摘要:

We are developing a computer program for automated detection of clustered microcalcifications on mammograms. In this study, we investigated the effectiveness of a signal classifier based on a convolution neural network (CNN) approach for improvement of the accuracy of the detection program. Fifty‐two mammograms with clustered microcalcifications were selected from patient files. The clusters on the mammograms were ranked by experienced mammographers and divided into an obvious group, an average group, and a subtle group. The average and subtle groups were combined and randomly divided into two sets, each of which was used as training or test set alternately. The obvious group served as an additional independent test set. Regions of interest (ROIs) containing potential individual microcalcifications were first located on each mammogram by the automated detection program. The ROIs from one set of the mammograms were used to train CNNs of different configurations with a back‐propagation method. The generalization capability of the trained CNNs was then examined by their accuracy of classifying the ROIs from the other set and from the obvious group. The classification accuracy of the CNNs for the ROIs was evaluated by receiver operating characteristic (ROC) analysis. It was found that CNNs of many different configurations can reach approximately the same performance level, with the area under the ROC curve (Az) of 0.9. We incorporated a trained CNN into the detection program and evaluated the improvement of the detection accuracy by the CNN using free response ROC analysis. Our results indicated that, over a wide range of true‐positive (TP) cluster detection rate, the CNN classifier could reduce the number of false‐positive (FP) clusters per image by more than 70%. For the obvious cases, at a TP rate of 100%, the FP rate reduced from 0.35 cluster per image to 0.1 cluster per image. For the average and subtle cases, the detection accuracy improved from a TP rate of 87% at an FP rate of four clusters per image to a TP rate of 90% at an FP rate of 1.5 clusters per image.

ISSN:0269-3127    

DOI:10.1118/1.597428

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

Spiculation is a primary sign of malignancy for masses detected by mammography. In this study, we developed a technique that analyzes patterns and quantifies the degree of spiculation present. Our current approach involves (1) automatic lesion extraction using region growing and (2) feature extraction using radial edge‐gradient analysis. Two spiculation measures are obtained from an analysis of radial edge gradients. These measures are evaluated in four different neighborhoods about the extracted mammographic mass. The performance of each of the two measures of spiculation was tested on a database of 95 mammographic masses using ROC analysis that evaluates their individual ability to determine the likelihood of malignancy of a mass. The dependence of the performance of these measures on the choice of neighborhood was analyzed. We have found that it is only necessary to accurately extract an approximate outline of a mass lesion for the purposes of this analysis since the choice of a neighborhood that accommodates the thin spicules at the margin allows for the assessment of margin spiculation with the radial edge‐gradient analysis technique. The two measures performed at their highest level when the surrounding periphery of the extracted region is used for feature extraction, yieldingAzvalues of 0.83 and 0.85, respectively, for the determination of malignancy. These are similar to that achieved when a radiologist's ratings of spiculation (Az=0.85) are used alone. The maximum value of one of the two spiculation measures (FWHM) from the four neighborhoods yielded anAzof 0.88 in the classification of mammographic mass lesions.

ISSN:0269-3127    

DOI:10.1118/1.597626

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

Measurements were made of the MTF(f), NPS(f), and DQE(f) of four generations of computed radiography (CR) imaging plates and three generations of CR image readers. The MTF generally showed only a minor change between generations of plates and readers, but the DQE(f) has improved substantially from a very early plate/reader combination to a more recent one. The DQE in the more recent plate/reader combination is 1.3× greater at low frequencies and about 3× greater at high frequencies than the much earlier versions. Thus there has been substantial improvement in the imaging performance obtainable with CR since some of the early observer studies which indicated poorer performance with CR than with screen–film.

ISSN:0269-3127    

DOI:10.1118/1.597627

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

We investigate a concept for making a large area, flat‐panel detector for digital radiology. It employs an x‐ray sensitive photoconductor to convert incident x‐radiation to a charge image which is then electronically read out with a large area integrated circuit. The large area integrated circuit, also called an active matrix, consists of a two‐dimensional array of thin film transistors (TFTs). The potential advantages of the flat‐panel detector for digital radiography include: instantaneous digital radiographs without operator intervention; compact size approaching that of a screen‐film cassette and thus compatibility with existing x‐ray equipment; high quantum efficiency combined with high resolution. Its potential advantages over the x‐ray image intensifier (XRII)/video systems for fluoroscopy include: compactness; geometric accuracy; high resolution, and absence of veiling glare. The feasibility of the detector for digital radiology was investigated using the properties of a particular photoconductor (amorphous selenium) and active matrix array (with cadmium selenide TFTs). The results showed that it can potentially satisfy the detector design requirements for radiography (e.g., chest radiography and mammography). For fluoroscopy, the images can be obtained in real‐time but the detector is not quantum noise limited below the mean exposure rate typically used in fluoroscopy. Possible improvements in x‐ray sensitivity and noise performance for the application in fluoroscopy are discussed.

ISSN:0269-3127    

DOI:10.1118/1.597628

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

The theoretical underpinnings of photoacoustic ultrasound (PAUS) reconstruction tomography are presented. A formal relationship between PAUS signals and the heterogeneous distribution of optical absorption within the object being investigated is developed. Based on this theory, a reconstruction approach, analogous to that used in x‐ray computed tomography, is suggested. Initial experimental results suggest that this approach produces “reasonable” reconstructions for absorbers distributed within a narrow plane embedded within a highly scattering medium.

ISSN:0269-3127    

DOI:10.1118/1.597429

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

A method based on group theory is proposed to quantify the shape of an organ. It is proposed that a complex shape is characterized by a basic set of symmetry operations involving rotation about the center axis. Based on this concept, partial symmetry elements (PSE) were determined by stepwise rotation of the images and measurement of overlap area between the original and rotated images. These measurements were used to define a symmetry index (SI) and to “decompose” the complex shape into simpler shapes with regular geometry such as a circle, an ellipse, an equilateral triangle, a square, etc. Computer simulations and two‐dimensional (2‐D) ultrasonic images of the prostate were used to demonstrate the feasibility of the method. The method was also applied to a series of clinical ultrasonic cardiac images to determine the dynamic shape changes in left ventricle (LV) of a normal heart and an heart with infarction. The results of this study show that SI increased with an increase in the symmetry of the simulation models and the prostate. Cross‐sectional images of prostate with triangular and ellipsoidal shapes exhibited larger proportion of three‐fold and two‐fold symmetries, respectively. Symmetry index of the LV changed during the cardiac cycle and exhibited a loop‐like structure when plotted as a function of size. Quantitative measurements of SI indicated that the left ventricular chamber is rounder during the diastolic phase of the cardiac cycle. The loop‐like structure was considerably reduced in the heart with infarction. Also, larger distortions in cardiac shape were observed over cardiac cycle in the heart with infarction, as compared to the normal heart. In conclusion, it is feasible to quantitatively characterize 2‐D shapes of the organs by symmetry operations based on group theoretical analysis.

ISSN:0269-3127    

DOI:10.1118/1.597430

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

The question of how to include radioactive decay during biological modeling with first‐order differential equations was considered. Modeling may involve either experimental datay(t) or decay‐corrected dataz(t) [≡exp(λt)y(t) where λ is the decay constant] for each compartment. It is sometimes assumed that the latter are solutions to corresponding purely pharmacokinetic models (no decay). We primarily compared the two analyses in the case where the model did not require simultaneous consideration of both labeled and unlabeled material. A general theorem was found which limits the use of decay‐corrected data to pharmacokinetic models containing linear, homogeneous differential equations. By way of verification, an example of this model type was analyzed for a chimeric monoclonal antibody biodistribution in man. Even in this case, statistically significant differences between the two solutions showed that one may find different model parameters depending upon which data set (yorz) was analyzed. For other mathematical forms, the analyst must include the physical decay in all relevant compartments. By analyzing an open, quadratic model, effects of not including decay were seen to be maximized if the biological rate constant was ≥λ, the physical decay constant. Finally, using monoclonal antibody–antigen reactions, similar discrepancies between thezfunctions and the pharmacokinetic variables were demonstrated. This result was found to persist even if competitive molecules were included. We conclude that decay‐corrected data may be shown, but should not be entered into the modeling equations unless the latter are of the linear, homogeneous form.

ISSN:0269-3127    

DOI:10.1118/1.597421

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

In this paper a method of modeling the distribution of scattered events in emission projection data is developed and applied. This method is based on the use of a transmission map to define the inhomogeneous scattering object. The key point is the use of the set of line integrals calculated as part of the attenuation correction technique, as the basis of a model of the distribution of scattered events. The probability of a photon being scattered through a given angle and being detected in the emission energy window is approximated using a Gaussian function. The parameters of this Gaussian are determined using Monte Carlo generated parallel‐beam scatter line spread functions from a nonuniformly attenuating phantom. The model is incorporated into a two‐dimensional projector–backprojector and used with the Expectation‐Maximization‐Maximum‐Likelihood algorithm for the reconstruction of fan‐beam phantom data. The correction is shown to perform well for a phantom that varies slowly in the axial direction. For the more clinically realistic situation of a torso phantom, the method produces improvements in terms of blood pool to myocardium contrast, but does not restore the contrast to the level exhibited in a reconstruction from “scatter free” data.

ISSN:0269-3127    

DOI:10.1118/1.597422

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

The majority of radiation absorbed dose estimates for radioimmunotherapy (RIT) with I‐131 labeled antibodies have been calculated based oninvivoquantitation of activity using the conjugate view approach with planar Anger camera images. Scatter and septal penetration events contributed by a small fraction of high‐energy photons emitted by I‐131 with an energy exceeding 600 keV lead to a significant degradation of I‐131 images acquired with an Anger camera, which blurs the images of uptake sites and complicates the definition of background regions. The objective of this study was to evaluate a triple energy window (TEW) subtraction method that has been used to remove these interfering events from I‐131 images. In the method, a primary photopeak image for I‐131 is obtained after sequential subtraction of septal penetration and scatter events by using scatter multipliers derived from a photopeak window and two adjacent scatter window images. Qualitative improvement in image contrast was demonstrated with this technique, together with more accurate and reproducible quantitation for I‐131 in the organs of an abdominal phantom. This TEW scatter subtraction method can be used to provide more precise dosimetry estimates for radionuclide therapy and RIT with I‐131.

ISSN:0269-3127    

DOI:10.1118/1.597423

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY

摘要:

The growing interest in using MRI to measure distributions of dose in specialized materials has prompted the need to improve existing methods for measuring NMR relaxation rates throughout large volume phantoms. The change in transverse relaxation rate (R2) has proven to be a more sensitive measure of dose than longitudinal relaxation rate (R1) in some materials; however, the accuracy ofR2 measurements is limited by the presence of artifacts resulting from imperfect MRI tip angles throughout large phantoms. A novel echo quotient technique allowing accurate measurement of changes inR2 (ΔR2) in regions of imperfect tip angle is presented. Application of this technique to the measurement of a dynamically wedged photon beam dose distribution demonstrates the improved accuracy of the echo quotient technique compared with other methods. Agreement of the MRI data with film and ion chamber data is within 2.5% throughout the volume of interest, indicating the potential for ΔR2 MRI dosimetry to become a clinically useful tool.

ISSN:0269-3127    

DOI:10.1118/1.597424

出版商:American Association of Physicists in Medicine    

年代:1998

数据来源: WILEY