摘要:

A novel wavelet‐based neural network (WNN) filter is proposed for image restoration as required for imaging of beta emitters by bremsstrahlung detection using a gamma camera. Quantitative imaging of beta emitters is important for theinvivomanagement of antibody therapy using either P‐32 or Y‐90. The theoretical basis for the general case forM‐channel multiresolution wavelet decomposition of the nuclear image into different subimages is developed with the objective of isolating the signal from noise. A modified Hopfield neural network (NN) architecture is then used for multichannel image restoration using the dominant signal subimages. The NN model avoids the common inverse problem associated with other image restoration filters such as the Wiener filter. The relative performance of the WNN for image restoration, forM=2 channel, is compared to a previously reported order statistic neural network hybrid (OSNNH) filter. Initially simulated degraded images of known structures with different noise levels are used. Quantitative metrics such as the normalized mean square error (NMSE) and signal‐to‐noise ratio (SNR) are used to compare filter performance. The WNN yields comparable results for image restoration with suggested slightly better performance for the images with higher noise levels as often encountered in bremsstrahlung detection. Attenuation measurements were performed using two radionuclides,32P and90Y as required for calibration of the gamma camera for quantitative measurements. Similar values for an effective attenuation coefficient were observed for the restored images using the OSNNH filters (32P: μ=0.122 cm−1,90Y: μ=0.135 cm−1) and WNN (32P: μ=0.122 cm−1,90Y: μ=0.135 cm−1) filters with slightly higher values obtained for the raw data (32P: μ=0.142 cm−1,90Y: μ=0.142 cm−1) for a 3.5‐cm source size. The WNN, however, was computationally more efficient by a factor of 4 to 6 compared to the OSNNH filter. The filter architecture, in turn, is also optimum for parallel processing or VLSI implementation as required for planar and particularly for SPECT mode of detection.

ISSN:0269-3127    

DOI:10.1118/1.597868

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

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×0.1 mm. Our recent study also showed that detection accuracy by computer decreases as the pixel size increases from 0.035 mm×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.

ISSN:0269-3127    

DOI:10.1118/1.597871

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

The goal of this study was to develop a technique to distinguish benign and malignant breast lesions in secondarily digitized mammograms. A set of 51 mammograms (two views/patient) containing lesions of known pathology were evaluated using six different morphological descriptors: circularity, μR/σR(where μR=mean radial distance of tumor boundary, σR=standard deviation); compactness,P2/A(whereP=perimeter length of tumor boundary andA=area of the tumor); normalized moment classifier; fractal dimension; and a tumor boundary roughness (TBR) measurement (the number of angles in the tumor boundary with more than one boundary point divided by the total number of angles in the boundary). The lesion was segmented from the surrounding background using an adaptive region growing technique. Ninety‐seven percent of the lesions were segmented using this approach. An ROC analysis was performed for each parameter and the results of this analysis were compared to each other and to those obtained from a subjective review by two board‐certified radiologists who specialize in mammography. The results of the analysis indicate that all six parameters are diagnostic for malignancy with areas under their ROC curves ranging from 0.759 to 0.928. We observed a trend towards increased specificity at low false‐negative rates (0.01 and 0.001) with the TBR measurement. Additionally, the diagnostic accuracy of a classification model based on this parameter was similar to that of the subjective reviewers.

ISSN:0269-3127    

DOI:10.1118/1.597707

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

While the qualitative effects of grid misalignment are known, we have quantified the effect of different degrees of grid misalignment on image contrast and patient exposure. Radiographs were made of a phantom consisting of five lead disks on top of a 15 cm block of lucite. Four 60 lines/cm grids, having grid ratios of 3:1, 4:1, 6:1, and 8:1 were used. When the tube was angled more than three degrees across the grid lines, the contrast improvement factor decreased substantially for all four grids, as much as 46% for an 8:1 grid with a 12° misalignment. There was a concomitant decrease in film optical density, which if compensated for by an increase in patient exposure, would lead to a higher effective bucky factor. With the exception of the 3:1 grid, if the grid is misaligned by more than 6°, higher signal‐to‐noise ratios can be attained by removing the grid and using the increased patient exposure to reduce noise.

ISSN:0269-3127    

DOI:10.1118/1.597767

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

To evaluate the potential clinical usefulness of a new screen‐film system (advanced screen‐film system; AD system) for chest radiography, its fundamental imaging properties compared with a conventional screen‐film system (HR‐4/HR‐S) were investigated. The basic imaging properties were evaluated by measuring characteristic (H&D) curves, relative speeds, MTFs (modulation transfer functions), WS (Wiener spectra), and x‐rayattenuationsof screens. The detail visibilities and pathological details of various diseases in chest radiographs of patients were evaluated subjectively. The film gradient of the AD system was slightly lower at low radiographic density, and higher at high density, as compared with a conventional screen‐film system.ThescreenspeedoftheADsystemwas212%greaterthanthatoftheconventionalsystem,andthefilmspeedwas53%thatoftheconventionalfilm.Astheresult,thetotalspeedoftheADsystemwasslightlyhighercomparedwiththeconventionalsystem. The spatial resolution of the AD system was comparable to or slightly lower than that of the conventional system. The noise level of the AD system was considerably lower than that of the conventional system at low (D=0.5) and middle (D=1.0) radiographic density levels. However, it was high at high radiographic density (D=1.8).TheradiographicdensitiesintheunderpenetratedareaswiththeADsystemweregreaterthanthoseoftheconventionalsystemwhenthelungdensitiesarematchedcomparable.ImprovementinnoiselevelwiththeADsystematlowandmiddledensitylevelsmaybeusefulfordetectionofvariousdiseasesinchestradiographs.

ISSN:0269-3127    

DOI:10.1118/1.597709

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

A method is described for automatically tracking spatial modulation of magnetizationtaglines on gated cardiac images. The method differs from previously reported methods in that it uses Fourier based spatial frequency and phase information to separately track horizontal and vertical tag lines. Use of global information from the frequency spectrum of an entire set of tag lines was hypothesized to result in a robust algorithm with decreased sensitivity to noise. The method was validated in several ways: first, actual tagged cardiac images at end diastole were deformed known amounts, and the algorithm's predictions compared to the known deformations. Second, tagged, gated images of the thigh muscle (assumed to have similar signal to noise characteristics as cardiac images, but to not deform with time) were created. Again the algorithmic predictions could be compared to the known (zero magnitude) deformations and to thigh images which had been artificially deformed. Finally, actual cardiac tagged images were acquired, and comparisons made between manual, visual, determinations of tag line locations, and those predicted by the algorithm. At 0.5 T, the mean bias of the method was<0.34 mm even at large deformations and at late (noisy) times. The standard deviation of the method, estimated from the tagged thigh images, was<0.7 mm even at late times. The method may be expected to have even lower error at higher field strengths.

ISSN:0269-3127    

DOI:10.1118/1.597819

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

In this article an automatic process is presented for the extraction of the left ventricular contours. The only operator‐supplied information is the location of a single point within the left ventricle. The raw data for processing are short‐axis cardiac MR images, obtained using various MRI protocols. In this study we considered only protocols for which the blood appears dark. The proposed algorithm performs smoothing and thresholding of the original MR images, and uses the segmented images to form edge images. From those edge images the endocardial curve is extracted using a modified contour following approach. The algorithm can accommodate moderate noise and small gaps in the contour. A smooth curve that approximates the epicardium is then obtained using a method combining radial search and FFT based smoothing. The quality of the obtained contours was visually assessed by three observers. Their evaluation shows that the algorithm is successful when images of reasonable quality are processed. Further evaluation work is required in order to fully assess the performance and limitation of the suggested method.

ISSN:0269-3127    

DOI:10.1118/1.597870

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

Post‐treatment dose verification refers to the process of reconstructing delivered dose distributions internal to a patient from information obtained during the treatment. The exit dose is commonly used to describe the dose beyond the exit surface of the patient from a megavoltage photon beam. Portal imaging provides a method of determining the dose in a plane distal to a patient from a megavoltage therapeutic beam. This exit dose enables reconstruction of the dose distribution from external beam radiation throughout the patient utilizing the convolution/superposition method and an extended phantom. An iterative convolution/superposition algorithm has been created to reconstruct dose distributions in patients from exit dose measurements during a radiotherapy treatment. The method is based on an extended phantom that includes the patient CT representation and an electronic portal imaging device (EPID). The convolution/superposition method computes the dose throughout the extended phantom, which allows the portal dose image to be predicted in the EPID. The process is then reversed to take the portal dose measurement and infer what the dose distribution must have been to produce the measured portal dose. The dose distribution is modeled without knowledge of the incident intensity distribution, and includes the effects of scatter in the computation. The iterative method begins by assuming that the primary energy fluence (PEF) at the portal image plane is equal to the portal dose image, the PEF is then back‐projected through the extended phantom and convolved with the dose deposition kernel to determine a new prediction of the portal dose image. The image of the ratio of the computed PEF to the computed portal dose is then multiplied by the measured portal dose image to produce a better representation of the PEF. Successive iterations of this process then converge to the exiting PEF image that would produce the measured portal dose image. Once convergence is established, the dose distribution is determined by back‐projecting the PEF and convolving with the dose deposition kernel. The method is accurate, provided the patient representation during treatment is known. The method was used on three phantoms with a photon energy of 6 MV to verify convergence and accuracy of the algorithm. The reconstructed dose volumes agree to within 3% of the forward computation dose volumes. Furthermore, this technique assumes no prior knowledge of the incident fluence and therefore may better represent the dose actually delivered.

ISSN:0269-3127    

DOI:10.1118/1.597872

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

Purpose:When a deviation in the treatment setup is identified, when or how should it be corrected? With the aim of improving the precision and reducing the systematic errors while maintaining the workload to a minimum, a study was performed to define the proper set of actions for the correction of tangential breast setup deviations during the course of the treatment.Patients and Methods:Clinical data were taken from a prospective study of more than 2200 on‐line Electronic Portal Images (EPI) from 20 patients treated with tangential irradiation following a partial mastectomy for an early stage (I and II) breast cancer. The values of the central lung distance were then entered in the verification/correction procedure and modified to correct only the portion of the deviation most likely attributed to systematic errors according to the maximum likelihood. The verification/correction procedure uses an action level equal to FML⋅α/√N, where α is proportional to the standard deviation (α=nσ) andNis the number of consecutive fractions delivered after the start of the treatment. FML is the fraction of the deviation due to the systematic errors estimated from the maximum likelihood of the two distributions. In addition ton, one needs to specifyNmax, the maximum number of consecutive measurements without correction, to apply the procedure. The combination of these two parametersC(n,Nmax) will determine the reduction of systematic errors (gain) and the number of measurements and corrections (workload) associated to the procedure.Results:The effects of three combinations,C(1,1),C(2,2), andC(3,4) were studied. Also, the analysis of the results after application of the procedure with and without the factor of maximum likelihood made individually for each patient demonstrates the importance of the FML.Conclusions:The verification/correction procedure with the inclusion of the FML can effectively improve the accuracy when applied to clinical data. With the specific workload related to measurements and corrections performed at our institution, we have found that in the case of the tangential breast treatment, the optimum combination uses an action level equal to 2σ and a maximum of two consecutive measurements without correction.

ISSN:0269-3127    

DOI:10.1118/1.597722

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

A comparison was made of different treatment plans to determine the effect on the three‐dimensional dose distributions of varying the allowed parameters in linac‐based stereotactic radiosurgery with circular collimators; these parameters are arc position, length, and weighting, and collimator diameter. For the class of eccentrically shaped target volumes that are not so irregular as to require several separate isocenters, it was found that superior dose distributions could be achieved by varying arc length, arc position, arc weighting, and collimator diameter. An analysis of the results achieved with an automated planning program indicates that, in general, the variables of arc position and arc length are of greater importance than collimator size or beam weighting. However, there are cases where varying these latter two parameters does result in markedly better dose distributions. A deeper investigation into the effects of multiple collimators on the dose distribution in the area of steepest gradient demonstrated that multiple collimator sizes do not significantly degrade the dose falloff, which is in fact mostly determined by the effects of intersecting arcs.

ISSN:0269-3127    

DOI:10.1118/1.597723

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY