Ultrasonic Assessment of Liver Fibrosis Using One-Dimensional Convolutional Neural Networks Based on Frequency Spectra of Radiofrequency Signals with Deep Learning Segmentation of Liver Regions in B-Mode Images: A Feasibility Study.

The early detection of liver fibrosis is of significant importance. Deep learning analysis of ultrasound backscattered radiofrequency (RF) signals is emerging for tissue characterization as the RF signals carry abundant information related to tissue microstructures. However, the existing methods only used the time-domain information of the RF signals for liver fibrosis assessment, and the liver region of interest (ROI) is outlined manually. In this study, we proposed an approach for liver fibrosis assessment using deep learning models on ultrasound RF signals. The proposed method consisted of two-dimensional (2D) convolutional neural networks (CNNs) for automatic liver ROI segmentation from reconstructed B-mode ultrasound images and one-dimensional (1D) CNNs for liver fibrosis stage classification based on the frequency spectra (amplitude, phase, and power) of the segmented ROI signals. The Fourier transform was used to obtain the three kinds of frequency spectra. Two classical 2D CNNs were employed for liver ROI segmentation: U-Net and Attention U-Net. ROI spectrum signals were normalized and augmented using a sliding window technique. Ultrasound RF signals collected (with a 3-MHz transducer) from 613 participants (Group A) were included for liver ROI segmentation and those from 237 participants (Group B) for liver fibrosis stage classification, with a liver biopsy as the reference standard (Fibrosis stage: F0 = 27, F1 = 49, F2 = 51, F3 = 49, F4 = 61). In the test set of Group A, U-Net and Attention U-Net yielded Dice similarity coefficients of 95.05% and 94.68%, respectively. In the test set of Group B, the 1D CNN performed the best when using ROI phase spectrum signals to evaluate liver fibrosis stages ≥F1 (area under the receive operating characteristic curve, AUC: 0.957; accuracy: 89.19%; sensitivity: 85.17%; specificity: 93.75%), ≥F2 (AUC: 0.808; accuracy: 83.34%; sensitivity: 87.50%; specificity: 78.57%), and ≥F4 (AUC: 0.876; accuracy: 85.71%; sensitivity: 77.78%; specificity: 94.12%), and when using the power spectrum signals to evaluate ≥F3 (AUC: 0.729; accuracy: 77.14%; sensitivity: 77.27%; specificity: 76.92%). The experimental results demonstrated the feasibility of both the 2D and 1D CNNs in liver parenchyma detection and liver fibrosis characterization. The proposed methods have provided a new strategy for liver fibrosis assessment based on ultrasound RF signals, especially for early fibrosis detection. The findings of this study shed light on deep learning analysis of ultrasound RF signals in the frequency domain with automatic ROI segmentation.

Design of Path-Planning System for Interventional Thermal Ablation of Liver Tumors Based on CT Images.

OBJECTIVE: Aiming at the shortcomings of artificial surgical path planning for the thermal ablation of liver tumors, such as the time-consuming and labor-consuming process, and relying heavily on doctors' puncture experience, an automatic path-planning system for thermal ablation of liver tumors based on CT images is designed and implemented. METHODS: The system mainly includes three modules: image segmentation and three-dimensional reconstruction, automatic surgical path planning, and image information management. Through organ segmentation and three- dimensional reconstruction based on CT images, the personalized abdominal spatial anatomical structure of patients is obtained, which is convenient for surgical path planning. The weighted summation method based on clinical constraints and the concept of Pareto optimality are used to solve the multi-objective optimization problem, screen the optimal needle entry path, and realize the automatic planning of the thermal ablation path. The image information database was established to store the information related to the surgical path. RESULTS: In the discussion with clinicians, more than 78% of the paths generated by the planning system were considered to be effective, and the efficiency of system path planning is higher than doctors' planning efficiency. CONCLUSION: After improvement, the system can be used for the planning of the thermal ablation path of a liver tumor and has certain clinical application value.

Ultrasound Homodyned-K Contrast-Weighted Summation Parametric Imaging Based on H-scan for Detecting Microwave Ablation Zones.

The homodyned-K (HK) distribution is a generalized model of envelope statistics whose parameters α (the clustering parameter) and k (the coherent-to-diffuse signal ratio) can be used to monitor the thermal lesions. In this study, we proposed an ultrasound HK contrast-weighted summation (CWS) parametric imaging algorithm based on the H-scan technique and investigated the optimal window side length (WSL) of the HK parameters estimated by the XU estimator (an estimation method based on the first moment of the intensity and two log-moments, which was used in the proposed algorithm) through phantom simulations. H-scan diversified ultrasonic backscattered signals into low- and high-frequency passbands. After envelope detection and HK parameter estimation for each frequency band, the α and k parametric maps were obtained, respectively. According to the contrast between the target region and background, the (α or k) parametric maps of the dual-frequency band were weighted and summed, and then the CWS images were yielded by pseudo-color imaging. The proposed HK CWS parametric imaging algorithm was used to detect the microwave ablation coagulation zones of porcine liver ex vivo under different powers and treatment durations. The performance of the proposed algorithm was compared with that of the conventional HK parametric imaging and frequency diversity and compounding Nakagami imaging algorithms. For two-dimensional HK parametric imaging, it was found that a WSL equal to 4 pulse lengths of the transducer was sufficient for estimating the α and k parameters in terms of both parameter estimation stability and parametric imaging resolution. The HK CWS parametric imaging provided an improved contrast-to-noise ratio over conventional HK parametric imaging, and the HK αcws parametric imaging achieved the best accuracy and Dice score of coagulation zone detection.

A Review of Quantitative Ultrasound-Based Approaches to Thermometry and Ablation Zone Identification Over the Past Decade.

Percutaneous thermal therapy is an important clinical treatment method for some solid tumors. It is critical to use effective image visualization techniques to monitor the therapy process in real time because precise control of the therapeutic zone directly affects the prognosis of tumor treatment. Ultrasound is used in thermal therapy monitoring because of its real-time, non-invasive, non-ionizing radiation, and low-cost characteristics. This paper presents a review of nine quantitative ultrasound-based methods for thermal therapy monitoring and their advances over the last decade since 2011. These methods were analyzed and compared with respect to two applications: ultrasonic thermometry and ablation zone identification. The advantages and limitations of these methods were compared and discussed, and future developments were suggested.

Liver Fibrosis Assessment Using Radiomics of Ultrasound Homodyned-K imaging Based on the Artificial Neural Network Estimator.

The homodyned-K distribution is an important ultrasound backscatter envelope statistics model of physical meaning, and the parametric imaging of the model parameters has been explored for quantitative ultrasound tissue characterization. In this paper, we proposed a new method for liver fibrosis characterization by using radiomics of ultrasound backscatter homodyned-K imaging based on an improved artificial neural network (iANN) estimator. The iANN estimator was used to estimate the ultrasound homodyned-K distribution parameters k and α from the backscattered radiofrequency (RF) signals of clinical liver fibrosis (n = 237), collected with a 3-MHz convex array transducer. The RF data were divided into two groups: Group I corresponded to liver fibrosis with no hepatic steatosis (n = 94), and Group II corresponded to liver fibrosis with mild to severe hepatic steatosis (n = 143). The estimated homodyned-K parameter values were then used to construct k and α parametric images using the sliding window technique. Radiomics features of k and α parametric images were extracted, and feature selection was conducted. Logistic regression classification models based on the selected radiomics features were built for staging liver fibrosis. Experimental results showed that the proposed method is overall superior to the radiomics method of uncompressed envelope images when assessing liver fibrosis. Regardless of hepatic steatosis, the proposed method achieved the best performance in staging liver fibrosis ≥F1, ≥F4, and the area under the receiver operating characteristic curve was 0.88, 0.85 (Group I), and 0.85, 0.86 (Group II), respectively. Radiomics has improved the ability of ultrasound backscatter statistical parametric imaging to assess liver fibrosis, and is expected to become a new quantitative ultrasound method for liver fibrosis characterization.

Fetal Electrocardiogram Signal Extraction Based on Fast Independent Component Analysis and Singular Value Decomposition.

Fetal electrocardiograms (FECGs) provide important clinical information for early diagnosis and intervention. However, FECG signals are extremely weak and are greatly influenced by noises. FECG signal extraction and detection are still challenging. In this work, we combined the fast independent component analysis (FastICA) algorithm with singular value decomposition (SVD) to extract FECG signals. The improved wavelet mode maximum method was applied to detect QRS waves and ST segments of FECG signals. We used the abdominal and direct fetal ECG database (ADFECGDB) and the Cardiology Challenge Database (PhysioNet2013) to verify the proposed algorithm. The signal-to-noise ratio of the best channel signal reached 45.028 dB and the issue of missing waveforms was addressed. The sensitivity, positive predictive value and F1 score of fetal QRS wave detection were 96.90%, 98.23%, and 95.24%, respectively. The proposed algorithm may be used as a new method for FECG signal extraction and detection.

Fetal Ultrasound Image Segmentation for Automatic Head Circumference Biometry Using Deeply Supervised Attention-Gated V-Net.

Automatic computerized segmentation of fetal head from ultrasound images and head circumference (HC) biometric measurement is still challenging, due to the inherent characteristics of fetal ultrasound images at different semesters of pregnancy. In this paper, we proposed a new deep learning method for automatic fetal ultrasound image segmentation and HC biometry: deeply supervised attention-gated (DAG) V-Net, which incorporated the attention mechanism and deep supervision strategy into V-Net models. In addition, multi-scale loss function was introduced for deep supervision. The training set of the HC18 Challenge was expanded with data augmentation to train the DAG V-Net deep learning models. The trained models were used to automatically segment fetal head from two-dimensional ultrasound images, followed by morphological processing, edge detection, and ellipse fitting. The fitted ellipses were then used for HC biometric measurement. The proposed DAG V-Net method was evaluated on the testing set of HC18 (n = 355), in terms of four performance indices: Dice similarity coefficient (DSC), Hausdorff distance (HD), HC difference (DF), and HC absolute difference (ADF). Experimental results showed that DAG V-Net had a DSC of 97.93%, a DF of 0.09 ± 2.45 mm, an AD of 1.77 ± 1.69 mm, and an HD of 1.29 ± 0.79 mm. The proposed DAG V-Net method ranks fifth among the participants in the HC18 Challenge. By incorporating the attention mechanism and deep supervision, the proposed method yielded better segmentation performance than conventional U-Net and V-Net methods. Compared with published state-of-the-art methods, the proposed DAG V-Net had better or comparable segmentation performance. The proposed DAG V-Net may be used as a new method for fetal ultrasound image segmentation and HC biometry. The code of DAG V-Net will be made available publicly on https://github.com/xiaojinmao-code/ .

[Monitoring microwave ablation using ultrasound backscatter homodyned K imaging: Comparison of estimators].

The feasibility of ultrasound backscatter homodyned K model parametric imaging (termed homodyned K imaging) to monitor coagulation zone during microwave ablation was investigated. Two recent estimators for the homodyned K model parameter, RSK (the estimation method based on the signal-to-noise ratio, the skewness, and the kurtosis of the amplitude envelope of ultrasound) and XU (the estimation method based on the first moment of the intensity of ultrasound, X statistics and U statistics), were compared. Firstly, the ultrasound backscattered signals during the microwave ablation of porcine liver ex vivo were processed by the noise-assisted correlation algorithm, envelope detection, sliding window method, digital scan conversion and color mapping to obtain homodyned K imaging. Then 20 porcine livers' microwave ablation experiments ex vivo were used to evaluate the effect of homodyned K imaging in monitoring the coagulation zone. The results showed that the area under the receiver operating characteristic curve of the RSK method was 0.77 ± 0.06 (mean ± standard deviation), and that of the XU method was 0.83 ± 0.08 (mean ± standard deviation). The accuracy to monitor the coagulation zone was (86 ± 10)% (mean ± standard deviation) by the RSK method and (90 ± 8)% (mean ± standard deviation) by the XU method. Compared with the RSK method, the Bland-Altman consistency for the coagulation zone estimated by the XU method and that of actual porcine liver tissue was higher. The time for parameter estimation and imaging by the XU method was less than that by the RSK method. We conclude that ultrasound backscatter homodyned K imaging can be used to monitor coagulation zones during microwave ablation, and the XU method is better than the RSK method.

[Advances in Ultrasound Tissue Characterization and Its Application in Thermal Ablation of Tumors].

The methods of monitoring the thermal ablation of tumor are compared and analyzed in recent years. The principle method results and insufficient of ultrasound elastography and quantitative ultrasound imaging are discussed. The results show that ultrasonic tissue signature has great development space in the field of real-time monitoring of thermal ablation, but there are still some problems such as insufficient monitoring accuracy difficulty in whole-course monitoring and insufficient in vivo experiments, so it is impossible to realize clinical application. It is necessary to further study the monitoring method which can realize accurate and real-time detection of ablation zone and transition zone and can be easily combined with the existing ultrasonic equipment.

IRN-MLSQR: An improved iterative reweight norm approach to the inverse problem of electrocardiography incorporating factorization-free preconditioned LSQR.

The inverse problem of electrocardiography (ECG) of computing epicardial potentials from body surface potentials, is an ill-posed problem and needs to be solved by regularization techniques. The L2-norm regularization can cause considerable smoothing of the solution, while the L1-norm scheme promotes a solution with sharp boundaries/gradients between piecewise smooth regions, so L1-norm is widely used in the ECG inverse problem. However, large amount of computation and long computation time are needed in the L1-norm scheme. In this paper, by combining iterative reweight norm (IRN) with a factorization-free preconditioned LSQR algorithm (MLSQR), a new IRN-MLSQR method was proposed to accelerate the convergence speed of the L1-norm scheme. We validated the IRN-MLSQR method using experimental data from isolated canine hearts and clinical procedures in the electrophysiology laboratory. The results showed that the IRN-MLSQR method can significantly reduce the number of iterations and operation time while ensuring the calculation accuracy. The number of iterations of the IRN-MLSQR method is about 60%-70% that of the conventional IRN method, and at the same time, the accuracy of the solution is almost the same as that of the conventional IRN method. The proposed IRN-MLSQR method may be used as a new approach to the inverse problem of ECG.

Ultrasound Backscatter Envelope Statistics Parametric Imaging for Liver Fibrosis Characterization: A Review.

Early detection and diagnosis of liver fibrosis is of critical importance. Currently the gold standard for diagnosing liver fibrosis is biopsy. However, liver biopsy is invasive and associated with sampling errors and can lead to complications such as bleeding. Therefore, developing noninvasive imaging techniques for assessing liver fibrosis is of clinical value. Ultrasound has become the first-line tool for the management of chronic liver diseases. However, the commonly used B-mode ultrasound is qualitative and can cause interobserver or intraobserver difference. Ultrasound backscatter envelope statistics parametric imaging is an important group of quantitative ultrasound techniques that have been applied to characterizing different kinds of tissue. However, a state-of-the-art review of ultrasound backscatter envelope statistics parametric imaging for liver fibrosis characterization has not been conducted. In this paper, we focused on the development of ultrasound backscatter envelope statistics parametric imaging techniques for assessing liver fibrosis from 1998 to September 2019. We classified these techniques into six categories: constant false alarm rate, fiber structure extraction technique, acoustic structure quantification, quantile-quantile probability plot, the multi-Rayleigh model, and the Nakagami model. We presented the theoretical background and algorithms for liver fibrosis assessment by ultrasound backscatter envelope statistics parametric imaging. Then, the specific applications of ultrasound backscatter envelope statistics parametric imaging techniques to liver fibrosis evaluation were reviewed and analyzed. Finally, the pros and cons of each technique were discussed, and the future development was suggested.

A Wearable Electrocardiogram Telemonitoring System for Atrial Fibrillation Detection.

In this paper we proposed a wearable electrocardiogram (ECG) telemonitoring system for atrial fibrillation (AF) detection based on a smartphone and cloud computing. A wearable ECG patch was designed to collect ECG signals and send the signals to an Android smartphone via Bluetooth. An Android APP was developed to display the ECG waveforms in real time and transmit every 30 s ECG data to a remote cloud server. A machine learning (CatBoost)-based ECG classification method was proposed to detect AF in the cloud server. In case of detected AF, the cloud server pushed the ECG data and classification results to the web browser of a doctor. Finally, the Android APP displayed the doctor's diagnosis for the ECG signals. Experimental results showed the proposed CatBoost classifier trained with 17 selected features achieved an overall F1 score of 0.92 on the test set (n = 7,270). The proposed wearable ECG monitoring system may potentially be useful for long-term ECG telemonitoring for AF detection.

Conformal coverage of liver tumors by the thermal coagulation zone in 2450-MHz microwave ablation.

Purpose: To optimize treatment schemes using 2450-MHz microwave ablation (MWA), a novel conformal coverage method based on bipolar-angle mapping is proposed that determines whether a liver tumor is completely encompassed by thermal coagulation zones. Materials and methods: Firstly, three-dimensional (3-D) triangular mesh data of liver tumors were reconstructed from clinical computed tomography (CT) slices using the Marching cubes (MC) algorithm. Secondly, characterization models of thermal coagulation zones were established based on finite element simulation results of 40, 45, 50, 55, and 60 W ablations. Finally, coagulation zone models and tumor surface data were mapped and fused on a two-dimensional (2-D) plane to achieve conformal coverage of liver tumors by comparing the corresponding polar radii. Results: Optimal parameters for ablation treatment of liver tumors were efficiently obtained with the proposed conformal coverage method. Fifteen liver tumors were obtained with maximal diameters of 12.329-78.612 mm (mean ± standard deviation, 39.094 ± 19.447 mm). The insertion positions and orientations of the MWA antenna were determined based on 3-D reconstruction results of these tumors. The ablation patterns and durations of tumors were planned according to the minimum mean standard deviations between the ablative margin and tumor surface. Conclusion: The proposed method can be applied to computer-assisted MWA treatment planning of liver tumors, and is expected to guide clinical procedures in future.

A Fetal ECG Monitoring System Based on the Android Smartphone.

In this paper, a fetal electrocardiogram (ECG) monitoring system based on the Android smartphone was proposed. We designed a portable low-power fetal ECG collector, which collected maternal abdominal ECG signals in real time. The ECG data were sent to a smartphone client via Bluetooth. Smartphone app software was developed based on the Android system. The app integrated the fast fixed-point algorithm for independent component analysis (FastICA) and the sample entropy algorithm, for the sake of real-time extraction of fetal ECG signals from the maternal abdominal ECG signals. The fetal heart rate was computed using the extracted fetal ECG signals. Experimental results showed that the FastICA algorithm can extract a clear fetal ECG, and the sample entropy can correctly determine the channel where the fetal ECG is located. The proposed fetal ECG monitoring system may be feasible for non-invasive, real-time monitoring of fetal ECGs.

Monitoring Microwave Ablation Using Ultrasound Echo Decorrelation Imaging: An ex vivo Study.

In this study, a microwave-induced ablation zone (thermal lesion) monitoring method based on ultrasound echo decorrelation imaging was proposed. A total of 15 cases of ex vivo porcine liver microwave ablation (MWA) experiments were carried out. Ultrasound radiofrequency (RF) signals at different times during MWA were acquired using a commercial clinical ultrasound scanner with a 7.5-MHz linear-array transducer. Instantaneous and cumulative echo decorrelation images of two adjacent frames of RF data were calculated. Polynomial approximation images were obtained on the basis of the thresholded cumulative echo decorrelation images. Experimental results showed that the instantaneous echo decorrelation images outperformed conventional B-mode images in monitoring microwave-induced thermal lesions. Using gross pathology measurements as the reference standard, the estimation of thermal lesions using the polynomial approximation images yielded an average accuracy of 88.60%. We concluded that instantaneous ultrasound echo decorrelation imaging is capable of monitoring the change of thermal lesions during MWA, and cumulative ultrasound echo decorrelation imaging and polynomial approximation imaging are feasible for quantitatively depicting thermal lesions.

Three-dimensional Visualization of Ultrasound Backscatter Statistics by Window-modulated Compounding Nakagami Imaging.

In this study, the window-modulated compounding (WMC) technique was integrated into three-dimensional (3D) ultrasound Nakagami imaging for improving the spatial visualization of backscatter statistics. A 3D WMC Nakagami image was produced by summing and averaging a number of 3D Nakagami images (number of frames denoted as N) formed using sliding cubes with varying side lengths ranging from 1 to N times the transducer pulse. To evaluate the performance of the proposed 3D WMC Nakagami imaging method, agar phantoms with scatterer concentrations ranging from 2 to 64 scatterers/mm3 were made, and six stages of fatty liver (zero, one, two, four, six, and eight weeks) were induced in rats by methionine-choline-deficient diets (three rats for each stage, total n = 18). A mechanical scanning system with a 5-MHz focused single-element transducer was used for ultrasound radiofrequency data acquisition. The experimental results showed that 3D WMC Nakagami imaging was able to characterize different scatterer concentrations. Backscatter statistics were visualized with various numbers of frames; N = 5 reduced the estimation error of 3D WMC Nakagami imaging in visualizing the backscatter statistics. Compared with conventional 3D Nakagami imaging, 3D WMC Nakagami imaging improved the image smoothness without significant image resolution degradation, and it can thus be used for describing different stages of fatty liver in rats.

A Review of Ultrasound Tissue Characterization with Mean Scatterer Spacing.

Tissues exhibiting quasi-periodic structures can be modeled as a collection of diffuse scatterers and coherent scatterers. The mean scatterer spacing (MSS) of coherent and quasi-periodic components is directly related to tissue microstructure and has become an important quantitative ultrasound (QUS) parameter in the characterization of quasi-periodic tissues. In this paper, a review of the literature on the development of MSS as a QUS parameter was conducted. First, a unified theoretical background of MSS estimates was provided. Then, the application of MSS estimates was summarized with respect to liver, spleen, breast, bone, muscle, and other tissues. MSS estimation techniques were applied to (a) the diagnosis of hepatitis, liver fibrosis and cirrhosis, and lesions in tissues such as liver, breast, and spleen; (b) the differentiation between benign and malignant breast tumors, and the grading of breast cancer; (c) the detection of cancellous bone; and (d) the monitoring of the efficacy of treatments such as thermal ablation, with various levels of success. Future developments were also discussed in terms of real-time implementation of MSS estimates, local MSS estimation, relationship of MSS to other QUS parameters, combination of MSS with other QUS parameters, in vivo validation of MSS estimates, MSS parametric imaging, and three-dimensional ultrasound tissue characterization.

A Computer-Aided Diagnosis Scheme For Detection Of Fatty Liver In Vivo Based On Ultrasound Kurtosis Imaging.

Fatty liver disease is a common disease caused by alcoholism, obesity, and diabetes, resulting in triglyceride accumulation in hepatocytes. Kurtosis coefficient, a measure of the peakedness of the probability distribution, has been applied to the analysis of backscattered statistics for characterizing fatty liver. This study proposed ultrasound kurtosis imaging as a computer-aided diagnosis (CAD) method to visually and quantitatively stage the fatty liver. A total of 107 patients were recruited to participate in the experiments. The livers were scanned using a clinical ultrasound scanner with a 3.5-MHz curved transducer to acquire the raw ultrasound backscattered signals for kurtosis imaging. The kurtosis image was constructed using the sliding window technique. Experimental results showed that kurtosis imaging has the ability to visualize and quantify the variation of backscattered statistics caused by fatty infiltration. The kurtosis coefficient corresponding to liver parenchyma decreased from 5.41 ± 0.89 to 3.68 ± 0.12 with increasing the score of fatty liver from 0 (normal) to 3 (severe), indicating that fatty liver reduces the degree of peakedness of backscattered statistics. The best performance of kurtosis imaging was found when discriminating between normal and fatty livers with scores ≥1: the area under the curve (AUC) is 0.92 at a cutoff value of 4.36 (diagnostic accuracy =86.9 %, sensitivity =86.7 %, specificity =87.0 %). The current findings suggest that kurtosis imaging may be useful in designing CAD tools to assist in physicians in early detection of fatty liver.

Classification of Benign and Malignant Breast Tumors in Ultrasound Images with Posterior Acoustic Shadowing Using Half-Contour Features.

Posterior acoustic shadowing (PAS) can bias breast tumor segmentation and classification in ultrasound images. In this paper, half-contour features are proposed to classify benign and malignant breast tumors with PAS, considering the fact that the upper half of the tumor contour is less affected by PAS. Adaptive thresholding and disk expansion are employed to detect tumor contours. Based on the detected full contour, the upper half contour is extracted. For breast tumor classification, six quantitative feature parameters are analyzed for both full contours and half contours, including standard deviation of degree (SDD), which is proposed to describe tumor irregularity. Fifty clinical cases (40 with PAS and 10 without PAS) were used. Tumor circularity (TC) and SDD were both effective full- and half-contour parameters in classifying images without PAS. Half-contour TC [74 % accuracy, 72 % sensitivity, 76 % specificity, 0.78 area under the receiver operating characteristic curve (AUC), p > 0.05] significantly improved the classification of breast tumors with PAS compared to that with full-contour TC (54 % accuracy, 56 % sensitivity, 52 % specificity, 0.52 AUC, p > 0.05). Half-contour SDD (72 % accuracy, 76 % sensitivity, 68 % specificity, 0.81 AUC, p < 0.05) improved the classification of breast tumors with PAS compared to that with full-contour SDD (62 % accuracy, 80 % sensitivity, 44 % specificity, 0.61 AUC, p > 0.05). The proposed half-contour TC and SDD may be useful in classifying benign and malignant breast tumors in ultrasound images affected by PAS.

Semi-automatic breast ultrasound image segmentation based on mean shift and graph cuts.

Computerized tumor segmentation on breast ultrasound (BUS) images remains a challenging task. In this paper, we proposed a new method for semi-automatic tumor segmentation on BUS images using Gaussian filtering, histogram equalization, mean shift, and graph cuts. The only interaction required was to select two diagonal points to determine a region of interest (ROI) on an input image. The ROI image was shrunken by a factor of 2 using bicubic interpolation to reduce computation time. The shrunken image was smoothed by a Gaussian filter and then contrast-enhanced by histogram equalization. Next, the enhanced image was filtered by pyramid mean shift to improve homogeneity. The object and background seeds for graph cuts were automatically generated on the filtered image. Using these seeds, the filtered image was then segmented by graph cuts into a binary image containing the object and background. Finally, the binary image was expanded by a factor of 2 using bicubic interpolation, and the expanded image was processed by morphological opening and closing to refine the tumor contour. The method was implemented with OpenCV 2.4.3 and Visual Studio 2010 and tested for 38 BUS images with benign tumors and 31 BUS images with malignant tumors from different ultrasound scanners. Experimental results showed that our method had a true positive rate (TP) of 91.7%, a false positive (FP) rate of 11.9%, and a similarity (SI) rate of 85.6%. The mean run time on Intel Core 2.66 GHz CPU and 4 GB RAM was 0.49 ± 0.36 s. The experimental results indicate that the proposed method may be useful in BUS image segmentation.