William A. Sandham

MRI fuzzy segmentation of brain tissue using neighborhood attraction with neural-network optimization

Image segmentation is an indispensable process in the visualization of human tissues, particularly during clinical analysis of magnetic resonance (MR) images. Unfortunately, MR images always contain a significant amount of noise caused by operator performance, equipment, and the environment, which can lead to serious inaccuracies with segmentation. A robust segmentation technique based on an extension to the traditional fuzzy c-means (FCM) clustering algorithm is proposed in this paper. A neighborhood attraction, which is dependent on the relative location and features of neighboring pixels, is shown to improve the segmentation performance dramatically. The degree of attraction is optimized by a neural-network model. Simulated and real brain MR images with different noise levels are segmented to demonstrate the superiority of the proposed technique compared to other FCM-based methods. This segmentation method is a key component of an MR image-based classification system for brain tumors, currently being developed.

Ultrasound imaging in lower limb prosthetics

The biomechanical interaction between the residual limb and the prosthetic socket determines the quality of fit of the socket in lower limb prosthetics. An understanding of this interaction and the development of quantitative measures to predict the quality of fit of the socket are important for optimal socket design. Finite-element modeling is used widely for biomechanical modeling of the limb/socket interaction and requires information on the internal and external geometry of the residual limb. Volumetric imaging methods such as X-ray computed tomography, magnetic resonance imaging, and ultrasound have been used to obtain residual limb shape information. Of these modalities, ultrasound has been introduced most recently and its development for visualization in prosthetics is the least mature. This paper reviews ultrasound image acquisition and processing methods as they have been applied in lower limb prosthetics.

Geophysical applications of artificial neural networks and fuzzy logic

This book is the first major text to encompass the wide diversity of geophysical applications of artificial neural networks (ANNs) and fuzzy logic (FZ). Each chapter, written by internationally-renowned experts in their field, represents a specific geophysical application, ranging from first-break picking and trace editing encountered in seismic exploration, through well-log lithology determination, to electromagnetic exploration and earthquake seismology. The book offers a well-balanced division of contributions from industry and academia, and includes a comprehensive, up-to-date bibliography covering all major publications in geophysical applications of ANNs and FZ. A special feature of this volume is the preface written by Professor Fred Aminzadeh, eminent authority in the field of artificial intelligence and geophysics.

Blood glucose prediction using artificial neural networks trained with the AIDA diabetes simulator: a proof-of-concept pilot study

Diabetes mellitus is a major, and increasing, global problem. However, it has been shown that, through good management of blood glucose levels (BGLs), the associated and costly complications can be reduced significantly. In this pilot study, Elman recurrent artificial neural networks (ANNs) were used to make BGL predictions based on a history of BGLs, meal intake, and insulin injections. Twenty-eight datasets (from a single case scenario) were compiled from the freeware mathematical diabetes simulator, AIDA. It was found that the most accurate predictions were made during the nocturnal period of the 24 hour daily cycle. The accuracy of the nocturnal predictions was measured as the root mean square error over five test days (RMSE5day) not used during ANN training. For BGL predictions of up to 1 hour a RMSE5 day of (±SD) 0.15±0.04 mmol/L was observed. For BGL predictions up to 10 hours, a RMSE5 day of (±SD) 0.14±0.16 mmol/L was observed. Future research will investigate a wider range of AIDA case scenarios, real-patient data, and data relating to other factors influencing BGLs. ANN paradigms based on real-time recurrent learning will also be explored to accommodate dynamic physiology in diabetes.

Neural network and neuro-fuzzy systems for improving diabetes therapy

Expert management of diabetes mellitus, through good glycaemic control, is necessary development of serious short-term complications, due to the persistence of either low or high blood glucose levels (BGLs), respectively. In this paper, the use of a recurrent artificial neural network (ANN) is described which is able to predict BGL for a specific patient. This predicted BGL may then be used in a neuro-fuzzy expert system to offer short-term therapeutic advice regarding the patients diet, exercise and insulin regime (for insulin-dependent or Type 1 diabetics). ANN training requirements are discussed, and BGL predictions for two Type 1 diabetic patients are compared with actual BGL measurements.

ANN compression of morphologically similar ECG complexes.

A compression algorithm for electrocardiogram signals is presented, based on an auto-associative neural network. Issues of weight and activation coding are considered, and compression performances of various network sizes are compared. A unique feature is the performance improvement achieved using DC level removal. A comparison with existing techniques is provided.

Preprocessing and segmentation of brain magnetic resonance images

This paper describes a process for improving the segmentation of brain magnetic resonance (MR) images. It involves two stages; preprocessing and segmentation. During preprocessing, the image intensities are first standardized using the pixel histograms. Morphological processing is then used to remove the non-brain regions. During the segmentation process, normal and abnormal brain tissues are segmented using both the traditional fuzzy c-means (FCM) clustering algorithm, and a new improved FCM algorithm. Neighborhood effects are considered in the latter method to overcome noise. Segmentation results show that this method is more robust to noise and can improve the integrity of the segmentation performance.

Fuzzy clustering based applications to medical image segmentation

Medical image segmentation is an indispensable process in the visualization of human tissues. However, medical images always contain a large amount of noise caused by operator performance, equipment and environment. This leads to inaccuracy with segmentation. A robust segmentation technique is required. In this paper, based on the traditional fuzzy c-means (FCM) clustering algorithm, the neighborhood attraction is shown to improve the segmentation performance. Two factors of the neighborhood attraction depend on relative location and features of neighboring pixels in the image. Simulated and real brain magnetic resonance (MR) images are segmented to demonstrate the superiority of the proposed method compared to the conventional FCM method.

A new approach to brain tumour diagnosis using fuzzy logic based genetic programming

Brain tumour diagnosis generally requires a histological analysis, involving invasive surgery which can cause pain and discomfort to patients. In this paper, a new brain tumour diagnostic procedure is described using magnetic-resonance imaging (MRI) only. First, the MR images are preprocessed, using standardizing, non-brain removal and enhancement. Second, an improved fuzzy clustering algorithm is applied to segment the brain into different tissues. Finally, brain tumour diagnosis is performed using fuzzy logic based genetic programming (GP) to search for classification rules. Classification results on a variety of MR images for different pathologies, indicate this technique to be promising.

3-D seismic tracking with probabilistic data association

In this paper, a new approach to the problem of tracking a seismic event through a 3-D data set is presented. The method under consideration was originally developed for tracking targets in a cluttered environment and uses Probabilistic Data Association (PDA) to assess the probability of each return being the correct return. This theory has been modified for use in seismic event tracking, which unlike target tracking, is a static problem, and this new approach has been tested on both real and synthetic 3-D data sets. The tracker successfully picks out the chosen horizon in both the synthetic and real 3-D data sets. The accuracy of the 3-D tracker may be improved by tracking through the data set in two perpendicular directions and correlating the results. Results show that it is also possible to include a diagonal track in the correlation.