Temporal subtraction chest radiography

Radiologist are commonly required to compare a sequence of two or more chest radiographs of a given patient obtained over a period of time, which may range from a few hours to many years. In such cases, the task is one of detecting interval change. In the case of patients who have had a previous chest radiograph, an opportunity exists to enhance selectively areas of interval change, including regions with new or altered pathology, by using the previous radiographs as a subtraction mask. With temporal subtraction, the previous image is superimposed and registered with the current image, using automated two-dimensional warping to compensate for any differences in positioning. A “difference image” is then created, by subtracting the previous from the current radiograph. In this temporal subtraction image, areas that are unchanged appear as uniform gray, while regions of new opacity, such as due to pneumonia or cancer, appear as prominent dark foci on a lighter background. By cancelling out the complex anatomical background, temporal subtraction can provide dramatically enhanced visibility of new areas of disease.     

Computer-aided diagnosis and artificial intelligence in clinical imaging

Computer-aided diagnosis (CAD) is rapidly entering the radiology mainstream. It has already become a part of the routine clinical work for the detection of breast cancer with mammograms. The computer output is used as a "second opinion" in assisting radiologists' image interpretations. The computer algorithm generally consists of several steps that may include image processing, image feature analysis, and data classification via the use of tools such as artificial neural networks (ANN). In this article, we will explore these and other current processes that have come to be referred to as "artificial intelligence." One element of CAD, temporal subtraction, has been applied for enhancing interval changes and for suppressing unchanged structures (eg, normal structures) between 2 successive radiologic images. To reduce misregistration artifacts on the temporal subtraction images, a nonlinear image warping technique for matching the previous image to the current one has been developed. Development of the temporal subtraction method originated with chest radiographs, with the method subsequently being applied to chest computed tomography (CT) and nuclear medicine bone scans. The usefulness of the temporal subtraction method for bone scans was demonstrated by an observer study in which reading times and diagnostic accuracy improved significantly. An additional prospective clinical study verified that the temporal subtraction image could be used as a "second opinion" by radiologists with negligible detrimental effects. ANN was first used in 1990 for computerized differential diagnosis of interstitial lung diseases in CAD. Since then, ANN has been widely used in CAD schemes for the detection and diagnosis of various diseases in different imaging modalities, including the differential diagnosis of lung nodules and interstitial lung diseases in chest radiography, CT, and position emission tomography/CT. It is likely that CAD will be integrated into picture archiving and communication systems and will become a standard of care for diagnostic examinations in daily clinical work.     

Clinical efficacy of a new software developed for dental digital subtraction radiography

OBJECTIVES: The aim of this study was to test and compare the efficacy of software developed recently for digital subtraction radiography (DSR) in vivo. METHODS: An algorithm performing both manual and automated image reconstructions and contrast correction was developed for the manipulation of radiographic images. Pre- and post-operative radiographic images of ten patients were obtained and the automated subtraction analyses were performed using four different softwares (new software, Emago, Photoshop 8.0 and Paintshop Pro 9). Ten experienced dental specialists evaluated the clinical efficacy of each program and scored the softwares by using visual analogue scales (VAS). The results were statistically analysed and alpha was set as 0.05. RESULTS: The newly developed algorithm received higher scores than the others (new software =67.89, Emago = 64.26, Paintshop Pro 9 = 33.41 and Photoshop 8.0 = 27.24, respectively). The clinical efficacies of the new software and Emago were not significantly different (P = 0.720); likewise, Photoshop 8.0 and Paintshop Pro 9 performed comparably (P = 0.295). CONCLUSIONS: Considering this study, the new software and Emago would be suggested for DSR in dental practice.     

Detection of acute cerebral infarction by dual echo subtraction technique in MR imaging

The purpose of this study was to develop an image enhancement technique to detect acute cerebral infarct regions in brain MR images. Transverse relaxation times for abnormal changes tend to be longer than those for normal tissues. In order to obtain MR images with two different echo times, we employed the fast spin echo sequence. We then employed the image subtraction technique using two T(2)-weighted images to enhance acute cerebral infarct regions. As a result, the areas of acute cerebral infarct regions were enhanced as regions of higher signal than normal regions of brain tissue. Further, high signal areas in dual echo subtraction images corresponded to cerebral infarct regions of high signal areas in diffusion weighted images (DWI). We found that the image subtraction technique is useful to enhance very subtle regions of acute cerebral infarction in MR images. Because we employ the difference between transverse relaxation times for normal and abnormal tissues, which does not depend on the strength of the magnetic field, the dual echo subtraction method can be used in many hospitals.     

Improved detection of lung cancer arising in diffuse lung diseases on chest radiographs using temporal subtraction

RATIONALE AND OBJECTIVES: The purpose of this study was to evaluate the usefulness of temporal subtraction for the detection of lung cancer arising in pneumoconiosis, idiopathic pulmonary fibrosis, and pulmonary emphysema. MATERIALS AND METHODS: Fifteen cases of lung cancer arising in diffuse lung diseases, including three cases of pneumoconiosis, six of idiopathic pulmonary fibrosis, and six of pulmonary emphysema, were evaluated. Pathologic proof was obtained by surgery or transbronchial lung biopsy. The average interval between previous and current radiographs was 356 days (range, 31-947 days). All chest radiographs were obtained with a computed radiography system, and temporal subtraction images were produced by subtracting of a previous image from a current one with a nonlinear image-warping technique. The effect of the temporal subtraction image was evaluated by observer performance study with receiver operating characteristic analysis. RESULTS: The average observer performance with temporal subtraction was significantly improved (Az = 0.935) compared with that without temporal subtraction (Az = 0.857, P < .0001). CONCLUSION: The temporal subtraction technique is useful for the detection of lung cancer arising in pneumoconiosis, idiopathic pulmonary fibrosis, and pulmonary emphysema.     

Effect of temporal subtraction images on radiologists' detection of lung cancer on CT: results of the observer performance study with use of film computed tomography images

RATIONALE AND OBJECTIVES: To evaluate the effect of temporal subtraction images on the radiologists' detection of early primary lung cancer in computed tomography (CT) scans. MATERIALS AND METHODS: Fourteen cases with primary lung cancer and 16 normal cases were used for this study from a database of low-dose CT images, which were obtained from a lung cancer screening program in Nagano, Japan. Images were obtained with a single-detector helical CT scanner using 10 mm collimation and 2:1 pitch. Each case had both previous and current CT scans. Temporal subtraction images were obtained by subtracting the warped previous images from the current images. Seven radiologists, including four attendings and three residents, provided their confidence levels for the presence or absence of lung cancers with use of film CT images without and with temporal subtraction images. Receiver operating characteristic analysis was used to compare their performance without and with temporal subtraction images. RESULTS: The mean Az values (area under the receiver operating characteristic curve) of seven observers without and with temporal subtraction images were 0.868 and 0.930, respectively. Diagnostic accuracy was significantly improved by using temporal subtraction images (P = .007). Temporal subtraction images were especially useful when a nodule was present near the pulmonary hilum, where radiologists tended to overlook it. CONCLUSION: The temporal subtraction technique can significantly improve the sensitivity and specificity for detection of lung cancer on CT scans.     

Temporal subtraction of thin-section thoracic computed tomography based on a 3-dimensional nonlinear geometric warping technique

The authors developed a computer system that automatically produces temporal subtraction images on thin-section thoracic computed tomography based on a 3-dimensional nonlinear geometric warping technique. The automated subtraction system was applied to data from 20 subjects with lung diseases. Nineteen (95%) of 20 results showed good or excellent image quality. The remaining 1 (5%) showed adequate image quality. Our preliminary computer system demonstrates the potential for a clinically useful automated subtraction system.     

Computer-aided diagnosis in the detection of temporal changes in sequential chest radiographs

To aid radiologists in the diagnosis of screening chest radiographs, a temporal subtraction technique using digital image processing was developed. The accurate image registration of two sequential images enables us to detect even subtle changes in the "difference image" between them. In this report, a new method based on matching "lung markings" is introduced. Twenty-nine pairs of sequential posteroanterior chest radiographs with and without temporal changes were selected from cases examined with the computed radiography system. Image registration was employed, with the local matching of "lung markings" in previous and current radiographs. Observer performance tests were carried out by eight radiologists, with and without the "difference image." Observer performance tests with the temporal subtraction image showed that six of eight observers diagnosed them with higher sensitivity (mean, 43.9% vs. 55.3%) and a comparable false positive response. Mean area under the AFROC (alternative free-response receiver operating characteristics) curve also improved from 0.596 to 0.647, a statistically significant difference. The subtraction image using this registration technique improved diagnostic accuracy for subtle temporal changes.     

Usefulness of temporal subtraction images for identification of interval changes in successive whole-body bone scans: JAFROC analysis of radiologists' performance

RATIONALE AND OBJECTIVES: We evaluated the usefulness of temporal subtraction images obtained from two successive whole-body bone scans, in terms of improvement in radiologists' diagnostic accuracy in detecting interval changes and of a reduction in reading time, by use of a jackknife free-response receiver operating characteristic (JAFROC) analysis method. MATERIALS AND METHODS: Twenty pairs of successive whole-body bone scans (72 consented interval changes) and their temporal subtraction images were used for an observer performance study. Our institutional review board approved the use of this database and the participation of radiologists in this study. In the first session of the observer study, without temporal subtraction images, the previous and current images were shown to five radiologists independently for their marking of the locations on current images and confidence ratings on potential interval changes from previous images. In the second session, temporal subtraction images were shown together with the modified previous and current images. JAFROC analysis was used for assessing the statistical significance of differences between radiologists' performance without and with temporal subtraction images. RESULTS: The average sensitivity for detecting interval changes was improved from 58.6% to 73.2% at a false-positive rate of two per case by use of temporal subtraction images, and the difference was statistically significant by use of JAFROC analysis (P = .035). In addition, the mean reading time per case was reduced considerably from 134 seconds to 91 seconds (P < .01). CONCLUSIONS: Temporal subtraction imaging for successive whole-body bone scans has the potential greatly to assist radiologists by increasing both their accuracy and productivity.     

A simple method for generating temporal subtraction image in bone SPECT-initial evaluation in pelvic region-

We proposed and optimized a simple method of temporal subtraction image between successive bone single photon emission computed tomography (SPECT) images for supporting interpretation of temporal changes, and we evaluated its clinical utility. This method consisted of image registration, count normalization, and image subtraction. For image registration, we used a BEAT-Tl software. For count normalization, a pixel value of the normal accumulation part in a SPECT image was used as a reference region. We evaluated accuracy of image registration and optimized the normalization procedure. The accuracy of image registration ranged within 1 pixel in all directions (x, y, x-axis, and rotation). As the reference region, the second lumbar vertebra showed the best results in terms of the normalization procedure. Our method simply allowed the production of a temporal subtraction image. Because the software used in this method can be used free, this method would be available in every institution.     

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.     

Composite SISCOM images in mesial temporal lobe epilepsy: technique and illustration of regions of hyperperfusion

OBJECTIVE: To outline and validate a technique for subtraction ictal single photon emission tomography (SPET) co-registered to magnetic resonance imaging (MRI) (SISCOM), which allows non-linear co-registration of groups of subtraction SPET images to a single, template image. METHODS: In patients with mesial temporal lobe epilepsy, we used linear and non-linear transformation steps to co-register ictal and interictal images to a template SPET, then used the resultant ictal and interictal images to produce subtraction images. Statistical changes in subtraction SPET before and after the transformation steps in individual subjects were documented to validate the technique. Subtraction SPET images were thresholded to include pixel values 1 standard deviation (SD) above zero, and converted to binary. Images were combined by simple addition of images to form the final composite image. Final results were co-registered to a template MRI to demonstrate the regions most commonly hyperperfused in mesial temporal lobe epilepsy. RESULTS: Linear and non-linear transformation steps induced only small changes in standard deviations of subtraction SPET, typically approximately 1 pixel value. The final composite SISCOM images showed the anterior temporal lobe, insula and basal ganglia as the most commonly hyperperfused regions during mesial temporal lobe epileptic seizures. CONCLUSION: Our technique resulted in only small changes in statistical characteristics of individual subtraction SPET studies, and was acceptable for the purpose of creating composite SISCOM images.     

Perfusion Imaging of the Brain Using Z-Score and Dynamic Images Obtained by Subtracting Images from before and after Contrast Injection

Objective

The aim of this study was to examine the feasibility of perfusion imaging of the brain using the Z-score and subtraction dynamic images obtained from susceptibility contrast MR images.

Materials and Methods

Five patients, each with a normal MRI, Moya-moya, a middle cerebral artery occlusion, post-trauma syndrome, and a metastatic brain tumor, were selected for a presentation. A susceptibility-contrast echo-planar image after a routine MRI was taken as the source image with a rapid manual injection of 0.1 mmol/kg of Gd-DTPA. The inflow and washout patterns were observed from the time-signal intensity curve of the serial scans using the standard program of an MRI machine. The repeated Z-score images of the peak and late phases were made using the threshold Z-score values between 1.4 and 2.0 in four to five studies of the pre-contrast, peak, and late phases. Dynamic subtraction images were produced by subtracting sequential post-contrast images from a pre-contrast image and coloring these images using a pseudocolor mapping method.

Results

In the diseases with perfusion abnormalities, the Z-score images revealed information about the degree of perfusion during the peak and late phases. However, the quality varied with the Z-score threshold and the studies selected in a group. The dynamic subtraction images were of sufficient quality with no background noise and more clearly illustrated the temporal changes in perfusion and delayed perfusion.

Conclusion

The Z-scores and dynamic subtraction images illustrated the degree of perfusion and sequential changes in the pattern of perfusion, respectively. These images can be used as a new complimentary method for observing the perfusion patterns in brain diseases.     

Topography of the human corpus callosum using diffusion tensor tractography

OBJECTIVE: To evaluate the crossing fiber trajectory through the corpus callosum using distortion-corrected diffusion tensor tractography in the human brain. METHODS: After correcting distortion associated with large-diffusion gradients, T2-weighted echo planar images (EPIs) acquired from 10 right-handed healthy men were coregistered into T2-weighted fast spin echo images using linear through sixth-order nonlinear, 3-dimensional, polynomial warping functions. The optimal transformation parameters were also applied to the distortion-corrected diffusion-weighted EPIs. Diffusion tensor tractography through the corpus callosum was reconstructed, employing the "1 or 2 regions of interest" method. RESULTS: Compared with the lines through the genu, those through the rostrum ran more inferiorly and seemed to enter the orbital gyrus. Those lines entering posterior temporal white matter (tapetum) crossed through the ventral portion of the splenium and were clearly distinguished from lines that reached parieto-occipital white matter (forceps major). CONCLUSION: Diffusion tensor tractography is a feasible noninvasive tool to evaluate commissural fiber trajectory.     

Carotid bifurcation: results of spliced hybrid subtraction angiography compared with standard temporal and hybrid images

Hybrid subtraction angiography can provide a higher percentage of two-view examinations of the carotid bifurcation. Unfortunately, hybrid images are lower in signal-to-noise ratio (S/N) than temporal subtraction angiographic images and are often described as "noisy." The present study was designed to determine if the lower S/N of hybrid subtraction angiography could be restricted to the contralateral carotid bifurcation by "splicing" a small hybrid region of interest into a larger temporal image. Twenty-two selected images were examined on four image quality scales by seven readers. Similarity of reader scoring was characterized by cluster analysis and was good. Ipsilateral carotid arteries were preferred as temporal images, and contralateral carotid arteries were preferred as hybrid images. The splice image was preferred over both. The splice image retained the superiority of temporal subtraction angiography in the region of the ipsilateral carotid artery and incorporated the superiority of hybrid subtraction angiography in the region of the contralateral carotid artery. The splice image incorporated the best features of both image modalities.     

Usefulness of temporal subtraction images of chest computed radiography for detection of metastatic pulmonary nodules

Temporal subtraction is a technique by which a previous chest radiograph is subtracted from a current radiograph in order to enhance interval changes. Our purpose in this study was to evaluate the usefulness of temporal subtraction for the detection of metastatic pulmonary nodules. We examined 19 cases of metastatic nodules less than 15 mm in diameter (8.4 mm on average). Temporal subtraction images were created based on the matching of local lung areas in pairs of chest radiographs. By using the subtraction images, the detectability of nodules was clearly improved in 5 cases and moderately improved in 8 cases; there was no improvement in 6 cases. The subtraction images were especially useful for nodules superimposed over normal structures, such as rib, mediastinum, and diaphragm, as well as for small nodules. In 6 of the 19 cases, the nodules had been missed clinically; however, the temporal subtraction images clearly demonstrated half of missed small nodules. Temporal subtraction made it possible to enhance subtle interval changes and helped in detecting small pulmonary metastases. This technique seems promising for augmenting the capabilities of computed radiography of the chest.     

Improved detection of lung nodules by using a temporal subtraction technique

PURPOSE: To evaluate the effect of a temporal subtraction technique for digital chest radiography with regard to the accuracy of detection of lung nodules. MATERIALS AND METHODS: Twenty solitary lung nodules smaller than 30 mm in diameter, including 10 lung cancers and 10 benign nodules, were used. The nodules were grouped subjectively according to their subtlety. For non-nodular cases, 20 nodules without perceptible interval changes were selected. All chest radiographs were obtained by using a computed radiographic system, and temporal subtraction images were produced by using a program developed at the University of Chicago. The effect of the temporal subtraction image was evaluated by using an observer performance study, with use of receiver operating characteristic analysis. RESULTS: Observer performance with temporal subtraction images was substantially improved (A(z) = 0.980 and 0.958), as compared with that without temporal subtraction images (A(z) = 0.920 and 0.825) for the certified radiologists and radiology residents, respectively. The temporal subtraction technique clearly improved diagnostic accuracy for detecting lung nodules, especially subtle cases. CONCLUSION: The temporal subtraction technique is useful for improving detection accuracy for peripheral lung nodules on digital chest radiographs.     

Effect of temporal subtraction technique on the diagnosis of primary lung cancer with chest radiography

PURPOSE: The purpose of this study was to assess the diagnostic accuracy of the temporal subtraction technique in the detection of primary lung cancers by readers with different levels of experience. METHODS: Previous and current chest radiographs from 40 patients with histologically proven lung cancer and 40 controls were studied. Temporal subtraction images were produced using an automated digital subtraction technique. We evaluated the effect of temporal subtraction images in the diagnosis of lung cancer with chest radiographs via an observer performance study with the use of receiver operating characteristic analysis. Six experienced radiologists and six residents participated as observers. RESULTS: Observer performance for all observers was superior when temporal subtraction images were used (mean Az value increased from 0.764 to 0.836, p=0.0006). Although the average Az value for residents increased significantly, from 0.707 to 0.795 (p=0.0038), the average Az value for experienced radiologists increased only from 0.821 to 0.878 (n.s.). CONCLUSION: In conclusion, the temporal subtraction technique clearly improves diagnostic accuracy for the detection of primary lung cancer. The results indicated that the use of temporal subtraction images was more beneficial for the residents than for the experienced radiologists. This method would compensate to some extent for experience-dependent diagnostic accuracy in the detection of lung cancer.     

Parathyroid subtraction imaging--pitfalls in diagnosis

Since 1985 when parathyroid subtraction scintigraphy was presented in atlas form in this journal, the T1-201-Tc-99m computer subtraction studies have become more and more popular. Although the study has been quite accurate in localizing abnormal parathyroid pathology, the rate of false positive studies has been somewhat excessive, usually due to thyroid pathology. This atlas is an update of that previous work with the goal of presenting a scheme of interpretation that lessens the incidence of false positive studies as well as presenting many of the pitfalls in diagnosis of parathyroid disease.     

Image normalization and background subtraction in T1-201/Tc-99m parathyroid subtraction scintigraphy. Effect on lesion detection.

The authors have developed two computer algorithms for T1-201/Tc-99m parathyroid subtraction scintigraphy that was performed on patients who subsequently underwent surgical exploration of the neck. Both methods employed a region-of-interest drawn around the thyroid/parathyroid glands for image realignment. The first algorithm normalized the Tl-201 and Tc-99m images using the ratio of maximum counts over the thyroid in each image. The second computer algorithm incorporated Tl-201 image background correction and normalization by the average of the ratios of maximum counts computed over each quadrant in both images. In 10 patients with confirmed parathyroid adenomas or hyperplasia, the first method yielded a 44% sensitivity. Upon reanalysis with the second algorithm, the sensitivity improved to 100%. Subsequently, in a total of 22 patients with 30 abnormal glands analyzed with the second algorithm, a sensitivity of 80% (94% for adenoma and 62% for hyperplasia) was achieved, with a specificity of 91%, as confirmed by surgery.