Classification of mass and normal breast tissue: a convolution neural network classifier with spatial domain and texture images

Breast cancer diagnosis using self-organizing map for sonography

The purpose of this study was to evaluate the performance of neural network model self-organizing maps (SOM) in the classification of benign and malignant sonographic breast lesions. A total of 243 breast tumors (82 malignant and 161 benign) were retrospectively evaluated. When a sonogram was performed, the analog video signal was captured to obtain a digitized sonographic image. The physician selected the region of interest in the sonography. An SOM model using 24 autocorrelation texture features classified the tumor as benign or malignant. In the experiment, cases were sampled with k-fold cross-validation (k = 10) to evaluate the performance using receiver operating characteristic (ROC) curves. The ROC area index for the proposed SOM system is 0.9357 +/- 0.0152, the accuracy is 85. 6%, the sensitivity is 97.6%, the specificity is 79.5%, the positive predictive value is 70.8%, and the negative predictive value is 98. 5%. This computer-aided diagnosis system can provide a useful tool and its high negative predictive value could potentially help avert benign biopsies.

Computer-aided diagnosis applied to US of solid breast nodules by using neural networks

PURPOSE

To increase the capabilities of ultrasonographic (US) technology for the differential diagnosis of solid breast tumors by using a neural network.

MATERIALS AND METHODS

One hundred forty US images of solid breast nodules were evaluated. When a sonogram was obtained, an analog video signal from the VCR output of the scanner was transmitted to a notebook computer. A frame grabber connected to the printer port of the computer was then used to digitize the data. The suspicious tumor region on the digitized US image was manually selected. The texture information of the subimage was extracted, and a neural network classifier with autocorrelation features was used to classify the tumor as benign or malignant. In this experiment, 140 pathologically proved tumors (52 malignant and 88 benign tumors) were sampled with k-fold cross-validation (k = 10) to evaluate the performance with receiver operating characteristic curves.

RESULTS

The accuracy of neural networks for classifying malignancies was 95.0% (133 of 140 tumors), the sensitivity was 98% (51 of 52), the specificity was 93% (82 of 88), the positive predictive value was 89% (51 of 57), and the negative predictive value was 99% (82 of 83).

CONCLUSION

This system differentiated solid breast nodules with relatively high accuracy and helped inexperienced operators to avoid misdiagnoses. Because the neural network is trainable, it could be optimized if a larger set of tumor images is supplied.

A neural network to predict symptomatic lung injury

A nonlinear neural network that simultaneously uses pre-radiotherapy (RT) biological and physical data was developed to predict symptomatic lung injury. The input data were pre-RT pulmonary function, three-dimensional treatment plan doses and demographics. The output was a single value between 0 (asymptomatic) and 1 (symptomatic) to predict the likelihood that a particular patient would become symptomatic. The network was trained on data from 97 patients for 400 iterations with the goal to minimize the mean-squared error. Statistical analysis was performed on the resulting network to determine the model's accuracy. Results from the neural network were compared with those given by traditional linear discriminate analysis and the dose-volume histogram reduction (DVHR) scheme of Kutcher. Receiver-operator characteristic (ROC) analysis was performed on the resulting network which had Az = 0.833 +/- 0.04. (Az is the area under the ROC curve.) Linear discriminate multivariate analysis yielded an Az = 0.813 +/- 0.06. The DVHR method had Az = 0.521 +/- 0.08. The network was also used to rank the significance of the input variables. Future studies will be conducted to improve network accuracy and to include functional imaging data.

Single and multiscale detection of masses in digital mammograms

Scale is an important issue in the automated detection of masses in mammograms, due to the range of possible sizes masses can have. In this work, it was examined if detection of masses can be done at a single scale, or whether it is more appropriate to use the output of the detection method at different scales in a multiscale scheme. Three different pixel-based mass-detection methods were used for this purpose. The first method is based on convolution of a mammogram with the Laplacian of a Gaussian, the second method is based on correlation with a model of a mass, and the third is a new approach, based on statistical analysis of gradient-orientation maps. Experiments with simulated masses indicated that little can be gained by applying the methods at a number of scales. These results were confirmed by experiments on a set of 71 cases (132 mammograms) containing a malignant tumor. The performance of each method in a multiscale scheme was similar to the performance at the optimal single scale. A slight improvement was found for the correlation method when the output of different scales was combined. This was especially evident at low specificity levels. The correlation method and the gradient-orientation-analysis method have similar performances. A sensitivity of approximately 75% is reached at a level of one false positive per image. The method based on convolution with the Laplacian of the Gaussian performed considerably worse, in both a single and multiscale scheme.

Computerized analysis of mammographic microcalcifications in morphological and texture feature spaces

We are developing computerized feature extraction and classification methods to analyze malignant and benign microcalcifications on digitized mammograms. Morphological features that described the size, contrast, and shape of microcalcifications and their variations within a cluster were designed to characterize microcalcifications segmented from the mammographic background. Texture features were derived from the spatial gray-level dependence (SGLD) matrices constructed at multiple distances and directions from tissue regions containing microcalcifications. A genetic algorithm (GA) based feature selection technique was used to select the best feature subset from the multi-dimensional feature spaces. The GA-based method was compared to the commonly used feature selection method based on the stepwise linear discriminant analysis (LDA) procedure. Linear discriminant classifiers using the selected features as input predictor variables were formulated for the classification task. The discriminant scores output from the classifiers were analyzed by receiver operating characteristic (ROC) methodology and the classification accuracy was quantified by the area, Az, under the ROC curve. We analyzed a data set of 145 mammographic microcalcification clusters in this study. It was found that the feature subsets selected by the GA-based method are comparable to or slightly better than those selected by the stepwise LDA method. The texture features (Az = 0.84) were more effective than morphological features (Az = 0.79) in distinguishing malignant and benign microcalcifications. The highest classification accuracy (Az = 0.89) was obtained in the combined texture and morphological feature space. The improvement was statistically significant in comparison to classification in either the morphological (p = 0.002) or the texture (p = 0.04) feature space alone. The classifier using the best feature subset from the combined feature space and an appropriate decision threshold could correctly identify 35% of the benign clusters without missing a malignant cluster. When the average discriminant score from all views of the same cluster was used for classification, the Az value increased to 0.93 and the classifier could identify 50% of the benign clusters at 100% sensitivity for malignancy. Alternatively, if the minimum discriminant score from all views of the same cluster was used, the Az value would be 0.90 and a specificity of 32% would be obtained at 100% sensitivity. The results of this study indicate the potential of using combined morphological and texture features for computer-aided classification of microcalcifications.

Design of a high-sensitivity classifier based on a genetic algorithm: application to computer-aided diagnosis

A genetic algorithm (GA) based feature selection method was developed for the design of high-sensitivity classifiers, which were tailored to yield high sensitivity with high specificity. The fitness function of the GA was based on the receiver operating characteristic (ROC) partial area index, which is defined as the average specificity above a given sensitivity threshold. The designed GA evolved towards the selection of feature combinations which yielded high specificity in the high-sensitivity region of the ROC curve, regardless of the performance at low sensitivity. This is a desirable quality of a classifier used for breast lesion characterization, since the focus in breast lesion characterization is to diagnose correctly as many benign lesions as possible without missing malignancies. The high-sensitivity classifier, formulated as the Fisher's linear discriminant using GA-selected feature variables, was employed to classify 255 biopsy-proven mammographic masses as malignant or benign. The mammograms were digitized at a pixel size of 0.1 mm x 0.1 mm, and regions of interest (ROIs) containing the biopsied masses were extracted by an experienced radiologist. A recently developed image transformation technique, referred to as the rubber-band straightening transform, was applied to the ROIs. Texture features extracted from the spatial grey-level dependence and run-length statistics matrices of the transformed ROIs were used to distinguish malignant and benign masses. The classification accuracy of the high-sensitivity classifier was compared with that of linear discriminant analysis with stepwise feature selection (LDAsfs). With proper GA training, the ROC partial area of the high-sensitivity classifier above a true-positive fraction of 0.95 was significantly larger than that of LDAsfs, although the latter provided a higher total area (Az) under the ROC curve. By setting an appropriate decision threshold, the high-sensitivity classifier and LDAsfs correctly identified 61% and 34% of the benign masses respectively without missing any malignant masses. Our results show that the choice of the feature selection technique is important in computer-aided diagnosis, and that the GA may be a useful tool for designing classifiers for lesion characterization.

Microscopic image analysis for quantitative measurement and feature identification of normal and cancerous colonic mucosa

The development of an automated algorithm for the categorization of normal and cancerous colon mucosa is reported. Six features based on texture analysis were studied. They were derived using the co-occurrence matrix and were angular second moment, entropy, contrast, inverse difference moment, dissimilarity, and correlation. Optical density was also studied. Forty-four normal images and 58 cancerous images from sections of the colon were analyzed. These two groups were split equally into two subgroups: one set was used for supervised training and the other to test the classification algorithm. A stepwise selection procedure showed that correlation and entropy were the features that discriminated most strongly between normal and cancerous tissue (P < 0.0001). A parametric linear-discriminate function was used to determine the classification rule. For the training set, a sensitivity and specificity of 93.1% and 81.8%, respectively, were achieved, with an overall accuracy of 88.2%. These results were confirmed with the test set, with a sensitivity and specificity of 93.1% and 86.4%, respectively, and an overall accuracy of 90.2%.

Linear and neural models for classifying breast masses

Computational methods can be used to provide an initial screening or a second opinion in medical settings and may improve the sensitivity and specificity of diagnoses. In the current study, linear discriminant models and artificial neural networks are trained to detect breast cancer in suspicious masses using radiographic features and patient age. Results on 139 suspicious breast masses (79 malignant, 60 benign, biopsy proven) indicate that a significant probability of detecting malignancies can be achieved at the risk of a small percentage of false positives. Receiver operating characteristic (ROC) analysis favors the use of linear models, however, a new measure related to the area under the ROC curve (AZ) suggests a possible benefit from hybridizing linear and nonlinear classifiers.

False-positive reduction technique for detection of masses on digital mammograms: global and local multiresolution texture analysis

We investigated the application of multiresolution global and local texture features to reduce false-positive detection in a computerized mass detection program. One hundred and sixty-eight digitized mammograms were randomly and equally divided into training and test groups. From these mammograms, two datasets were formed. The first dataset (manual) contained four regions of interest (ROIs) selected manually from each of the mammograms. One of the four ROIs contained a biopsy-proven mass and the other three contained normal parenchyma, including dense, mixed dense/fatty, and fatty tissues. The second dataset (hybrid) contained the manually extracted mass ROIs, along with normal tissue ROIs extracted by an automated Density-Weighted Contrast Enhancement (DWCE) algorithm as false-positive detections. A wavelet transform was used to decompose an ROI into several scales. Global texture features were derived from the low-pass coefficients in the wavelet transformed images. Local texture features were calculated from the suspicious object and the peripheral subregions. Linear discriminant models using effective features selected from the global, local, or combined feature spaces were established to maximize the separation between masses and normal tissue. Receiver Operating Characteristic (ROC) analysis was conducted to evaluate the classifier performance. The classification accuracy using global features were comparable to that using local features. With both global and local features, the average area, Az, under the test ROC curve, reached 0.92 for the manual dataset and 0.96 for the hybrid dataset, demonstrating statistically significant improvement over those obtained with global or local features alone. The results indicated the effectiveness of the combined global and local features in the classification of masses and normal tissue for false-positive reduction.

Computerized classification of malignant and benign microcalcifications on mammograms: texture analysis using an artificial neural network

We investigated the feasibility of using texture features extracted from mammograms to predict whether the presence of microcalcifications is associated with malignant or benign pathology. Eighty-six mammograms from 54 cases (26 benign and 28 malignant) were used as case samples. All lesions had been recommended for surgical biopsy by specialists in breast imaging. A region of interest (ROI) containing the microcalcifications was first corrected for the low-frequency background density variation. Spatial grey level dependence (SGLD) matrices at ten different pixel distances in both the axial and diagonal directions were constructed from the background-corrected ROI. Thirteen texture measures were extracted from each SGLD matrix. Using a stepwise feature selection technique, which maximized the separation of the two class distributions, subsets of texture features were selected from the multi-dimensional feature space. A backpropagation artificial neural network (ANN) classifier was trained and tested with a leave-one-case-out method to recognize the malignant or benign microcalcification clusters. The performance of the ANN was analysed with receiver operating characteristic (ROC) methodology. It was found that a subset of six texture features provided the highest classification accuracy among the feature sets studied. The ANN classifier achieved an area under the ROC curve of 0.88. By setting an appropriate decision threshold, 11 of the 28 benign cases were correctly identified (39% specificity) without missing any malignant cases (100% sensitivity) for patients who had undergone biopsy. This preliminary result indicates that computerized texture analysis can extract mammographic information that is not apparent by visual inspection. The computer-extracted texture information may be used to assist in mammographic interpretation, with the potential to reduce biopsies of benign cases and improve the positive predictive value of mammography.