Wen-Jia Kuo

Diagnosis of breast tumors with sonographic texture analysis using wavelet transform and neural networks

To increase the ability of ultrasonographic technology for the differential diagnosis of solid breast tumors, we describe a novel computer-aided diagnosis (CADx) system using neural networks for classification of breast tumors. Tumor regions and surrounding tissues are segmented from the physician-located region-of-interest (ROI) images by applying our proposed segmentation algorithm. Cooperating with the segmentation algorithm, three feasible features, including variance contrast, autocorrelation contrast and distribution distortion of wavelet coefficients, were extracted from the ROI images for further classification. A multilayered perceptron (MLP) neural network trained using error back-propagation algorithm with momentum was then used for the differential diagnosis of breast tumors on sonograms. In the experiment, 242 cases (including benign breast tumors from 161 patients and carcinomas from 82 patients) were sampled with k-fold cross-validation (k = 10) to evaluate the performance. The receiver operating characteristic (ROC) area index for the proposed CADx system is 0.9396 +/- 0.0183, the sensitivity is 98.77%, the specificity is 81.37%, the positive predictive value is 72.73% and the negative predictive value is 99.24%. Experimental results showed that our diagnosis model performed very well for breast tumor diagnosis.

Data mining with decision trees for diagnosis of breast tumor in medical ultrasonic images

To increase the ability of ultrasonographic (US) technology for the differential diagnosis of solid breast tumors, we describe a novel computer-aided diagnosis (CADx) system using data mining with decision tree for classification of breast tumor to increase the levels of diagnostic confidence and to provide the immediate second opinion for physicians. Cooperating with the texture information extracted from the region of interest (ROI) image, a decision tree model generated from the training data in a top-down, general-to-specific direction with 24 co-variance texture features is used to classify the tumors as benign or malignant. In the experiments, accuracy rates for a experienced physician and the proposed CADx are 86.67% (78/90) and 95.50% (86/90), respectively.

Retrieval technique for the diagnosis of solid breast tumors on sonogram

We evaluated a series of pathologically proven breast tumors using an image-retrieval technique for classifying benign and malignant lesions. A total of 263 breast tumors (129 malignant and 134 benign) were retrospectively evaluated. The physician located regions-of-interest (ROI) of ultrasonic images and texture parameters (contrast, covariance and dissimilarity) were used in the process of the content-based image-retrieval technique. The accuracy of using the retrieval technique for classifying malignancies was 92.55% (236 of 255), the sensitivity was 94.44% (119 of 126), the specificity was 90.70% (117 of 129), the positive predictive value was 90.84% (119 of 131), and negative predictive value was 94.35% (117 of 124) for the proposed computer-aided diagnostic system. This computer-aided diagnosis system can provide a useful tool and its high negative predictive value could potentially help avert benign biopsies. It is unnecessary to perform any training procedures. This computer-aided diagnosis system can provide a second opinion for a sonographic interpreter; the main advantage in this proposed system is that we do not need any training. Historical cases can be directly added into the database and training of the diagnosis system again is not needed. With the growth of the database, more and more information can be collected and used as reference cases while performing diagnoses. This increases the flexibility of our diagnostic system.

Computer-Aided Diagnosis of Breast Tumors with Different US Systems

RATIONALE AND OBJECTIVES The authors performed this study to determine whether a computer-aided diagnostic (CAD) system was suitable from one ultrasound (US) unit to another after parameters were adjusted by using intelligent selection algorithms. MATERIALS AND METHODS The authors used texture analysis and data mining with a decision tree model to classify breast tumors with different US systems. The databases of training cases from one unit and testing cases from another were collected from different countries. Regions of interest on US scans and co-variance texture parameters were used in the diagnosis system. Proposed adjustment schemes for different US systems were used to transform the information needed for a differential diagnosis. RESULTS Comparison of the diagnostic system with and without adjustment, respectively, yielded the following results: accuracy, 89.9% and 82.2%; sensitivity, 94.6% and 92.2%; specificity, 85.4% and 72.3%; positive predictive value, 86.5% and 76.8%; and negative predictive value, 94.1% and 90.4%. The improvement in accuracy, specificity, and positive predictive value was statistically significant. Diagnostic performance was improved after the adjustment. CONCLUSION After parameters were adjusted by using intelligent selection algorithms, the performance of the proposed CAD system was better both with the same and with different systems. Different resolutions, different setting conditions, and different scanner ages are no longer obstacles to the application of such a CAD system.

Use of the bootstrap technique with small training sets for computer-aided diagnosis in breast ultrasound

The purpose of this study was to test the efficacy of using small training sets in computer-aided diagnostic systems (CAD) and to increase the capabilities of ultrasound (US) technology in the differential diagnosis of solid breast tumors. A total of 263 sonographic images of solid breast nodules, including 129 malignancies and 134 benign nodules, were evaluated by using a bootstrap technique with 10 original training samples. Texture parameters of a region-of-interest (ROI) were resampled with a bootstrap technique and a decision-tree model was used to classify the tumor as benign or malignant. The accuracy was 87.07% (229 of 263 tumors), the sensitivity was 95.35% (123 of 129), the specificity was 79.10% (106 of 134), the positive predictive value was 81.46% (123 of 151), and the negative predictive value was 94.64% (106 of 112). This analysis method provides a second opinion for physicians with high accuracy. The new method shows a potential to be useful in future application of CAD, especially when a large database cannot be obtained for training or a newly developed ultrasonic system has smaller sets of samples.

Adaptive predictive multiplicative autoregressive model for medical image compression

An adaptive predictive multiplicative autoregressive (APMAR) method is proposed for lossless medical image coding. The adaptive predictor is used for improving the prediction accuracy of encoded image blocks in our proposed method. Each block is first adaptively predicted by one of the seven predictors of the JPEG lossless mode and a local mean predictor. It is clear that the prediction accuracy of an adaptive predictor is better than that of a fixed predictor. Then the residual values are processed by the multiplicative autoregressive (MAR) model with Huffman coding. Comparisons with other methods [MAR, space-varying MAR (SMAR) and adaptive JPEG (AJPEG) models] on a series of test images show that our method is suitable for reversible medical image compression.In this paper, an adaptive predictive multiplicative autoregressive (APMAR) method is proposed for lossless medical image coding. The adaptive predictor is used for improving the prediction accuracy of encoded image blocks in our proposed method. Each block is first adaptively predicted by one of the seven predictors of the JPEG lossless mode and a local mean predictor. It is clear that the prediction accuracy of an adaptive predictor is better than that of a fixed predictor. Then the residual values are processed by the MAR model with Huffman coding. Comparisons with other methods [MAR, SMAR, adaptive JPEG (AJPEG)] on a series of test images show that our method is suitable for reversible medical image compression.

Image retrieval on uncompressed and compressed domains

With the rapid growth of the Internet and multimedia systems, digital images have been enormously used in many applications now. The way of how to retrieve the image we want from a large image database correctly plays a more important role. Currently, most proposed methods for image retrieval can only be applied with images in the same representing domain. We propose a novel methodology that extracts the feature image directly from its original domain without any decompressing procedures. The feature is further modified to achieve the rotation-invariant and scaling property. Our proposed method can be used for image retrieval on the three most frequently adopted types of images nowadays: spatial, DCT compressed, and wavelet-compressed images. Experimental results show that our proposed method is quite effective not only for the performance but also is not afraid of scale and rotation.

Edge-based motion compensated classified DCT with quadtree for image sequence coding

Abstract The motion compensated discrete cosine transform coding (MCDCT) is an efficient image sequence coding technique. In order to further reduce the bit-rate for the quantizied DCT coefficients and keep the visual quality, we propose an adaptive edge-based quadtree motion compensated discrete cosine transform coding (EQDCT). In our proposed algorithm, the overhead moving information is encoded by a quadtree structure and the nonedge blocks will be encoded at lower bit-rate but the edge blocks will be encoded at higher bit-rate. The edge blocks will be further classified into four different classes according to the orientations and locations of the edges. Each class of edge blocks selects the different set of the DCT coefficients to be encoded. By this method, we can just preserve and encode a few DCT coefficients, but still maintain the visual quality of the images. In the proposed EQDCT image sequence coding scheme, the average bit-rate of each frame is reduced to 0.072 bit/pixel and the average PSNR value is 32.11 dB.

Two-pass side-match finite-state vector quantization

Among the image coding techniques, vector quantization (VQ) has been considered to be an effective method for coding images at low bit rate. Side-match finite-state vector quantizer (SMVQ) exploits the correlations between the neighboring blocks (vectors) to avoid large gray level transition across block boundaries. In this paper, an improved SMVQ technique named two-pass side-match finite-state vector quantization (TPSMVQ) has been proposed. In TPSMVQ, the size of state codebook in the first pass is decided by the variances of neighboring blocks. In the second pass, we will improve the blocks encoded in the first pass whose variances are greater than a threshold. Moreover, not only the left and upper blocks but also the down and right blocks are used for constructing the state codebook. In our experiment results, the improvement of second pass is up to 1.5 dB in PSNR over the fist pass. In comparison to ordinary SMVQ, the improvement is upt to 1.54 dB at nearly the same bit rate.