-
[show abstract]
[hide abstract]
ABSTRACT: The authors investigated a new method to optimize artificial neural networks (ANNs) with adaptive filtering used in computer-assisted detection schemes in digitized mammograms and to assess performance changes when averaging classification scores from three sets of optimized schemes. Two independent training and testing image databases involving 978 and 830 digitized mammograms, respectively, were used in this study. In the training data set, initial filtering and subtraction resulted in the identification of 592 mass regions and 3790 suspicious, but actually negative regions. These regions (including both true-positive and negative regions) were segmented into three subsets three times based on the calculation of the values of three features as segmentation indices. The indices were "mass" size multiplied by their digital value contrast, conspicuity, and circularity. Nine ANN-based classifiers were separately optimized using a genetic algorithm for each subset of regions. Each region was assigned three classification scores after applying the three adaptive ANNs. The performance gain of the CAD scheme after averaging the three scores for each suspicious region was tested using an independent data set and a ROC methodology. The experimental results showed that the areas under ROC curves (Az) for the testing database using three sets of optimized ANNs individually were 0.84+/-0.01, 0.83+/-0.01, and 0.84+/-0.01, respectively. The between-index correlations of three A values were 0.013, -0.007, and 0.086. Similar to averaging diagnostic ratings from independent observers, by averaging three ANN-generated scores for each testing region, the performance of the CAD scheme was significantly improved (p<0.001) with Az value of 0.95+/-0.01.
Medical Physics 12/2001; 28(11):2302-8. · 2.83 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: The purpose of this work was to develop and evaluate a computer-aided detection (CAD) scheme for the improvement of mass identification on digitized mammograms using a knowledge-based approach. Three hundred pathologically verified masses and 300 negative, but suspicious, regions, as initially identified by a rule-based CAD scheme, were randomly selected from a large clinical database for development purposes. In addition, 500 different positive and 500 negative regions were used to test the scheme. This suspicious region pruning scheme includes a learning process to establish a knowledge base that is then used to determine whether a previously identified suspicious region is likely to depict a true mass. This is accomplished by quantitatively characterizing the set of known masses, measuring "similarity" between a suspicious region and a "known" mass, then deriving a composite "likelihood" measure based on all "known" masses to determine the state of the suspicious region. To assess the performance of this method, receiver-operating characteristic (ROC) analyses were employed. Using a leave-one-out validation method with the development set of 600 regions, the knowledge-based CAD scheme achieved an area under the ROC curve of 0.83. Fifty-one percent of the previously identified false-positive regions were eliminated, while maintaining 90% sensitivity. During testing of the 1,000 independent regions, an area under the ROC curve as high as 0.80 was achieved. Knowledge-based approaches can yield a significant reduction in false-positive detections while maintaining reasonable sensitivity. This approach has the potential of improving the performance of other rule-based CAD schemes.
Medical Physics 05/2001; 28(4):455-61. · 2.83 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: The authors' purpose was to assess the effects of Joint Photographic Experts Group (JPEG) image data compression on the performance of computer-assisted detection (CAD) schemes for the detection of masses and microcalcification clusters on digitized mammograms.
This study included 952 mammograms that were digitized and compressed with a JPEG-compatible image-compression scheme. A CAD scheme, previously developed in the authors' laboratory and optimized for noncompressed images, was applied to reconstructed images after compression at five levels. The performance was compared with that obtained with the original noncompressed digitized images.
For mass detection, there were no significant differences in performance between noncompressed and compressed images for true-positive regions (P = .25) or false-positive regions (P = .40). In all six modes the scheme identified 80% of masses with less than one false-positive region per image. For the detection of microcalcification clusters, there was significant performance degradation (P < .001) at all compression levels. Detection sensitivity was reduced by 4%-10% as compression ratios increased from 17:1 to 62:1. At the same time, the false-positive detection rate was increased by 91%-140%.
The JPEG algorithm did not adversely affect the performance of the CAD scheme for detecting masses, but it did significantly affect the detection of microcalcification clusters.
Academic Radiology 08/2000; 7(8):595-602. · 1.69 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: The authors compared two computerized methods, the arc and cartesian straight-line, for the localization of breast lesions in two mammographic views.
A total of 571 craniocaudal and 571 mediolateral oblique matched mammographic image pairs (or 1142 individual images) depicting 290 pathology-verified masses on both views were selected from our image database. Using a previously developed computer-aided detection scheme, all 290 masses and 3992 suspicious but negative regions were identified. After pairing all identified regions from both views, all masses (true-positive-true-positive matched pairs) and a total of 10330 false-positive pairs (including false-positive-false-positive, true-positive-false-positive, and false-positive-true positive pairs) were assessed as to their position in relation to the nipple using both the arc and the cartesian straight-line methods. Receiver operating characteristic methodology was used to evaluate the performance levels for each method in determining, based solely on location, whether a pair of suspicious regions represented a true mass or a false-positive combination.
The areas under the receiver operating characteristic curves (Az) were 0.79 and 0.78 for the arc and cartesian straight-line methods, respectively. The difference between the two techniques (as measured by Az) was not statistically significant (P > 0.99).
These preliminary results demonstrated that the two methods are comparable in identifying true masses from triangulated observations on two views. However, the arc method is somewhat favorable because only the nipple location is required for localization.
Investigative Radiology 10/1999; 34(9):585-8. · 4.59 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: To investigate optimization of feature selection for computerized mass detection in digitized mammograms, and to compare the effectiveness of a genetic algorithm (GA) in such optimization with that of an "exhaustive" search of all feature permutations.
A Bayesian belief network (BBN) was used to classify positive and negative regions for masses depicted in digitized mammograms; 20 features were computed for each of 592 positive and 3,790 negative regions in two databases. Conditional probabilities for the BBN were computed by using a "training" database of 288 positive and 2,204 negative regions. Performance was measured by the area under the receiver operating characteristic curve (A) by using the remainder database (304 positive and 1,586 negative regions). The optimal set was first found by using an "exhaustive" (complete permutation) searching method. A GA-based search for the optimal set then was applied, and the results of the two approaches were compared.
As the number of features in the classifier increased, the A value increased until it reached a maximum performance for 11 features of 0.876 +/- 0.008. The A value then decreased monotonically as the number of features increased from 11 to 20. Using 100 random chromosomes (seeds) in the first generation, the GA identified the same optimal set of features but reduced the total computation time by a factor of 65.
A GA-based search might be an efficient and effective approach to selecting an optimal feature set.
Academic Radiology 07/1999; 6(6):327-32. · 1.69 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: This study investigates a simple Bayesian belief network for the diagnosis of breast cancer, and specifically addresses the question of whether integrating image and non-image based features into a single network can yield better performance than hybrid combinations of independent networks. From a dataset of 419 cases, including 92 malignancies, 13 features relating to mammographic findings, physical examinations and patients' clinical histories, were extracted to build three Bayesian belief networks. The scenarios tested included a network incorporating all features and two hybrids which combined the outputs of sub-networks corresponding to the image or non-image features. Average areas (Az) under the corresponding ROC curves were used as measures of performance. The network incorporating only image based features performed better (Az =0.81) than that using nonimage features (Az = 0.71). Both hybrid classifiers yielded better performance (Az =0.85 for averaging and Az = 0.87 for logistic regression), but neither hybrid was as accurate as the network incorporating all features (Az = 0.89). This preliminary study suggests that, like human observers who concurrently consider different types of information, a single classifier that simultaneously evaluates both image and non-image information can achieve better diagnostic performance than the hybrid combinations considered here.
International Journal of Medical Informatics 06/1999; 54(2):115-26. · 2.41 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: A mathematical morphology-based computer-aided detection (CAD) scheme for the identification of clustered microcalcifications was developed and tested. The potential for improving either sensitivity or specificity by combining the results with those previously reported was investigated.
The CAD scheme presented here is based on mathematical morphology and a series of simple rule-based criteria for the identification of clustered microcalcifications. A database of 105 digitized mammograms was used for training and rule setting of the scheme. A test set of 191 digitized mammograms was used to evaluate its performance. The same test set had been used to evaluate a multilayer, topography-based scheme. The results obtained by the two schemes were then combined using logical OR and AND operations.
The morphology-based and topography-based CAD schemes performed at sensitivities of 82.9% and 89.5%, with false-positive detection rates of 1.3 and 0.4 per image, respectively. A logical OR operation resulted in 95.4% sensitivity. An AND operation achieved 76.2% sensitivity, with no false identifications on 93% of images.
By combining the results of the morphology-based and the topography-based schemes, either sensitivity or specificity can be improved.
Investigative Radiology 11/1998; 33(10):746-51. · 4.59 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: The authors evaluated whether a hybrid classifier of two independent computer-aided diagnosis (CAD) schemes, the set enumeration (SE) trees approach and an artificial neural network (ANN), could improve the detection of masses on digitized mammograms. The potential benefits resulting from the interpretability of the SE trees model was also explored.
Two hundred thirty verified mass regions and 230 negative but suspicious regions were randomly selected from 618 digitized mammograms. Each region was represented by a 24-parameter feature vector. These features were used as input data for the SE trees and ANN-based schemes. After the positive and negative regions were randomly segmented into five exclusive partitions, a fivefold cross-validation method was applied to evaluate and compare the performance of the SE trees, ANN, and hybrid system in the identification of masses.
The performance of the SE trees approach was comparable to that of the ANN. The average area under the receiver operating characteristic (ROC) curves for all five partitions was 0.88 (standard deviation, 0.04). Owing to the relatively low correlation between the region-based results of the SE trees and ANN methods, the hybrid classifier yielded a significantly improved performance, with an area under the ROC curve of 0.94 (standard deviation, 0.02; P < .05).
The hybrid CAD scheme significantly improved performance. The amenability of the SE trees models to interpretation may aid in the assessment of the importance of specific features.
Academic Radiology 03/1998; 5(3):181-7. · 1.69 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: The authors assessed the performance changes of a computer-assisted diagnosis (CAD) scheme as a function of the number of regions used for training (rule-setting).
One hundred twenty regions depicting actual masses and 400 suspicious but actually negative regions were selected as a testing data set from a database of 2,146 regions identified as suspicious on 618 mammograms. An artificial neural network using 24 and 16 region-based features as input neurons was applied to classify the regions as positive or negative for the presence of a mass. CAD scheme performance was evaluated on the testing data set as the number of regions used for training increased from 60 to 496.
As the number of regions in the training sets increased, the results decreased and plateaued beyond a sample size of approximately 200 regions. Performance with the testing data set continued to improve as the training data set increased in size.
A trend in a system's performance as a function of training set size can be used to assess adequacy of the training data set in the development of a CAD scheme.
Academic Radiology 08/1997; 4(7):497-502. · 1.69 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: The authors assessed the performance of an existing computer-aided diagnosis (CAD) scheme for the detection of clustered microcalcifications in a large image database.
A previously developed, rule-based system was used to assess detectability of microcalcification clusters in a set of 386 digitized mammograms with 239 verified clusters visible on 191 images. The test was performed without any reoptimization of the scheme. None of the 386 images had been used in any previous scheme development or testing procedures.
The CAD scheme achieved 89.5% sensitivity at an average false-positive detection rate of 0.39 per image. In 75% of all images, no false-positive findings occurred. Twenty-three of 25 false-negative findings (misses) occurred during the last two stages in the detection process.
This scheme produced reasonable results in a large data set of images with a large variety of cluster characteristics.
Academic Radiology 07/1997; 4(6):415-8. · 1.69 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: As research efforts for developing computer-aided diagnosis (CAD) schemes of digitized mammograms increase and interscheme results are compared, the desire to establish an acceptable consistent reporting protocol of the distribution of abnormal characteristic is becoming an issue. "Mass contrast" is very important and frequently reported in current CAD studies. In this report, 100 verified mass regions were analyzed systemically using 6 different definitions of "mass contrast." Measured variability in mass contrast was demonstrated by the distribution shift in this group masses. The need for universally accepted and largely standardized descriptors of objects of interest is clearly demonstrated.
Medical Physics 01/1997; 23(12):2007-9. · 2.83 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Using two independent computer-assisted diagnosis (CAD) schemes, we investigated the potential to improve the sensitivity of mass detection by applying a logical "or" operation and to improve the specificity using a logical "and" operation.
Two independent mass detectors, one with Gaussian bandpass filtering and multilayer topographic feature analysis and the other with a five-stage search for a single suspicious region, were applied to a large image database that included 428 digitized mammograms with 220 verified masses. The performance of the two schemes and a combination of them in the form of either logical "or" or logical "and" operations were compared.
In this preliminary study, a multilayer topographic feature analysis CAD scheme (CAD-1) achieved a sensitivity of 96% and had a false-positive detection rate of 0.79 per image. A five-stage search method scheme (CAD-2) achieved a sensitivity of 94% and had a false-positive detection rate of 1.69 per image. With an "or" operation, the combined results yielded 100% sensitivity with a false-positive detection rate of 2.07 per image. A logical "and" operation produced a reduction of the false-positive detection rate to 0.4 per image, but sensitivity also decreased to 90%.
Similar to an independent double-reading approach and depending upon the relevant clinical question, sensitivity or specificity can be improved by combining the results of several independent CAD schemes.
American Journal of Roentgenology 01/1997; 167(6):1421-4. · 2.78 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: We investigated an adaptive rule-based computer-aided diagnosis (CAD) scheme for digitized mammograms that can be optimized by using an image difficulty index as determined from global measures of image characteristics.
First, we defined an image "difficulty" index based on image feature measurements in both the spatial and frequency domains. The CAD scheme then segmented the database into three groups. An image database of 428 digitized mammograms with 220 verified masses was randomly divided into two subsets, one for training (rule-setting) and the other for testing the adaptive CAD scheme. Each of the image difficulty groups in the training set was optimized independently to achieve a low false-positive detection rate while maintaining high detection sensitivity. Scheme performance was then evaluated with the test set, and the results were compared with a global rule-based system that was optimized without the adaptive method.
In this preliminary study, a relatively simple adaptive scheme reduced false-positive mass detections compared with the nonadaptive scheme from 0.85 to 0.53 per image. At the same time sensitivity was not significantly changed.
This adaptive CAD scheme has distinct advantages in improving CAD scheme performance as long as the training database includes a large number of cases in each image difficulty group with a variety of true-positive abnormalities.
Academic Radiology 11/1996; 3(10):806-14. · 1.69 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: The authors assess the robustness of a computer-aided diagnosis (CAD) scheme with five rule-based stages to identify regions suspicious for mass in digitized mammograms.
With a database of 428 mammograms, 234 of which had not been analyzed by this scheme before, the authors evaluated the performance robustness of their CAD scheme. The following four issues were investigated to assess the variability of the scheme's performance due to: (1) the maximum permissible number of "masses" detected at each stage; (2) exclusion of selected individual rule-based stages; (3) added image noise; and (4) repeated digitizations of the same image.
Enabling the CAD scheme to select a maximum of two suspicious mass regions at any one stage increased sensitivity by as much as 4% (from 93% to 97%), but it increased the false-positive detection rate by as much as 1.2 per image (from 1.7 to 2.9). Eliminating any individual stage decreased sensitivity by as much as 6%, but this reduced the false-positive detection rate by as much as 0.4 per image (from 1.7 to 1.3). The addition of reasonable noise levels decreased sensitivity by as much as 4% without substantially affecting the false-positive detections. Repeated digitizations of selected images demonstrated a scheme sensitivity of 93% +/- 1.8% with more than a 90% overlap of the false-positive regions.
The results of this preliminary study clearly indicate that this scheme is reasonably robust to the variables investigated here.
Investigative Radiology 10/1996; 31(9):563-8. · 4.59 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: A simple and effective computerized detection scheme was developed to identify suspicious mass regions in digitized mammograms.
This method identifies a maximum of five suspicious mass regions per image and was tested with a database of 510 images, including 162 verified masses. It includes a series of five rule-based processes that select one region with each of the following characteristics: 1) a global minimum of optical density in a smoothed image; 2) a local minimum of optical density in the original image; 3) a local minimum of optical density in a filtered image; 4) a small "mass" of low contrast; and 5) a small "mass" of high contrast.
This multi-stage process achieved a sensitivity of 95% while limiting false-positive detection rates to below an average of two per image.
Because this method limits the initial number of suspicious mass regions while retaining high sensitivity, it may be applicable to clinically usable computer-aided diagnosis schemes.
Investigative Radiology 04/1996; 31(3):146-53. · 4.59 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Two methods--single-image segmentation and bilateral-image subtraction--have been used commonly as the first stage in computer-aided detection (CAD) schemes to detect masses on digitized mammograms. In the current study, we investigated and compared the advantages and disadvantages of the two methods in achieving a high sensitivity for mass detection.
Two CAD schemes were tested. One used Gaussian filtering based on single-image segmentation, and the other used bilateral-image subtraction based on left-right image pairs to identify suspicious mass regions. A clinical database that contained 152 verified mass cases was used to compare the two approaches.
The single-image segmentation method yielded 100% sensitivity and had a somewhat higher number of initial suspicious regions. The bilateral-image subtraction method missed several true-positive regions at the initial phase. Each approach achieved more than 90% sensitivity at a false-positive rate of approximately 0.8 per image.
Optimal initial image segmentation schemes may depend on the complete detection and classification method used. Single-image segmentation methods may perform comparably with bilateral-image segmentation schemes, and these techniques appear to be more versatile and easily adaptable to future clinical CAD applications.
Academic Radiology 01/1996; 2(12):1056-61. · 1.69 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: We developed and evaluated a computer-aided detection (CAD) scheme for masses in digitized mammograms.
A multistep CAD scheme was developed and tested. The method uses a technique of single-image segmentation with Gaussian bandpass filtering to yield a high sensitivity for mass detection. A rule-based multilayer topographic feature analysis method is then used to classify suspected regions. A set of 260 cases, including 162 verified masses, was divided into two subsets; one set was used to set the rule-based classification and one was used to test the performance of the scheme.
In a preliminary clinical study, the implemented detection scheme yielded 98% sensitivity with a false-positive detection rate of less than one false-positive region per image.
Single-image segmentation methods seem to have high sensitivity in selecting true-positive mass regions in the first stage of a CAD scheme. A multilayer topographic image feature analysis method in the second stage of a CAD scheme has the potential to significantly reduce the false-positive detection rate.
Academic Radiology 12/1995; 2(11):959-66. · 1.69 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: We investigated a computer-aided detection (CAD) scheme for clustered microcalcifications in digitized mammograms.
A multistage CAD scheme was developed and tested. To increase sensitivity, the scheme uses a Gaussian band-pass filter and nonlinear threshold. A multistage local minimum searching routine and a multilayer topographic feature analysis are used to reduce the false-positive detection rate. One hundred ten digitized mammograms were used in this preliminary test, with 55 images containing one or two verified microcalcification clusters.
The CAD scheme achieved 100% sensitivity and had an average false-positive detection rate of 0.18 per image.
The CAD scheme performs as well as many published schemes and has some unique advantages to further improve detection sensitivity and specificity of future CAD schemes.
Academic Radiology 09/1995; 2(8):655-62. · 1.69 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Fractal analysis of digitized images has been investigated in recent years as a potential measure of structural bone strength. Several technical issues associated with such measurements are assessed.
In a series of experiments using a hand phantom, the effects of system noise and modulation transfer function on fractal dimension were explored. The authors evaluated a method for correcting the estimated power spectrum using a step-wedge image exposed and digitized under identical conditions as a reference.
System noise and modulation transfer function significantly affect estimated fractal dimension in bony regions computed from conventional radiographs.
Before conventional radiographs are used for fractal analysis in the clinical environment, many of the technical problems associated with this methodology must be addressed.
Investigative Radiology 07/1994; 29(6):624-9. · 4.59 Impact Factor
