiographs should be digitized with 100 microns pixel in 12 bit density resolution.     

Detection of pneumothorax: comparison of digital and conventional chest imaging

To investigate radiologists' performance at interpreting digital radiographic images, we compared the detectability of pneumothoraces on computed radiographic chest images with 0.2-mm pixel size (2.5 Ip/mm) with their detectability on matched conventional screen-film images (5 Ip/mm). Eight radiologists reviewed 50 computed and 50 screen-film chest radiographs from 25 patients with pneumothoraces and 25 patients with other (or no) abnormalities. Four of the readers who best detected pneumothoraces on screen-film examinations performed worse when interpreting computed radiographic studies; the other four readers detected pneumothoraces similarly with both techniques. No relationship was found between the size of a pneumothorax and its likelihood of detection by either technique. These results raise concerns about implementing computed radiography for comprehensive chest imaging.     

Evaluation of CRT images of digitized film angiographs

X-ray sheet film images of the test chart, the vascular phantom and angiography were digitized at sampling pitch of 0.2 mm and 0.15 mm using film digitizer TFR-01 (Toshiba) and transferred to a device for image storage and display system with 1635-line display monitor (TDF-500AS, Toshiba). Comparison of image qualities between film- and CRT-images was performed in fundamental and clinical studies. Resolution of the test chart image of conventional radiography was worse on CRT than on the original film, although it was improved when film image was digitized at resolution of 0.15 mm/pixel in comparison with that at resolution of 0.2 mm/pixel. Moiré stripes which occurred due to interference were found on CRT images taken using a grid technique. On CRT images of X-ray sheet film using direct magnification technique moiré stripes were not produced because of non grid technique, and the resolution approached that of the original film. In the study using vascular phantom, the optimal image on CRT could be obtained by various image processing procedures, and image quality on CRT with resolution of 0.15 mm approached that of original film. In case of direct magnification CRT images were superior to film images. Subtraction image of the vascular phantom at resolution of 0.2 mm/pixel was obtained on CRT and compared with film subtraction image. On conventional subtraction CRT image moiré stripes impaired the image quality in comparison with the film subtraction. However, magnification subtraction image of the vascular phantom on CRT was superior to the film subtraction. The results obtained in the test chart studies and phantom studies were also confirmed in clinical studies using various kind of angiograms. In addition, ROC study using clinical angiograms showed no significant statistical differences between the original film and CRT image even with 0.2 mm matrix size. Angiographic image on CRT at resolution of 0.15 mm/pixel or less is available for clinical use in place of conventional film image.     

Digital camera resolution and proximal caries detection

OBJECTIVES: To evaluate the diagnostic accuracy of proximal caries detection from digitized film images captured by a digital camera at different resolution settings. METHODS: Twenty-five periapical radiographs of 50 premolar and 25 molar teeth were photographed using a digital camera, Sony Cyber-shot, DSC-S75 at three different resolution settings: 640 x 480, 1280 x 960 and 1600 x 1200. Seventy-five digital images were transferred to a computer, saved and opened using ACDSee software. In addition, a PowerPoint slide was made from each digital image. Five observers scored three groups of images (the films, the displayed 1:1 digital images on the ACDSee software, and the PowerPoint slides) for the existence of proximal caries using a 5-point confidence scale, and the depth of caries on a 4-point scale. Ground sections of the teeth were used as the gold standard. Az values under the receiver operating characteristic (ROC) curve of each group of images and at different resolutions were compared using the Friedman and Wilcoxon signed rank tests. Mean different values between the lesions' depth interpreted by the observers and that of the gold standard were analysed. RESULTS: Films showed the highest Az values. Only the 1280 x 960 images on the ACDSee software showed no significant difference of the Az value from the films (P=0.28). The digital images from three resolution settings on the PowerPoint slides showed no significant differences, either among each other or between them and the films. For caries depth, the 1280 x 960 images showed lower values of mean difference in enamel lesions compared with the other two resolution groups. CONCLUSIONS: This study showed that in order to digitize conventional films, it was not necessary to use the highest camera resolution setting to achieve high diagnostic accuracy for proximal caries detection. The 1280 x 960 resolution setting of the digital camera demonstrated comparable diagnostic accuracy with film and was adequate for digitizing radiographs for caries detection.     

Digital gastrointestinal imaging: the effect of pixel size on detection of subtle mucosal abnormalities

Five radiographs of double-contrast colon examinations demonstrating subtle mucosal changes of inflammatory bowel disease and five radiographs of healthy colonic mucosa were selected and digitized to four levels of resolution. Pixel sizes of 0.1 mm, 0.2 mm, 0.4 mm, and 0.8 mm were used. Ten radiologists interpreted the images, which were displayed on laser-printed film. Analysis of variance with repeated measures was performed and receiver operator characteristic curves were determined. The results demonstrate that the sensitivity in detecting subtle mucosal abnormalities improved as the resolution improved, with the best sensitivity at the highest resolution; more experienced readers detected details well even at the poorer levels of resolution; the resolution necessary for successfully evaluating the colonic mucosa was lower than expected; and given low noise levels, the matrix size used in conventional television fluoroscopy would be adequate for mucosal evaluation.     

Pneumothorax and other lung diseases: effect of altered resolution and edge enhancement on diagnosis with digitized radiographs

Prior studies have shown that pneumothorax is one of the more difficult entities to diagnose with digitized radiography. This study was designed to test whether increasing resolution from 1.25 to 2.5 line pairs per millimeter (lp/mm) and image processing (edge enhancement from unsharp masking) would increase accuracy and confidence in the diagnosis of pneumothorax, as well as normal cases and other forms of lung disease. Conventional radiographs were digitized with use of a laser reader and then reformatted as film hard copy. Eleven observers read 35 cases reformatted in three different ways (1.25 lp/mm, 2.5 lp/mm, 1.25 lp/mm unsharp mask). The images with finer resolution (2.5 lp/mm) and unsharp mask images were superior to those with coarser resolution (1.25 lp/mm) for the diagnosis of pneumothorax. There was no difference in diagnostic accuracy for normal patients. For abnormalities other than pneumothorax, the unsharp mask images were significantly worse. Confidence in the diagnosis of pneumothorax and other abnormalities was highest with the finest resolution (2.5 lp/mm).     

Clinical evaluation of life size image of Fuji computed radiography for detection of diffuse interstitial lung diseases]

To evaluate the diagnostic accuracy of Fuji Computed Radiography (FCR) in the detection of interstitial pulmonary infiltrates, FCR life-size images at a pixel size of 0.1 mm were compared with conventional radiographs taken on the same day. Seventeen radiologists assessed the radiographs and FCR images of 56 cases, including 39 cases of various interstitial lung diseases such as interstitial pneumonia, pulmonary abnormalities associated with collagen disease, sarcoidosis, multiple pulmonary metastases, diffuse panbronchiolitis and pulmonary emphysema, and 17 normal controls. All of the pulmonary abnormalities were confirmed by high resolution CT. Observer performance tests were carried out using receiver operating characteristic analysis. In 21 cases of increased pulmonary density revealed by high resolution CT, FCR was significantly superior to conventional radiographs in the detection of reticular or linear shadows. In 11 cases of subtle interstitial abnormalities, there was no difference between FCR and conventional radiographs in the detection of any pulmonary abnormality, ground-glass opacities and reticular or linear shadows. There was also no difference between the two images in the detection of diffuse nodular shadow and pulmonary emphysema. These results indicate that FCR life-size images at a pixel size of 0.1 mm are useful for the detection of diffuse interstitial lung diseases.     

Temporal subtraction for the detection of hazy pulmonary opacities on chest radiography

OBJECTIVE: The purpose of this study was to evaluate the accuracy of temporal subtraction with a commercially available computer-assisted diagnosis system for the detection of multifocal hazy pulmonary opacities on chest radiographs, which are sometimes difficult to detect directly on chest radiographs. MATERIALS AND METHODS: Thirty healthy patients and 30 patients with new multifocal hazy pulmonary opacities that were confirmed by serial chest CT examinations were evaluated with and without temporal subtraction images. Chest radiographs were taken from either film-screen or digital radiography images and were digitized with a spatial resolution of 0.171 mm per pixel. Temporal subtraction images were produced by an iterative image-warping technique. We designed an observer performance study in which observers (six chest radiologists and four residents) indicated their confidence level for the presence or absence of hazy pulmonary opacities on two sets of images, current and previous radiographs only (set A), and current and previous radiographs with temporal subtraction images (set B). Receiver operating characteristic curves were generated. RESULTS: For chest radiologists, observer performance with set B (with temporal subtraction images; A(z) = 0.947) was superior to that with set A (without temporal subtraction images; A(z) = 0.916) (p < 0.05). For residents, no statistically significant difference was found between sets A and B. CONCLUSION: The temporal subtraction technique clearly improves diagnostic accuracy for the detection of multifocal hazy pulmonary opacities on chest radiographs, especially when the observers are experienced chest radiologists who have sufficient skill to evaluate the patient's condition as revealed on the images.     

Clarity and diagnostic quality of digitized conventional intraoral radiographs

OBJECTIVES: To compare digital images of conventional radiographs with the original radiographs for perceived clarity of endodontic files, periapical lesions and carious lesions, and to establish the diagnostic value of the digital images. METHODS: Four groups of ten radiographs were used: standardized bitewings demonstrating carious lesions, periapical radiographs of apical lesions, periapical radiographs showing endodontic files of various sizes at working length, and standardized periapical radiographs with size 08 files at working length. Radiographs were photographed using an Olympus C 2500-L digital camera and a Nikon D1X digital camera and were scanned using a Nikon Supercoolscan 4000 ED film/slide scanner. The digital images were then transferred to a Toshiba Satellite 2210 laptop. Three general dental practitioners compared each conventional radiograph with the three matching digital images. Images were ranked for clarity and were assessed for diagnostic quality. Data were analysed using General Estimating Equations. RESULTS: The clarity and diagnostic quality of the conventional radiographs were superior to the digital images produced by the three techniques (P < 0.001). No significant difference was found between the Nikon D1X and Camedia 2500-L cameras for clarity or diagnostic quality. The scanner was equivalent to the Camedia 2500-L camera for diagnostic quality, but was otherwise inferior to both cameras. CONCLUSIONS: Digitizing conventional radiographs using current high-grade digital cameras or scanners does not produce images of diagnostic quality. Improved resolution of viewing monitors is necessary to fully harness the potential of digital technology.     

Digital subtraction radiography in artificial recurrent caries detection.

The diagnostic accuracy of digital subtraction radiography in detection of artificial recurrent caries lesions was assessed in this project. The use of digital subtraction radiography has been shown to markedly increase the accuracy of the detection of destruction in the periodontal bone, but the method has not been evaluated in secondary caries detection. Defects of three different sizes, simulating recurrent caries, were sequentially prepared in the interproximal cavity preparation margins of 28 teeth. Two composite restorative materials with different radiographic densities were used as posterior restorations, and a radiograph was obtained of each defect size and restorative material. The radiographs were digitized and subtracted from the reference images, and the conventional radiographs and the subtraction images were evaluated by seven observers. The data were analysed with ROC statistics. Subtraction radiography was found to be superior to conventional radiography in recurrent caries detection, mainly by reducing the false-positive diagnoses. The radiopacity of the restorative material had a significant effect on accuracy with conventional but not with subtraction radiography.     

Conventional radiographs vs digitized radiographs: image quality assessment

OBJECTIVES: The aim of this study was to compare image quality characteristics of conventional radiographs and their digital counterparts. METHODS: 100 conventional radiographs (E-speed) were used for the evaluation of density and contrast. The radiographs were made using a range of exposures under standardized conditions and were compared with digital images that resulted from scanning the above radiographs with a commercially available scanner. Resolution was evaluated with an experiment using 50 film packets and a resolution target exposed with different exposure times, using both the original radiographs and their scanned equivalents. RESULTS: The digitized radiographs appeared to be of higher density than the conventional ones. Moreover, they demonstrated a narrower density range. Resolution was similar for both types of images. CONCLUSIONS: Although there is an agreement with the literature that the digitized radiographs are of higher density, further investigation is required to detect the various factors that may have an effect on the quality of the digitized images.     

Pulmonary nodule detection of chest phantom with CRT images--comparison with conventional radiography (film-screen system) and FCR (Fuji Computed Radiography) hardcopy

Conventional Radiography, Fuji Computed Radiography (FCR) hardcopy and CRT images were evaluated about the detectability of pulmonary nodule using the chest phantom. Conventional and digitized chest radiographs (FCR) were used, including 45 normal cases and 45 abnormal cases with a variety pulmonary nodule. Observer performance tests were conducted to compare the effects on diagnostic accuracy of Conventional Radiography, FCR hardcopy and CRT images, and diagnostic accuracy was determined with receiver operating characteristic (ROC) analysis. ROC study was performed in which six radiologists were asked to locate nodule on three modalities. Two CRT monitors were used to observed CRT images. The left side CRT monitor was divided into two parts and displayed two images which were done as same image processing as FCR hardcopy. The right side CRT monitor was divided into four parts and displayed four images as a subsidiary diagnosis. The upper two images were displayed to diagnose the nodule of cardiac and diaphragmatic area, and lower two images were displayed to diagnose the nodule of the lung field. The results were summarized as follows: 1) CRT images were superior in sensitivity (78.5%) to the others, and FCR hardcopys were superior in specificity (95.9%). 2) About the accuracy of 5 and 8 mm in diameter nodule detection, there was no significant difference among three modalities. 3) Diagnostic accuracy of 3 mm in diameter nodule detection was significantly greater with digitized radiographs (FCR 44.4%, CRT images 54.4%) than with conventional radiograph (17.7%).(ABSTRACT TRUNCATED AT 250 WORDS)     

Diagnostic quality of portable abdominal radiographs in neonates with necrotizing enterocolitis: digitized vs nondigitized images

Digital manipulation of radiographic images has prompted significant interest because of the potential for improving image quality and diagnostic accuracy. We compared conventional and digital radiographs in the evaluation of neonatal necrotizing enterocolitis (NEC). Fifty abdominal radiographs in neonates with suspected or autopsy-confirmed NEC and 50 similar radiographs of neonates without suspicion of NEC were digitized. Definition of intraabdominal anatomy was optimized by window width and level adjustment. Hard-copy radiographs of the digitally manipulated images were then produced by a laser printer. Twelve general radiologists each evaluated, without benefit of clinical information, a random mixture of 50 cases of NEC and normal controls, with both conventional and digital images for each case, for a total of 100 radiographs reviewed. Each image was evaluated for overall suspicion of NEC and the presence and severity of six radiographic signs of NEC. The radiologists also rated their confidence in their assessments. The results were compared with those from a similar analysis by an experienced pediatric radiologist to ensure validity of image evaluation. No statistically significant differences were found between the conventional and digital imaging formats for the assessment of the signs of NEC (p = .15) or for determination of the overall suspicion of NEC (p = .07). Our results show the digitized and the conventional, nondigitized radiographs to be at least equally useful for evaluating the radiographic signs of NEC and suggesting an appropriate diagnosis.     

Morphometric analysis of experimentally induced periapical lesions: radiographic vs histopathological findings

OBJECTIVES: To evaluate the accuracy and reliability of conventional (Kodak Ektaspeed Plus film) and digitized radiographic images to detect the presence as well to estimate the size, as measured by an image analysis programme, of periapical radiolucencies induced in dog teeth in comparison with the histomorphometric data obtained from the same lesions by conventional and fluorescence microscopy. METHOD: After the removal of pulp, the root canals of five premolars from the same animal were left exposed for 7 days after which they were sealed for 60 days. At day 53, three more premolars were opened and left exposed to the oral cavity for 7 days. Intact premolars were used as control. Conventional radiographs were taken at day 0, day 7, day 30, day 45 and day 60. Morphometry in digitized radiographic images and histological sections were compared at day 7 and day 60 after setting the experimental series. RESULTS: Radiographically, periapical lesions were only detected 30 days after coronal sealing. A progressively increasing radiolucent lesion area was observed at day 45 and day 60. Histopathologically, 7 days after pulp removal dense inflammatory infiltrate and root resorption in the periapical region was observed. At day 7 and day 60, the lesion sizes were similar when evaluated by both conventional and fluorescence microscopy. Lesion size was about 20% larger in digitized radiographs in comparison with histological measurements. CONCLUSIONS: Although image digitization could not improve the detection of the early stages of periapical lesions, it provides a valuable quantitative assessment of extensive periapical lesions. In addition, fluorescence light microscopy enhances the visualization of the apical and periapical structures and seems to be a highly useful tool for histological evaluation, valuable for both qualitative and quantitative studies of periapical disease.     

Influence of film digitization on radiological interpretation

Our objective was to evaluate the influence of changing from analogue to digital imaging on interobserver and intraobserver image interpretation. Three radiologists interpreted 96 three image series of occipitomental radiographs of paranasal sinuses from the films and from the corresponding digitized images from the screen. Images were classified according to degree of abnormality as either normal, with mucosal thickening of less than 5 mm, with mucosal thickening of 5 mm or more, total opacity, air-fluid level or polyp or cyst of maxillary sinuses. In the present study we found that there were more differences between two radiologist's interpretations with a single method than in a single radiologist's interpretations between the methods, although radiologists interpreted fewer pathological findings from the digitized images than from the corresponding films. Our data show that the results of image interpretation are preferentially dependent on the reader rather than on the method of reading.     

"Single-exposure" dual energy digital radiography in the detection of pulmonary nodules and calcifications

We studied the detectability of mineralized and non-mineralized simulated pulmonary nodules with dual energy digital radiography. "Soft tissue" and "bone" images (pixel size = 0.2 mm, 10 bits deep) were obtained with subtraction image processing after a single simultaneous exposure (100 kVp, 8 mAs, 17 mR skin exposure dose) of two storage phosphors with an interleaved 0.9 mm copper wafer. Three classes of paraffin-based nodules (0.5 to 3.0 cm) of varying mineral concentration (0, 120 and 190 mg/cm3 K2HPO4) were randomly positioned on the chest wall of two healthy volunteers to simulate calcified and non-calcified nodules. The average receiver operating characteristics (ROC) area of six readers (n = 2880 observations) showed that digital "bone" images (ROC area: 0.77 +/- 0.03) were significantly better (P less than 0.04) than conventional radiographs (OC Film, Lanex medium screens, 141 kVp, 19 mR skin exposure dose) (ROC area: 0.71 +/- 0.05) in detecting calcification in nodules. The unsubtracted digital images of lower kilovoltage were not superior to the 141 kVp conventional radiographs in a subgroup of two readers (ROC area: 0.73 +/- 0.02). Digital "soft tissue" images were equivalent to conventional chest radiographs in detecting soft tissue pulmonary nodules (ROC areas: 0.92 +/- 0.04 and 0.92 +/- 0.05, respectively.