S. S. Mohamed

Computer-aided diagnosis for prostate cancer using support vector machine

The work in this paper aims for analyzing texture features of the prostate using Trans-Rectal Ultra-Sound images (TRUS) images for tissue characterization. This research is expected to assist beginner radiologists with the decision making. Moreover it will also assist in determining the biopsy locations. Texture feature analysis is composed of four stages. The first stage is automatically identifying Regions Of Interest (ROI), a step that was usually done either by an expert radiologist or by dividing the whole image into smaller squares that represent regions of interest. The second stage is extracting the statistical features from the identified ROIs. Two different statistical feature sets were used in this study; the first is Grey Level Dependence Matrix features. The second feature set is Grey level difference vector features. These constructed features are then ranked using Mutual Information (MI) feature selection algorithm that maximizes MI between feature and class. The obtained feature sets, the combined feature set as well as the reduced feature subset were examined using Support Vector Machine (SVM) classifier, a well established classifier that is suitable for noisy data such as those obtained from the ultrasound images. The obtained sensitivity is 83.3%, specificity ranges from 90% to 100% and accuracy ranges from 87.5% to 93.75%.

Artificial life feature selection techniques for prostrate cancer diagnosis using TRUS images

This paper presents two novel feature selection techniques for the purpose of prostate tissue characterization based on Trans-rectal Ultrasound (TRUS) images. First, suspected cancerous regions of interest (ROIs) are identified from the segmented TRUS images using Gabor filters. Next, second and higher order statistical texture features are constructed for these ROIs. Furthermore, a representative feature subset with the best discriminatory power among the constructed features is selected using two artificial life techniques: the Particle Swarm Optimization (PSO) and the Ant Colony Optimization (ACO). Both the PSO and ACO are tailored to fit the binary nature of the feature selection problem. The results are compared to the results obtained using the Genetic Algorithm (GA) feature selection approach. When Support Vector Machine (SVM) classifier is applied for the purpose of tissue characterization, the features obtained using the PSO and ACO outperforms the features obtained using the GA, i.e., they are capable of discriminating between suspicious cancerous and non-cancerous in a better accuracy. The obtained results demonstrate excellent tissue characterization with 83.3% sensitivity, 100% specificity and 94% overall accuracy.

Prostate tissue texture feature extraction for cancer recognition in trus images using wavelet decomposition

In this paper, a wavelet based approach is proposed for the detection and diagnosis of prostate cancer in Trans Rectal UltraSound (TRUS) images. A texture feature extraction filter was implemented to extract textural features from TRUS images that characterize malignant and benign tissues. The filter is based on the wavelet decomposition. It is demonstrated that the wavelet decomposition reveals details in the malignant and benign regions in TRUS images of the prostate which correlate with their pathological representations. The wavelet decomposition is applied to enhance the visual distinction between the malignant and benign regions, which could be used by radiologists as a supplementary tool for making manual classification decisions. The proposed filter could be used to extract texture features which linearly separate the malignant and benign regions in the feature domain. The extracted feature could be used as an input to a complex classifier for automated malignancy region classification.

Prostate tissue characterization using TRUS image spectral features

In this paper focuses on extracting and analyzing spectral features from Trans-Rectal Ultra-Sound (TRUS) images for prostate tissue characterization. The information of the images’ frequency domain features and spatial domain features are used to achieve an accurate Region of Interest (ROI) identification. In particular, each image is divided into ROIs by the use of Gabor filters, a crucial stage, where the image is segmented according to the frequency response of the image pixels. Further, pixels with a similar response to the same filter are assigned to the same region to form a ROI. The radiologist’s experience is also integrated into the algorithm to identify the highly suspected ROIs. Next, for each ROI, different spectral feature sets are constructed. One set includes the power spectrum wedge and ring energies. The other sets are constructed using geometrical features extracted from the Power Spectrum Density (PSD). In particular, the estimated PSD in these sets is divided into two segments. Polynomial interpolation is used for the first segment and the obtained polynomial coefficients are used as features. The second segment is approximated by a straight line and the slope, the Y intercept as well as the first maximum reached by the PSD are considered as features. A classifier-based feature selection algorithm using CLONALG, a recently proposed optimization technique developed on the basis of clonal selection of the Artificial Immune System (AIS), is adopted and used to select an optimal subset from the above extracted features. Using different PSD estimation techniques, the obtained accuracy ranges from 72.2% to 93.75% using a Support Vector Machine classifier.

Prostate Tissue Texture Feature Extraction for Suspicious Regions Identification on TRUS Images

In this work, two different approaches are proposed for region of interest (ROI) segmentation using transrectal ultrasound (TRUS) images. The two methods aim to extract informative features that are able to characterize suspicious regions in the TRUS images. Both proposed methods are based on multi-resolution analysis that is characterized by its high localization in both the frequency and the spatial domains. Being highly localized in both domains, the proposed methods are expected to accurately identify the suspicious ROIs. On one hand, the first method depends on a Gabor filter that captures the high frequency changes in the image regions. On the other hand, the second method depends on classifying the wavelet coefficients of the image. It is shown in this paper that both methods reveal details in the ROIs which correlate with their pathological representations. It was found that there is a good match between the regions identified using the two methods, a result that supports the ability of each of the proposed methods to mimic the radiologist’s decision in identifying suspicious regions. Studying two ROI segmentation methods is important since the only available dataset is the radiologist’s suspicious regions, and there is a need to support the results obtained by either one of the proposed methods. This work is mainly a preliminary proof of concept study that will ultimately be expanded to a larger scale study whose aim will be introducing an assisting tool to help the radiologist identify the suspicious regions.

Region of interest identification in prostate TRUS images based on Gabor filter

This paper presents a new algorithm for prostate texture classification based on transrectal ultrasound (TRUS) images. A Gabor filter is designed to automatically identify the regions of interest (ROI) in the image. Furthermore, texture analysis for these regions is carried out by employing grey level co-occurrence matrix GLCM. Contrast feature is found to be useful for the differentiation between cancerous and non-cancerous tissues. The obtained results demonstrate that the contrast level in normal tissue is higher than that of cancerous tissue

Mass spectrometry-based proteomic pattern analysis for prostate cancer detection using neural networks with statistical significance test-based feature selection

Mass spectrometry-based proteomics provides a promising approach for accurate diagnosis of different diseases. However, there are some problems in the mass spectral data such as huge volume, data complexity and the presence of noise. These problems make analyzing the proteomic pattern difficult. In this paper, a neural network-based system is proposed for proteomic pattern analysis for prostate cancer screening. The system consists of three stages: feature selection based on statistical significant test, classification by a Radial Basis Function Neural Network (RBFNN) and a probabilistic neural network (PNN), and finally results optimization through ROC analysis. The experimental results show that the proposed systems performance is excellent in comparison with the existing tools. The high sensitivity (97.1%) and specificity (96.8%) of the proposed system when combined with prostatic biopsy are expected to help in early detection of prostate cancer.

ANN-based technique for fault location estimation using TLS-ESPRIT

In this paper, a methodology is proposed for identifying the fault location in transmission lines. The required features for the proposed algorithm is extracted from transient currents or voltages waveforms measured at the substation using the total least square-estimation of signal parameters via rotational invariance technique (TLS-ESPRIT). Since these transient waveforms are considered as a summation of damped sinusoids, TLS-ESPRIT is used to estimate different signal parameters mainly damping factors, frequencies and amplitudes of different modes contained in the signal. These parameters are functions of the fault location, system damping and fault time. Those features can then be employed for fault location identification. Artificial neural networks (ANNs) are used to estimate the fault location using this modal information. Training and testing data are generated using PSCAD/EMTDC simulations. It is crucial to indicate that no pre-fault data is required in this algorithm.

An automated neural-fuzzy approach to malignant tumor localization in 2D ultrasonic images of the prostate.

In this paper, a new neural-fuzzy approach is proposed for automated region segmentation in transrectal ultrasound images of the prostate. The goal of region segmentation is to identify suspicious regions in the prostate in order to provide decision support for the diagnosis of prostate cancer. The new automated region segmentation system uses expert knowledge as well as both textural and spatial features in the image to accomplish the segmentation. The textural information is extracted by two recurrent random pulsed neural networks trained by two sets of data (a suspicious tissues’ data set and a normal tissues’ data set). Spatial information is captured by the atlas-based reference approach and is represented as fuzzy membership functions. The textural and spatial features are synthesized by a fuzzy inference system, which provides a binary classification of the region to be evaluated.

Prostate TRUS Image Region-Based Feature Extraction and Evaluation

In this work a new informative feature set is proposed to identify suspicious Regions Of Interest (ROIs) in the prostate TransRectal UltraSound (TRUS) images. The proposed features are region based to overcome the limitations present in the pixel based feature extraction methods. First a thresholding algorithm integrated with the medical information is used to identify different candidate ROIs. Next, image registration is performed to transform the prostate image to a model based from which some of the proposed region based features are extracted. Subsequently, the proposed raw based and model based region features are extracted at the region level. Finally Mutual Information is used to evaluate the extracted features and compare their information content with both the typical pixel based features and the well known texture and grey level features. It was found that the proposed features provide more information than both the texture features and the pixel based features.