Computer correction of projective distortions in dental radiographs.

Distortions in radiographic images caused by changes in shape and position of the film relative to the x-ray beam can result in registration artifacts which interfere with interpretation of subtracted images. A warping transformation is described which maps known points of reference in one image into homologous points in another. The corrected image pair can then be superimposed and subtracted with significantly-reduced differences in registration to facilitate the detection of localized changes of diagnostic interest.

Enhancement of tomosynthetic images in dental radiology.

In conventional projection radiography, anatomical structures superimposed on details of interest increase the difficulty of visual interpretation. Part of this problem can be resolved by tomographic reconstruction of specific "tissue" slices. Computer-aided tomosynthesis is one method which can reduce the superposition problem, but the reconstruction can benefit from further enhancement. In this investigation, we studied two such enhancement schemes: high-pass spatial-frequency filtration and iterative deconvolution. The performance of each technique was judged qualitatively in terms of the depiction of specific details of diagnostic interest, and quantitatively in terms of the degree of success afforded by each in rendering adjacent tissue slices independent. Both techniques demonstrated improved performance over the status quo.

A simple model combining quantum noise and anatomical variation in radiographs.

A model is presented for the detection of uniform lesions by means of an ideal photon detector. The difference of observed photons between an area overlaying the embedded lesion and an adjacent reference area of equal size constitutes the signal to be detected. Application of ROC analysis reveals that the exact probability distribution of this photon count difference can be approximated well by a Gaussian, on condition that modulation less than 0.1 and signal-to-noise ratio (SNR) greater than 1. Moreover, within these constraints, SNR emerged as the more salient parameter characterizing detection performance. It is shown that in the absence of anatomical variation, lesions of arbitrarily small size may be detected at any prescribed level of confidence, provided one is willing to accept the required high photon exposure. The effect of anatomical variation on detection performance is conveniently demonstrated in a graph of SNR versus exposure. There, two global regions are identified, each characterized by an asymptote, corresponding to either photon-limited or photon-saturated imaging. Under the first condition, quantum fluctuations are dominating the noise, and thus, detection performance is influenced by the photon exposure. Under the second, anatomical variations limit the SNR to an upper value, irrespective of exposure magnitude. Data obtained from dental radiographs demonstrate that anatomical variation is amenable to experimental measurement, and that it sets the upper limit for the SNR achievable in the diagnostic task of detecting incipient carious lesions.

An optimal synthetic aperture for circular tomosynthesis.

In circular tomosynthesis obtained from a sufficient number of radiographs with projection directions forming a single cone, object detail at a distance from the plane of interest is blurred according to the zero order Bessel function. The main lobe of this window function defines slice thickness, while its ringing side lobes are undesirable because they permit further outlying structures to "leak" through. Using the orthogonality of Bessel functions, a sampling scheme consisting of multiple, concentric projection cones was designed to synthesize by a finite Fourier-Bessel series a slice window with superior side lobe suppression. The window that concentrates the most "energy" within a finite slice, and can be realized by a limited number of sampling cones, is the zero order circular prolate spheroidal function. Its application to tomosynthesis of 3 mm thick slices with a characteristic detail size of 0.5 mm, concentrated 95% of the total series expansion to only 3 concentric circular scans, yielding a theoretical suppression of the first side lobe of -38 dB, to be compared to -8 dB achieved with the Bessel function. Experimental implementation of this optimal sampling scheme using 3, 8, and 14 projection views, distributed over cones with opening half-anges of 1.92, 4.39, and 6.88 degrees, respectively, resulted in a side-lobe attenuation of at least -20 dB. This attenuation reduced significantly artifacts arising from out-of-plane detail of high spatial frequency (e.g., edges) in tomosynthetic images of clinical interest in dentistry.

Effect of periodontal therapy on alveolar bone as measured by subtraction radiography.

Changes in the periodontal alveolar bone are often evaluated by comparing a series of radiographs taken over time. This investigation used a technique that allowed the image registration to be geometrically standardized each time a radiograph was taken. Radiographs of 24 patients from an ongoing double-blind, clinical study using metronidazole were obtained: (1) before any treatment, (2) at the completion of scaling and root planing and surgery (when performed) and (3) during the maintenance phase. One hundred six (106) paired comparisons were analyzed by subtraction radiography using a computerized system. Of these, 95 (89%) exhibited a minimal degree of geometric distortion and could be successfully substracted. Most areas (67%) showed no change in bone structure following periodontal treatment. Bone gain was noted in 12% of the sites examined, while bone loss was seen in 21% of the sites. This bone loss was statistically associated with sites that had received some form of surgical treatment.

Computerized tomosynthesis of dental tissues.

A digital method that produces an arbitrary number of tomographic slices from a finite number of dental radiographs is described. The relation between the slice thickness of the tomograms, the number of radiographs required, and the tomographic angle is determined on the basis of existing theory. In order to test the method, a series of radiographs of a radiographic phantom were taken and subsequently digitized to facilitate computer processing. From eight radiographs with an angular disparity of 4.5 degrees, tomograms with a slice thickness of about 3 mm, were produced. These not only indicate the relative position of the dental tissues in space but also show clinical relationships that are not visible in the original radiographs.

Fractal dimension from radiographs of peridental alveolar bone. A possible diagnostic indicator of osteoporosis.

The purpose of this study was to investigate whether a radiographic estimate of osseous fractal dimension is useful in the characterization of structural changes in alveolar bone. Ten dry mandibular bone segments were radiographed from three controlled projection angles (-5, 0, +5 degrees), before and after acid-induced partial decalcification. Fractal dimension was estimated by regression analysis of power spectra computed by Fourier transform of selected regions of interest in digitized images of the radiographs. Repeated-measures ANOVA showed that fractal dimension so determined varied over anatomic locations (p less than .01), but increased after acid-induced demineralization (p less than .0005), irrespective of the radiographic projection angles (p greater than .99). In vivo fractal dimension was computed from randomly selected intraoral radiographs of six premenopausal (ages, 32.8 +/- 3.9) and six postmenopausal (ages, 62.5 +/- 4.1) women. A significantly (p less than .01) higher fractal dimension was observed in the older group.

Dental radiography: efficacy in the assessment of intraosseous lesions of the face and jaws in asymptomatic patients.

In this investigation the efficacy of dental radiography for the detection of occult intraosseous lesions of the face and jaws was evaluated. An analysis of 30 million health insurance records indicated that the period prevalence of malignant lesions was less than 5 cases/million/year, and for benign lesions approximately 100 cases/million/year. Data from a controlled observer-performance study showed that radiographic sensitivities ranged between 50% and 80%. The cost per true-positive finding was estimated to be +8.6 million per malignant case and +430,000 per benign case. An assessment of the dosimetric literature indicated that the benefits of radiographic screening as a means for early detection of a malignancy appear to be counterbalanced by the risk of causing a radiation-induced malignancy. Taken together, these data demonstrate that dental radiography is not efficacious for the purpose of detecting occult lesions.

Calibration errors in digital subtraction radiography.

Two methods for the measurements of osseous change detected by digital subtraction radiography have been evaluated. Estimates of errors introduced by soft and hard tissues that overlay the region of interest (ROI) or by soft tissue that replaced the bone are made. All estimates are made assuming that the radiation source is a 40 keV monoenergetic beam and the reference standard is equivalent to compact bone. These assumptions facilitate a theoretical analysis of calibration errors on a relative scale. The radiographic image method uses a calibration wedge on each of the two films. The mean gray value of the ROI on each film is converted into an equivalent thickness of bone by matching the ROI mean gray value to a gray value along the wedge. These thickness values are then subtracted to obtain a measurement of the amount of change. The subtraction image method makes use of a wedge on only one of the two films. The image of the subtraction of the two radiographs is used for detection of the area of change and the mean gray value of the ROI is matched to a gray value along the wedge. The thickness of the wedge at this point is the measured change. The errors introduced by inclusion of the cheek over the ROI and the replacement of bone by soft tissue are estimated for the radiographic image method. The influence of unequal beam attenuation between the reference ramp and the ROI is estimated for the subtraction image method. Other factors influencing the accuracy of estimates of osseous change which are nonlinear in their effects are also considered.(ABSTRACT TRUNCATED AT 250 WORDS)

Assessment of bone remodeling in the temporomandibular joint by serial uptake measurement of technetium 99m-labeled methylene diphosphonate with a cadmium telluride probe.

Regional uptake of technetium 99m-labeled methylene diphosphonate was measured serially over time to follow the healing course of surgically generated lesions, induced unilaterally in the condyles of beagles. A small portable cadmium telluride probe was used for radiation detection, with and without a gold collimator attached. Radioactivity was measured at six periarticular sites on both the operated and the contralateral control sides and expressed as a ratio with respect to a reference site over the sagittal suture. The use of a collimator was crucial to separate the activity associated with bone mineralization in the condylar head from that arising in the adjoining bones. The uptake in the condylar region increased 2 weeks after surgery and remained at a constant level above the presurgery baseline (p less than 0.05) until termination of the study (10 weeks). Postmortem histologic examinations confirmed the high bone turnover level by revealing newly formed bone as well as the presence of osteoclasts. The time course of nuclide uptake in the temporomandibular joint is different from the general pattern of rise and fall of activity displayed in other bones and indicates the persistence of bone remodeling processes during a prolonged period of at least 10 weeks.

An experimental study of new diagnostic methods for the examination of osseous lesions in the temporomandibular joint.

Digital subtraction radiography, tomosynthesis, bone uptake of radionuclide, and arthroscopy were evaluated for detection and quantification of bony lesions induced unilaterally in the condyles of six dogs. A stereotaxic head-holder facilitated acquisition of reproducible radiographs suitable for subtraction and for circular tomosynthesis. Bone uptake of technetium-99m methylene diphosphonate was measured with a hand-held collimated miniature detector. Arthroscopy was performed with an arthroscope of 2.4 mm diameter. Bone defect mass determined by subtraction radiography correlated highly (r = 0.92, p less than 0.001) with the calcium content of removed bone measured by atomic spectroscopy. Both subtraction radiography and tomosynthesis indicated reshaping of the condyle into a more anterior position over a 10-week follow-up period. Radionuclide uptake was significantly elevated (p less than 0.04) from 2 to 10 weeks after surgery and correlated (r = 0.73, p less than 0.05) with regained bone mass measured by subtraction radiography. Arthroscopy revealed progressive degeneration of cartilage with denudation in the fossa. Both radiographic techniques demonstrated the lesions and bone remodeling, but only subtraction provided quantitative results. Radionuclide uptake predicted quantitatively future bone mass changes, and arthroscopy revealed cartilage and soft tissue status not otherwise observable.

Determination of projections for subtraction radiography based on image similarity measurements.

The reliability of subtraction radiography strongly depends on the ability to obtain two identical projections. Generally this is done by connecting mechanically the X-ray source, the patient and the film. However, this method has certain disadvantages, which restrict its use to small groups of patients in other than routine clinical situations. A computer-aided radiographic imaging technique (tomosynthesis) brings the reconstruction of arbitrary projections within reach. This technique can be used to reconstruct the projection required for a proper subtraction. However, in order to do so, the coordinates of the source position of one projection, relative to the source positions of the set of projections used for fomosynthesis, has to be determined. A method is described based on similarity measurements, in pairs of images (expressed in the form of standard deviations) to achieve this. The coordinates of the unknown source position could be determined with an average accuracy of 0.513 degrees (range 0.000-1.289 degrees), which is well within the range of deviations tolerable for the clinical application of subtraction radiography.

Studies on focal alveolar bone healing with technetium (Tc)-99m labeled methylene diphosphonate and gold-collimated cadmium telluride probe.

The benefit of using a collimator for a miniaturized cadmium telluride probe was evaluated by monitoring the bone-healing processes for 13 weeks after the induction of small iatrogenic alveolar bone lesions in one side of the mandible in beagles. Technetium (Tc)-99m labeled methylene diphosphonate (200 to 300 MBq, 5.1 to 8.1 mCi, in a solution of 0.5 to 1 ml, intravenously) was used as a bone-seeking radiopharmaceutical. The radioactivity over the bone lesion (L) and the contralateral normal site (C) in the mandible were measured between 1.5 and 2 hours after injection of the tracer, and the activity ratio L/C served as an index of relative bone uptake. A study of six dogs revealed that the healing response to a hemispheric bone defect of 2 mm diameter in the cortical bone could not be detected by an uncollimated probe, and in a repeated study in two dogs the use of a gold collimator (5 mm in diameter, 5 mm in length) did not increase the L/C ratio significantly. A second study in six dogs with 5 mm lesions showed that although systematic trends in the time courses of the L/C ratio obtained both with and without the collimator could be demonstrated, the L/C ratio of collimated versus uncollimated measurements was significantly (p less than 0.005) increased. In three of the latter six dogs, abscesses developed after 9 weeks, leading to a second increase (p less than 0.05) of the L/C ratio with collimation compared with the noninflammation group; without collimation no significant (p greater than 0.15) difference between the two groups could be demonstrated.(ABSTRACT TRUNCATED AT 250 WORDS)