Ahmed M. Badawi

Classification algorithms for quantitative tissue characterization of diffuse liver disease from ultrasound images

Visual criteria for diagnosing diffused liver diseases from ultrasound images can be assisted by computerized tissue classification. Feature extraction algorithms are proposed in this paper to extract the tissue characterization parameters from liver images. The resulting parameter set is further processed to obtain the minimum number of parameters which represent the most discriminating pattern space for classification. This preprocessing step has been applied to over 120 distinct pathology-investigated cases to obtain the learning data for classification. The extracted features are divided into independent training and test sets, and are used to develop and compare both statistical and neural classifiers. The optimal criteria for these classifiers are set to have minimum classification error, ease of implementation and learning, and the flexibility for future modifications. Various algorithms of classification based on statistical and neural network methods are presented and tested. The authors show that very good diagnostic rates can be obtained using unconventional classifiers trained on actual patient data.

Fuzzy logic algorithm for quantitative tissue characterization of diffuse liver diseases from ultrasound images

Computerized ultrasound tissue characterization has become an objective means for diagnosis of liver diseases. It is difficult to differentiate diffuse liver diseases, namely cirrhotic and fatty liver by visual inspection from the ultrasound images. The visual criteria for differentiating diffused diseases are rather confusing and highly dependent upon the sonographers experience. This often causes a bias effects in the diagnostic procedure and limits its objectivity and reproducibility. Computerized tissue characterization to assist quantitatively the sonographer for the accurate differentiation and to minimize the degree of risk is thus justified. Fuzzy logic has emerged as one of the most active area in classification. In this paper, we present an approach that employs Fuzzy reasoning techniques to automatically differentiate diffuse liver diseases using numerical quantitative features measured from the ultrasound images. Fuzzy rules were generated from over 140 cases consisting of normal, fatty, and cirrhotic livers. The input to the fuzzy system is an eight dimensional vector of feature values: the mean gray level (MGL), the percentile 10%, the contrast (CON), the angular second moment (ASM), the entropy (ENT), the correlation (COR), the attenuation (ATTEN) and the speckle separation. The output of the fuzzy system is one of the three categories: cirrhosis, fatty or normal. The steps done for differentiating the pathologies are data acquisition and feature extraction, dividing the input spaces of the measured quantitative data into fuzzy sets. Based on the expert knowledge, the fuzzy rules are generated and applied using the fuzzy inference procedures to determine the pathology. Different membership functions are developed for the input spaces. This approach has resulted in very good sensitivities and specificity for classifying diffused liver pathologies. This classification technique can be used in the diagnostic process, together with the history information, laboratory, clinical and pathological examinations.

A Multimodal Hand Vein, Hand Geometry, and Fingerprint Prototype Design for High Security Biometrics

Prior research evidenced that the unimodal biometric systems have several tradeoffs like noisy data, intra-class variations, restricted degrees of freedom, non-universality, spoof attacks, and unacceptable error rates. In order for the biometric system to be more secure and to provide higher accuracy, more than one form of biometrics are required. Hence, the need arise for multimodal biometrics using combinations of different biometric modalities. We describe the design and development of whole hands biometrics prototype system that acquires left and right (L/R) index and ring fingerprints (FP), L/R near-infra-red (NIR) dorsal hand vein (HV) patterns, and L/R NIR dorsal hand geometry (HG) shape. Large database of 500-1000 subjects for whole hands is planned for data collection. The acquired sample images were found to have good quality for all features and patterns extraction to all modalities. The designed prototype can be considered for authentication and identification purposes. Advantages of this system over few existing multimodal systems are its being very hard to spoof attacks on the sensory level and the NIR HV and NIR HG thermal images are good signals for liveness detection.

Localization Accuracy of the Clinical Target Volume During Image-Guided Radiotherapy of Lung Cancer

PURPOSE To evaluate the position and shape of the originally defined clinical target volume (CTV) over the treatment course, and to assess the impact of gross tumor volume (GTV)-based online computed tomography (CT) guidance on CTV localization accuracy. METHODS AND MATERIALS Weekly breath-hold CT scans were acquired in 17 patients undergoing radiotherapy. Deformable registration was used to propagate the GTV and CTV from the first weekly CT image to all other weekly CT images. The on-treatment CT scans were registered rigidly to the planning CT scan based on the GTV location to simulate online guidance, and residual error in the CTV centroids and borders was calculated. RESULTS The mean GTV after 5 weeks relative to volume at the beginning of treatment was 77% ± 20%, whereas for the prescribed CTV, it was 92% ± 10%. The mean absolute residual error magnitude in the CTV centroid position after a GTV-based localization was 2.9 ± 3.0 mm, and it varied from 0.3 to 20.0 mm over all patients. Residual error of the CTV centroid was associated with GTV regression and anisotropy of regression during treatment (p = 0.02 and p = 0.03, respectively; Spearman rank correlation). A residual error in CTV border position greater than 2 mm was present in 77% of patients and 50% of fractions. Among these fractions, residual error of the CTV borders was 3.5 ± 1.6 mm (left-right), 3.1 ± 0.9 mm (anterior-posterior), and 6.4 ± 7.5 mm (superior-inferior). CONCLUSIONS Online guidance based on the visible GTV produces substantial error in CTV localization, particularly for highly regressing tumors. The results of this study will be useful in designing margins for CTV localization or for developing new online CTV localization strategies.

Optimizing principal component models for representing interfraction variation in lung cancer radiotherapy.

PURPOSE To optimize modeling of interfractional anatomical variation during active breath-hold radiotherapy in lung cancer using principal component analysis (PCA). METHODS In 12 patients analyzed, weekly CT sessions consisting of three repeat intrafraction scans were acquired with active breathing control at the end of normal inspiration. The gross tumor volume (GTV) and lungs were delineated and reviewed on the first week image by physicians and propagated to all other images using deformable image registration. PCA was used to model the target and lung variability during treatment. Four PCA models were generated for each specific patient: (1) Individual models for the GTV and each lung from one image per week (week to week, W2W); (2) a W2W composite model of all structures; (3) individual models using all images (weekly plus repeat intrafraction images, allscans); and (4) composite model with all images. Models were reconstructed retrospectively (using all available images acquired) and prospectively (using only data acquired up to a time point during treatment). Dominant modes representing at least 95% of the total variability were used to reconstruct the observed anatomy. Residual reconstruction error between the model-reconstructed and observed anatomy was calculated to compare the accuracy of the models. RESULTS An average of 3.4 and 4.9 modes was required for the allscans models, for the GTV and composite models, respectively. The W2W model required one less mode in 40% of the patients. For the retrospective composite W2W model, the average reconstruction error was 0.7 +/- 0.2 mm, which increased to 1.1 +/- 0.5 mm when the allscans model was used. Individual and composite models did not have significantly different errors (p = 0.15, paired t-test). The average reconstruction error for the prospective models of the GTV stabilized after four measurements at 1.2 +/- 0.5 mm and for the composite model after five measurements at 0.8 +/- 0.4 mm. CONCLUSIONS Retrospective PCA models were capable of reconstructing original GTV and lung shapes and positions within several millimeters with three to four dominant modes, on average. Prospective models achieved similar accuracy after four to five measurements.

Chromosomes classification based on neural networks, fuzzy rule based, and template matching classifiers

A new features extraction algorithms for G-banded chromosomes classification system based on neural networks, fuzzy rule based and template matching classifiers are proposed. Chromosomes image is acquired and processed, geometrical features and gray-scale features are extracted for 872 chromosomes. Neural networks, fuzzy rule based, template matching classifiers results were compared. Classification rates are found to be over 99% for training and over 96% for testing sets

Speckle Reduction in Medical Ultrasound: A Novel Scatterer Density Weighted Nonlinear Diffusion Algorithm Implemented as a Neural-Network Filter

This paper proposes a novel algorithm for speckle reduction in medical ultrasound imaging while preserving the edges with the added advantages of adaptive noise filtering and speed. We propose a nonlinear image diffusion algorithm that incorporates two local parameters of image quality, namely, scatterer density and texture-based contrast in addition to gradient, to weight the nonlinear diffusion process. The scatterer density is proposed to replace the existing traditional measures of quality of the ultrasound diffusion process such as MSE, RMSE, SNR, and PSNR. This novel diffusion filter was then implemented using backpropagation neural network for fast parallel processing of volumetric images. The experimental results show that weighting the image diffusion with these parameters produces better noise reduction and produces a better edge detection quality with reasonable computational cost. The proposed filter can be used as a preprocessing phase before applying any ultrasound segmentation or active contour model processes

3D Statistical Shape Models of Patella for Sex Classification

This paper proposes a new sex classification method from patellae using a novel automated feature extraction technique. A dataset of 228 patellae (95 females and 133 males) was collected and CT scanned. After the CT data was segmented, a set of features was automatically extracted, normalized, and ranked. These features include geometric features, moments, principal axes, and principal components. A feature vector of 45 dimensions for each subject was then constructed. A set of statistical and supervised neural network classification methods were used to classify the patellar feature vectors according to sex. Different classification methods were compared. Classification success ranged from 83.77% average classification rate with labeling using fuzzy C-means method (FCM), to 90.3% for linear discriminant function (LDF) analysis. We obtained results of 96.02% and 93.51% training and testing classification rates (respectively) using feedforward backpropagation neural networks (NN). These promising results encourage the usage of this method in forensic anthropology for identifying the sex from incomplete skeletons containing at least one patella

Fuzzy similarity measures for ultrasound tissue characterization

Computerized ultrasound tissue characterization has become an objective means for diagnosis of diseases. It is difficult to differentiate diffuse liver diseases, namely cirrhotic and fatty liver from a normal one, by visual inspection from the ultrasound images. The visual criteria for differentiating diffused diseases is rather confusing and highly dependent upon the sonographers experience. The need for computerized tissue characterization is thus justified to quantitatively assist the sonographer for accurate differentiation and to minimize the degree of risk from erroneous interpretation. In this paper we used the fuzzy similarity measure as an approximate reasoning technique to find the maximum degree of matching between an unknown case defined by a feature vector and a family of prototypes (knowledge base). The feature vector used for the matching process contains 8 quantitative parameters (textural, acoustical, and speckle parameters) extracted from the ultrasound image. The steps done to match an unknown case with the family of prototypes (cirr, fatty, normal) are: Choosing the membership functions for each parameter, then obtaining the fuzzification matrix for the unknown case and the family of prototypes, then by the linguistic evaluation of two fuzzy quantities we obtain the similarity matrix, then by a simple aggregation method and the fuzzy integrals we obtain the degree of similarity. Finally, we find that the similarity measure results are comparable to the neural network classification techniques and it can be used in medical diagnosis to determine the pathology of the liver and to monitor the extent of the disease.

Automatic tissue characterization from ultrasound imagery

In this work, feature extraction algorithms are proposed to extract the tissue characterization parameters from liver images. Then the resulting parameter set is further processed to obtain the minimum number of parameters representing the most discriminating pattern space for classification. This preprocessing step was applied to over 120 pathology-investigated cases to obtain the learning data for designing the classifier. The extracted features are divided into independent training and test sets and are used to construct both statistical and neural classifiers. The optimal criteria for these classifiers are set to have minimum error, ease of implementation and learning, and the flexibility for future modifications. Various algorithms for implementing various classification techniques are presented and tested on the data. The best performance was obtained using a single layer tensor model functional link network. Also, the voting k-nearest neighbor classifier provided comparably good diagnostic rates.