Image processing algorithms for digital mammography: a pictorial essay.

Digital mammography systems allow manipulation of fine differences in image contrast by means of image processing algorithms. Different display algorithms have advantages and disadvantages for the specific tasks required in breast imaging-diagnosis and screening. Manual intensity windowing can produce digital mammograms very similar to standard screen-film mammograms but is limited by its operator dependence. Histogram-based intensity windowing improves the conspicuity of the lesion edge, but there is loss of detail outside the dense parts of the image. Mixture-model intensity windowing enhances the visibility of lesion borders against the fatty background, but the mixed parenchymal densities abutting the lesion may be lost. Contrast-limited adaptive histogram equalization can also provide subtle edge information but might degrade performance in the screening setting by enhancing the visibility of nuisance information. Unsharp masking enhances the sharpness of the borders of mass lesions, but this algorithm may make even an indistinct mass appear more circumscribed. Peripheral equalization displays lesion details well and preserves the peripheral information in the surrounding breast, but there may be flattening of image contrast in the nonperipheral portions of the image. Trex processing allows visualization of both lesion detail and breast edge information but reduces image contrast.

Digital breast imaging: tomosynthesis and digital subtraction mammography.

Advances in the computer technology and the introduction of new digital imaging detectors offer the potential for digital image acquisition and several new mammography techniques, such as tomosynthesis and digital subtraction mammography. Tomosynthesis is a method of obtaining tomographic images of a breast. In tomosynthesis, any number of tomographic planes may be reconstructed from a set of images obtained as the X-ray source is moved in an arc above the breast. By shifting and adding the information obtained at different source positions, any plane of the breast can be brought into a sharp focus, while structures outside this selected plane are blurred. This may lead to improved lesion detection, especially in dense breast tissue. Thus, tomosynthesis may play a role in improving breast cancer screening and lesion characterization. Digital subtraction mammography is a method of breast angiography. It is performed by obtaining a digital radiographic image before, and one or more digital radiographic images after the injection of a contrast agent such as iodine. The pre- and post-contrast images are subtracted, resulting in an image of the vascular structures in the breast. Because breast cancer lesions have increased vascularity, digital subtraction mammography may play an important role in improving lesion detection, characterizing lesions, monitoring response to therapy, and determining lesion extent.Thus, both of these new digital techniques have the potential to address the major limitation of conventional mammography, namely the difficulty in detecting cancer in radiographically dense breasts.

Digital tomosynthesis in breast imaging.

PURPOSE: To describe and evaluate a method of tomosynthesis breast imaging with a full-field digital mammographic system. MATERIALS AND METHODS: In this tomosynthesis method, low-radiation-dose images were acquired as the x-ray source was moved in an arc above the stationary breast and digital detector. A step-and-expose method of imaging was used. Breast tomosynthesis and conventional images of two imaging phantoms and four mastectomy specimens were obtained. Three experienced readers scored the relative lesion visibility, lesion margin visibility, and confidence in the classification of six lesions. RESULTS: Tomosynthesis image-reconstruction algorithms allow tomographic imaging of the entire breast from a single arc of the x-ray source and at a radiation dose comparable with that in single-view mammography. Except for images of a large mass in a fatty breast, the tomosynthesis images were superior to the conventional images. CONCLUSION: Digital mammographic systems make breast tomosynthesis possible. Tomosynthesis may improve the specificity of mammography with improved lesion margin visibility and may improve early breast cancer detection, especially in women with radiographically dense breasts.

Image compression in digital mammography: effects on computerized detection of subtle microcalcifications.

Our previous receiver operating characteristic (ROC) study indicated that the detection accuracy of microcalcifications by radiologists is significantly reduced if mammograms are digitized at 0.1 mm x 0.1 mm. Our recent study also showed that detection accuracy by computer decreases as the pixel size increases from 0.035 mm x 0.035 mm. It is evident that very large matrix sizes have to be used for digitizing mammograms in order to preserve the information in the image. Efficient compression techniques will be needed to facilitate communication and archiving of digital mammograms. In this study, we evaluated two compression techniques: full frame discrete cosine transform (DCT) with entropy coding and Laplacian pyramid hierarchical coding (LPHC). The dependence of their efficiency on the compression parameters was investigated. The techniques were compared in terms of the trade-off between the bit rate and the detection accuracy of subtle microcalcifications by an automated detection algorithm. The mean-square errors in the reconstructed images were determined and the visual quality of the error images was examined. It was found that with the LPHC method, the highest compression ratio achieved without a significant degradation in the detectability was 3.6:1. The full frame DCT method with entropy coding provided a higher compression efficiency of 9.6:1 at comparable detection accuracy. The mean-square errors did not correlate with the detection accuracy of the microcalcifications. This study demonstrated the importance of determining the quality of the decompressed images by the specific requirements of the task for which the decompressed images are to be used. Further investigation is needed for selection of optimal compression technique for digital mammograms.

Enzyme replacement therapy for Gaucher disease: skeletal responses to macrophage-targeted glucocerebrosidase.

OBJECTIVES: Reversal of the hematologic and visceral abnormalities characteristic of Gaucher disease, the most common lipid storage disorder, with biweekly infusions of macrophage-targeted glucocerebrosidase (glucosylceramidase) is well documented. The extent to which the skeleton responds to enzyme replacement therapy has not been systematically investigated. METHODS: To assess the skeletal response to enzyme replacement therapy, we treated 12 patients with type 1 Gaucher disease, who had intact spleens, with macrophage-targeted glucocerebrosidase. The initial dose of enzyme was 60 U/kg body weight every 2 weeks for 24 months, followed by reduction in dosage to 30 and then 15 U/kg body weight every 2 weeks, each for 9 months. RESULTS: The lipid composition of bone marrow, determined by direct chemical analysis, began to improve after 6 months of treatment at a time when noninvasive imaging studies showed no significant changes. By 42 months, improvement in marrow composition was demonstrable on all noninvasive, quantitative imaging modalities (magnetic resonance score, quantitative xenon scintigraphy, and quantitative chemical shift imaging) used in this study. Quantitative chemical shift imaging, the most sensitive technique, demonstrated a dramatic normalization of the marrow fat content in all patients. Net increases in either cortical or trabecular bone mass, as assessed by combined cortical thickness measurements and dual-energy quantitative computed tomography, respectively, occurred in 10 patients. CONCLUSIONS: Prolonged treatment over 3 1/2 years with macrophage-targeted glucocerebrosidase produces objective reversal of disease in both the axial and appendicular skeleton in patients with Gaucher disease. Marked improvement occurs in marrow composition and bone mass in both children and adults.

The estimation of occupational effective dose in diagnostic radiology with two dosimeters.

Annual effective dose limits have been proposed by national and international radiation protection committees. Radiation protection agencies must decide upon a method of converting the radiation dose measured from dosimeters to an estimate of effective dose. A proposed method for the estimation of effective dose from the radiation dose to two dosimeters is presented. Correction factors are applied to an over-apron collar dose and an under-apron dose to estimate the effective dose. Correction factors are suggested for two cases, both with and without a thyroid shield. Effective dose may be estimated by the under-apron dose plus 6% of the over-collar dose if a thyroid shield is not worn or plus 2% of the over-collar dose if a thyroid shield is worn. This method provides a reasonable estimate of effective dose that is independent of lead apron thickness and accounts for the use of a thyroid shield.

Automated segmentation of regions of interest on hand radiographs.

Most radiologists do not use texture information contained in the trabecular patterns of hand radiographs to diagnose erosive changes and demineralization due to systemic inflammatory diseases that affect the skeletal system. However, high-resolution digitization achievable by a laser digitizer now makes it possible to access texture information that may not be perceived visually. We are studying the feasibility of computer-assisted early detection of these processes with particular attention to patients with hyperparathyroidism. In this paper the methods used to extract a region of interest (ROI) for texture analysis are discussed. The techniques include multiresolution sensing, automatic adaptive thresholding, detection of orientation angle, and projection taken perpendicular to the line of least second moment. The methods were tested on a database of 50 pairs of hand radiographs. We segmented the middle and the index fingers with an average success rate of 83% per hand. For the segmented finger strips, we located ROIs on both the middle and the proximal phalanges correctly over 84% of the times. Texture information was collected in the form of a concurrence matrix within the ROI. This study is a prelude to evaluating the correlation between classification based on texture analysis and diagnosis made by experienced radiologists.

Evaluation of a parallel hole grid for bedside chest imaging.

RATIONALE AND OBJECTIVES: The application of a new parallel hole grid designed for bedside radiography is investigated. This grid is constructed of sheets of holes and is designed to have more tolerance to positioning errors than conventional grids. METHODS: The parallel hole grid is compared to conventional 6:1 and 12:1 grids using phantoms. The contrast improvement, scatter fractions, exposure, and tolerance to positioning errors are measured. RESULTS: The parallel hole grid has much more tolerance to positioning errors than conventional grids. The contrast improvement and scatter rejection are significantly less than those obtained with a conventional 6:1 ratio grid. Compared with nongrid techniques, the parallel hole grid provides 13% to 20% higher lung contrast. CONCLUSIONS: The parallel hole grid may be used for bedside imaging without the need for accurate alignment. The parallel hole grid requires approximately 2.7 times the entrance exposure of a nongrid technique. Lung contrast improvement is approximately half of that from an accurately aligned 6:1 conventional grid. For grid angulation greater than 8 degrees, the parallel hole grid provides higher contrast than the 6:1 grid.

Digitization requirements in mammography: effects on computer-aided detection of microcalcifications.

We have developed a computerized method for detection of microcalcifications on digitized mammograms. The program has achieved an accuracy that can detect subtle microcalcifications which may potentially be missed by radiologists. In this study, we evaluated the dependence of the detection accuracy on the pixel size and pixel depth of the digitized mammograms. The mammograms were digitized with a laser film scanner at a pixel size of 0.035 mm x0.035 mm and 12-bit gray levels. Digitization with larger pixel sizes or fewer number of bits was simulated by averaging adjacent pixels or by eliminating the least significant bits, respectively. The SNR enhancement filter and the signal-extraction criteria in the computer program were adjusted to maximize the accuracy of signal detection for each pixel size. The overall detection accuracy was compared using the free response receiver operating characteristic curves. The results indicate that the detection accuracy decreases significantly as the pixel size increases from 0.035 mm x 0.035 mm to 0.07 mm x 0.07 mm (P < 0.007) and from 0.07 mm x 0.07 mm to 0.105 mm x 0.105 mm (P < 0.002). The detection accuracy is essentially independent of pixel depth from 12 to 9 bits and decreases significantly (P < 0.003) from 9 to 8 bits; a rapid decrease is observed as the pixel depth decreases further from 8 to 7 bits (P < 0.03) or from 7 to 6 bits (P < 0.02).(ABSTRACT TRUNCATED AT 250 WORDS)

Radiation dose reduction in CT of the pediatric pelvis.

PURPOSE: To determine whether a lower radiation dose technique can be used for computed tomography (CT) of the pediatric pelvis without significant loss of diagnostic image quality. MATERIALS AND METHODS: Thirty-six pediatric patients underwent CT at 80 mAs (experimental group), and 42 underwent CT at 240 mAs (control group). Anatomic details, image quality, and the degree of confidence in reaching a diagnosis were graded on a scale of 1 (poor) to 4 (excellent). RESULTS: The difference in perceived image quality between the experimental and control groups was not statistically significant. The mean scores for scans evaluated by the first reader were 3.88 for the experimental group and 3.92 for the control group (P = .2804). The mean scores for the second reader were 3.78 and 3.77 for the experimental and control groups, respectively (P = .8131). CONCLUSION: A substantial dose reduction can be achieved if pelvic CT is performed at 80 mAs, without a recognizable deterioration of diagnostic image quality.

Computer-aided diagnosis: detection and characterization of hyperparathyroidism in digital hand radiographs.

An automated method is developed for the detection and staging of skeletal changes due to hyperparathyroidism on digitized hand radiographs. Subperiosteal bony resorption, particularly along the radial margins of the middle and proximal phalanges, is among the earliest manifestations of secondary hyperparathyroidism. In order to quantify the severity of bone resorption in these regions, the computer method analyzes the roughness of the phalangeal margins, as projected on the radiograph. The regions of interest, which contain the phalanges, are obtained from the digitized hand radiographs by an image preprocessor. The radial margin of each phalanx is detected by a model-guided boundary-tracking scheme. The roughness of these boundaries is then quantified by the mean-square variation and the first moment of the power spectrum. A receiver operating characteristic (ROC) study comparing the computer detection of hyperparathyroidism with the diagnosis by three experienced skeletal radiologists was performed by evaluating 84 hand images from 22 patients. Our present computer method can achieve a true-positive rate of 94% and a true-negative rate of 92%. Such a computer-aided diagnosis system may assist radiologists in their assessment of primary and secondary hyperparathyroidism, since it is both accurate and not subject to either intra- or interobserver variations.

Interventional radiologists: occupational radiation doses and risks.

Interventional radiologists receive nonuniform occupational radiation doses, with relatively high doses to the head and extremities and low doses to the trunk, which is protected by a lead apron. Twenty-eight interventional radiologists from 17 institutions wore thermoluminescent dosimeters over their collars and under their aprons for a 2-month period. The estimated annual radiation dose was converted to effective dose as suggested by the International Commission on Radiological Protection. Effective dose is used to relate the risk associated with nonuniform dose to that associated with an equivalent uniform whole-body dose. The mean annual effective dose was 3.16 mSv (316 mrem), with a range of 0.37-10.1 mSv. The mean annual effective dose is approximately equal to the mean natural background dose of 3 mSv per year from radon and other natural sources and is only 6% of the National Council on Radiation Protection and Measurements' recommended effective dose equivalent limit of 50 mSv per year. The annual radiation risk of fatal cancer would be less than one per 10,000 for almost the entire career of an interventional radiologist.

Portable chest imaging: comparison of storage phosphor digital, asymmetric screen-film, and conventional screen-film systems.

The quality of chest images obtained with portable radiography was evaluated for a conventional screen-film system, a new asymmetric screen-film chest system, and computed radiography (CR). Sixty chest images were obtained in 20 patients in an intensive care unit. The CR system was ranked by all three evaluating radiologists as substantially better in overall diagnostic quality, interpretability of the lungs, and musculoskeletal detail and by two of the three observers as better for the visibility of catheters and lines. In the upper abdomen and mediastinum, there was not a clear preference. Standard deviations of film density were +/- 0.12, +/- 0.41, and +/- 0.39 for the CR, conventional, and asymmetric systems, respectively. For the same systems, phantom results indicated the relative lung contrast values were 1.2, 1.0, and 0.89, respectively. Similarly, the limiting resolution values in the lung were 2.0, 4.2, and 6.3 line pairs per millimeter. The CR system had twice the root-mean-square noise of the screen-film systems. Overall, the preferred system for portable chest imaging was the CR system.

A simple method for use in QC of CT for radiation therapy treatment planning.

A simple procedure for monitoring constancy of spatial measurement and CT number determination from CT images used in radiation therapy treatment planning is described. The procedure uses low-Z material rods glued to the underside of the CT table insert and does not require a special phantom. Measurements are made on the same patient images used for treatment planning. Deviations from predetermined baseline values outside quality control limits of +/- 2 mm in spatial resolution and +/- 20 CT numbers in density can be detected with a confidence level of 97% or better.

CT densitometry of pulmonary nodules in a frozen human thorax.

The influence of (1) calcium concentration, (2) exposure technique, (3) reconstruction algorithm, (4) nodule size, and (5) nodule location on the CT attenuation values (CT density) of pulmonary nodules was examined in a frozen human thorax. Nodules with calcium concentrations of 0-310 mg/ml and diameters of either 0.95 or 1.59 cm were inserted into a frozen, unembalmed human thorax. The nodules were placed either at the lung apex or 4 cm below the tracheal carina. Each nodule was scanned on a GE CT 9800 scanner; five different exposure techniques were used. The slice thickness was uniformly 1.5 mm. As expected, increasing the kilovoltage caused a significant decrease in CT nodule density in all nodules with calcium concentrations greater than 80 mg/ml. The inverse relationship between kilovoltage and nodule density was exaggerated with increasing calcium concentration. A high-resolution (bone) algorithm gave a significantly higher CT number than did a smoothed (standard) algorithm, regardless of nodule size and location, but this difference could be attributed almost entirely to the edge-enhancement effect of the bone algorithm. The CT density of the larger nodules was significantly higher than that of the smaller nodules at calcium concentrations greater than 65 mg/ml for both standard and bone algorithms. Densities were significantly higher in the mid lung than in the apex with a standard algorithm, but this was not the case with a bone algorithm. The GE CT 9800 scanner had a linear response between CT density and increasing calcium concentration within the confines of a human thorax. A high-resolution (bone) reconstruction algorithm has higher spatial resolution but does show an edge-enhancing effect not found with the smoothed algorithm. Two major variables in CT densitometry for pulmonary nodules are the kilo electron voltage of the X-ray beam and the reconstruction algorithm used; these two parameters should be standardized, with a high kilovoltage and high-resolution algorithm favored on the GE CT 9800 scanner.

Dual-energy digital radiographic quantification of calcium in simulated pulmonary nodules.

The presence or absence of calcium in solitary pulmonary nodules may indicate whether a nodule is benign or malignant. Because current techniques for measuring the amount of calcium in these nodules are unsatisfactory, a study was carried out to assess the capability of dual-energy digital chest radiography to identify and quantify the calcium content of simulated pulmonary nodules of known calcium content. Measurements were carried out on 280 nodules of various sizes and calcium content that were placed within the lungs of a frozen human-chest phantom. A new calcium quantification technique that uses a parallelogram was developed to eliminate the problem of nodule superimposition over ribs. Nodules containing more than 35 mg of calcium per square centimeter (i.e., 7, 30, 60, and 110 mg of calcium for spherical nodules 0.5, 1.0, 1.5, and 2.0 cm in diameter) were measured with a high degree of accuracy and reasonable precision. Dual-energy digital radiography is a simple and accurate method of measuring the calcium content of solitary pulmonary nodules in humans.

Simulated pulmonary nodules: detection with dual-energy digital versus conventional radiography.

Performance of a prototype dual-energy digital chest radiography unit in detecting calcified and noncalcified simulated pulmonary nodules was compared with that of a highly optimized, conventional system. Nodules ranging in size (0.5, 1.0, and 1.6 cm), in number (five to 11), and in calcium content (0-25 mg) were superimposed over the lungs of a frozen, unembalmed, human chest phantom. For each technique, six observers examined 50 posteroanterior projections with different randomized nodule locations. Detection consisted of locating and assigning a level of confidence to each perceived nodular opacity. The resulting plots of the true-positive fraction versus the mean number of false-positive calls per projection indicate that for both calcified and noncalcified nodules, the digital unit performed significantly better (P less than .01).

Calcification in pulmonary nodules: detection with dual-energy digital radiography.

The authors undertook a clinical study to determine the accuracy of dual-energy digital radiography in revealing nodule calcification because calcification in a pulmonary nodule almost excludes the possibility of malignancy. Over a 6-month period, 61 patients with pulmonary nodules (less than or equal to 3 cm) or masses (greater than 3 cm) were examined on a prototype scanned projection unit using a dual-energy detector. In 49 of 61 patients, nodules were noncalcified, and in 12, they were calcified. In 57 patients, the benignancy or malignancy of nodules was established beyond reasonable doubt by pathologic confirmation in 38 and by strong inference in 19 (four patients with noncalcified solitary pulmonary nodules either refused further investigation or surgery or their follow-up was too short to permit exclusion of malignancy). Dual-energy radiography was found to be highly accurate in assessing the presence or absence of calcification in pulmonary nodules and thus in determining their benignancy or possible malignancy.