Search

EEG signal classification is an important task to build an accurate Brain Computer Interface (BCI) system. Many machine learning and deep learning approaches have been used to classify EEG signals. Besides, many studies have involved the time and frequency domain features to classify EEG signals. On the other hand, a very limited number of studies combine the spatial and temporal dimensions of the EEG signal. Brain dynamics are very complex across different mental tasks, thus it is difficult to design efficient algorithms with features based on prior knowledge. Therefore, in this study, we utilized the 2D AlexNet Convolutional Neural Network (CNN) to learn EEG features across different mental tasks without prior knowledge. First, this study adds spatial and temporal dimensions of EEG signals to a 2D EEG topographic map. Second, topographic maps at different time indices were cascaded to populate a 2D image for a given time window. Finally, the topographic maps enabled the AlexNet to learn features from the spatial and temporal dimensions of the brain signals. The classification performance was obtained by the proposed method on a multiclass dataset from BCI Competition IV dataset 2a. The proposed system obtained an average classification accuracy of 81.09%, outperforming the previous state-of-the-art methods by a margin of 4% for the same dataset. The results showed that converting the EEG classification problem from a (1D) time series to a (2D) image classification problem improves the classification accuracy for BCI systems. Also, our EEG topographic maps enabled CNN to learn subtle features from spatial and temporal dimensions, which better represent mental tasks than individual time or frequency domain features.

Machine learning algorithms are currently being implemented in an escalating manner to classify and/or predict the onset of some neurodegenerative diseases; including Alzheimer's Disease (AD); this could be attributed to the fact of the abundance of data and powerful computers. The objective of this work was to deliver a robust classification system for AD and Mild Cognitive Impairment (MCI) against healthy controls (HC) in a low-cost network in terms of shallow architecture and processing. In this study, the dataset included was downloaded from the Alzheimer's disease neuroimaging initiative (ADNI). The classification methodology implemented was the convolutional neural network (CNN), where the diffusion maps, and gray-matter (GM) volumes were the input images. The number of scans included was 185, 106, and 115 for HC, MCI and AD respectively. Ten-fold cross-validation scheme was adopted and the stacked mean diffusivity (MD) and GM volume produced an AUC of 0.94 and 0.84, an accuracy of 93.5% and 79.6%, a sensitivity of 92.5% and 62.7%, and a specificity of 93.9% and 89% for AD/HC and MCI/HC classification respectively. This work elucidates the impact of incorporating data from different imaging modalities; i.e. structural Magnetic Resonance Imaging (MRI) and Diffusion Tensor Imaging (DTI), where deep learning was employed for the aim of classification. To the best of our knowledge, this is the first study assessing the impact of having more than one scan per subject and propose the proper maneuver to confirm the robustness of the system. The results were competitive among the existing literature, which paves the way for improving medications that could slow down the progress of the AD or prevent it.

Stroke is one of the leading causes of paralysis. It is defined by being the sudden death of some brain cells due to lack of oxygen when blood flow to the brain is lost. Stroke leaves a large number of people with cognitive, affective and motor impairments depending on assistance in their daily lives. Many studies have proved that motor imagery (MI) based Brain Computer Interface (BCI) is an effective technique for post-stroke rehabilitation. It requires a real time response with high achieved accuracy. In this paper, we present a real time rehabilitation system for post-stroke patients. The system consists of a Virtual Reality (VR) 3D game controlled by a BCI system. We utilized various schemes of pre-processing, features extraction, feature selectors and classifiers to ameliorate the performance of the system for dataset IV 2a [Four class motor imagery (001–2014) – Institute for Knowledge Discovery, Graz University of Technology]. We achieved a maximum accuracy of 79% for the signal processing part using CSP feature extractor and SVM classifier with signal processing time of 0.02 seconds and 0.6 seconds for the response of whole system as a current status.

Motor imagery patterns are extensively exploited in brain-computer interface systems in order to control outer devices without using peripheral nerves or muscles. Classification of these patterns can be based on the associated electroencephalogram (EEG) signals. Recent approaches addressed this classification problem through techniques exploiting mainly information from one or two EEG channels. However, these approaches overlook correlations between multiple EEG channels. In this paper, we create motor-imagery classification systems based on graph-theoretic models of multichannel EEG signals. In particular, multivariate autoregressive models are used to establish the relations between the EEG channels and construct directed graph signals. Also, we constructed undirected graph signal models with Gaussian-weighted distances between graph nodes. Then, a novel variant of the graph Fourier transform is applied to the directed and undirected graph models with and without edge weights. Distinctive features were thus extracted from the transform coefficients. Additional features were computed using common spatial patterns, polynomial representations and principal components of EEG signals. Significant performance improvements were achieved using extreme learning machine (ELM) classifiers. For Dataset Ia of the BCI Competition 2003, our approach led to a classification accuracy of 96.58% with fully-connected weighted directed graph features computed on the delta-band EEG signals. For the six subjects of the Dataset 1 of the BCI Competition IV, our approach compared well with other state-of-the-art methods in the alpha and beta EEG bands.

Computer-aided detection of malignant breast tumors in ultrasound images has been receiving growing attention. In this paper, we propose a deep learning methodology to tackle this problem. The training data, which contains several hundred images of benign and malignant cases, was used to train a deep convolutional neural network (CNN). Three training approaches are proposed: a baseline approach where the CNN architecture is trained from scratch, a transfer-learning approach where the pre-trained VGG16 CNN architecture is further trained with the ultrasound images, and a fine-tuned learning approach where the deep learning parameters are fine-tuned to overcome overfitting. The experimental results demonstrate that the fine-tuned model had the best performance (0.97 accuracy, 0.98 AUC), with pre-training on US images. Creating pre-trained models using medical imaging data would certainly improve deep learning outcomes in biomedical applications.

Elucidating protein subcellular localization is an essential topic in proteomics research due to its importance in the process of drug discovery. Unfortunately, experimentally uncovering protein subcellular targets is an arduous process that may not result in a successful localization. In contrast, computational methods can rapidly predict protein subcellular targets and are an efficient alternative to experimental methods for unannotated proteins. In this work, we introduce a new method to predict protein subcellular localization which increases the predictive power of generative probabilistic models while preserving their explanatory benefit. Our method exploits Markov models to produce a feature vector that records micro-similarities between the underlying probability distributions of a given sequence and their counterparts in reference models. Compared to ordinary Markov chain inference, we show that our method improves overall accuracy by 10% under 10-fold cross-validation on a dataset consisting of 10 subcellular locations. The source code is publicly available on https://github.com/aametwally/MC MicroSimilarities.

Medical equipment management raises a range of complex problems including those associated with maintenance process. In developing countries, hospitals rarely implement a coherent management plan in medical equipment management. One of the most significant challenges is to distinguish medical equipment that requires repair from those require replacement. Thus, a Decision Support System (DSS) is required to manage this pitfall properly. In this paper, a multi-criteria decision making model, Analytic Hierarchy Process (AHP) is presented to select an optimum maintenance strategy. With reference to literature review and experts' opinions; a set of criteria is employed to calculate a criticality score for each piece of equipment. Therefore, a list of equipment is ranked based on their scores and an optimum threshold is selected to differentiate between maintenance and replacement requirements. Fifty different types of medical equipment located in multiple public hospitals have been used in the validation of the proposed model. Results show that the proposed model can efficiently differentiate the equipment that requires repair and the others that needs to be scrapped.

We present a CAD scheme using DBN unsupervised path followed by NN supervised path.Our two-phase method 'DBN-NN' classification accuracy is higher than using one phase.Overall accuracy of DBN-NN reaches 99.68% with 100% sensitivity & 99.47% specificity.DBN-NN was tested on the Wisconsin Breast Cancer Dataset (WBCD).DBN-NN results show classifier performance improvements over previous studies. Over the last decade, the ever increasing world-wide demand for early detection of breast cancer at many screening sites and hospitals has resulted in the need of new research avenues. According to the World Health Organization (WHO), an early detection of cancer greatly increases the chances of taking the right decision on a successful treatment plan. The Computer-Aided Diagnosis (CAD) systems are applied widely in the detection and differential diagnosis of many different kinds of abnormalities. Therefore, improving the accuracy of a CAD system has become one of the major research areas. In this paper, a CAD scheme for detection of breast cancer has been developed using deep belief network unsupervised path followed by back propagation supervised path. The construction is back-propagation neural network with Liebenberg Marquardt learning function while weights are initialized from the deep belief network path (DBN-NN). Our technique was tested on the Wisconsin Breast Cancer Dataset (WBCD). The classifier complex gives an accuracy of 99.68% indicating promising results over previously-published studies. The proposed system provides an effective classification model for breast cancer. In addition, we examined the architecture at several train-test partitions.

Brain-computer interface (BCI) research is speedily growing and numerous innovative techniques are proposed for implementing BCI. One of the major drawbacks in BCI's application is the problem of searching out a response from one trial. Hence, many trials are performed for every element so as to decrease the error in prediction. This led to a long time before accurately predicting the user intent and intensive user training is needed. The objective of this paper is to investigate a new technique to process the signal from brain in a real time without any prior training. The new approach is applied to experimental data for motor imagery-based BCI and is compared to the classification results of the same data using the conventional processing techniques requiring prior calibration. Different classification methodologies were used as in time and frequency domain. It is concluded that wavelet transform get best performance reaches 82.14%. Therefore, these promising results recommend that this approach can reach accuracies not extremely far from those got with training.

In the past recognition of P300-based Brain Computer Interface (BCI), depends on the experience increased through adjustment or training sessions before usage by applying supervised training sets to construct the classification model. In this paper, a new technique has been introduced for BCI data that requires no earlier training. The fundamental rule of this unsupervised technique is that the trial with true activation signal inside every block must be distinctive from whatever remains of the trials inside that block. Consequently, a measure that is delicate to this difference can be utilized to settle on a choice taking into account a single block with no earlier training. In addition, different algorithms of aggregating data from many trials have been applied to extend communication speed and bit rate. Aggregation strategies include simple average, Principle Component Analysis (PCA) and Probabilistic Principle Component Analysis (PPCA). This new technique is applied to experimental data for P300-based BCI for both healthy and disabled subjects. By comparing our results to the classification output results of the same data utilizing the same processing techniques but with different classification method, it is found that the results by averaging the sample individual cases show that the proposed technique supported Singular Value Decomposition (SVD) with effective performance reaches 98.61%, while the other reached 95.83%.

Fourier volume rendering (FVR) is a significant visualization technique that has been used widely in digital radiography. As a result of itsO(N2log⁡N)time complexity, it provides a faster alternative to spatial domain volume rendering algorithms that areO(N3)computationally complex. Relying on theFourier projection-slice theorem, this technique operates on the spectral representation of a 3D volume instead of processing its spatial representation to generate attenuation-only projections that look likeX-ray radiographs. Due to the rapid evolution of its underlying architecture, the graphics processing unit (GPU) became an attractive competent platform that can deliver giant computational raw power compared to the central processing unit (CPU) on a per-dollar-basis. The introduction of the compute unified device architecture (CUDA) technology enables embarrassingly-parallel algorithms to run efficiently on CUDA-capable GPU architectures. In this work, a high performance GPU-accelerated implementation of the FVR pipeline on CUDA-enabled GPUs is presented. This proposed implementation can achieve a speed-up of 117x compared to a single-threaded hybrid implementation that uses the CPU and GPU together by taking advantage of executing the rendering pipeline entirely on recent GPU architectures.

Brain Computer Interface (BCI) is a channel of communication between the human brain and an external device through brain electrical activity. In this paper, we extracted different features to boost the classification accuracy as well as the mutual information of BCI systems. The extracted features include the magnitude of the discrete Fourier transform and the wavelet coefficients for the EEG signals in addition to distance series values and invariant moments calculated for the reconstructed phase space of the EEG measurements. Different preprocessing, feature selection, and classification schemes were utilized to evaluate the performance of the proposed system for dataset III from BCI competition II. The maximum accuracy achieved was 90.7% while the maximum mutual information was 0.76 bit obtained using the distance series features.

there are many reasons lead to increase the mortality of the respiratory failure patients, who are undergoing mechanical ventilation. From these reasons, the inability of an expert physician to follow up such patients where he/she cannot stay with them all time. In addition, in many cases, lack of physicians' experience leads to the inability to provide the necessary therapeutic service at the suitable time for such patients. In this work, we provide a new approach for tele-controlling and tele-monitoring the mechanical ventilator, while tele-monitoring a patient when is demanded. In this approach, we use a simple, a small, and a lightweight unit placed inside the device in an easy and a safe way. This unit allows experts to control the device remotely using a virtual keypad via internet/intranet, while monitoring the patient him/herself. Moreover, this system can tele-control the device from anywhere without causing any defect to the internal circuitry of the device. To evaluate this system, we applied it to three types of hospital mechanical ventilators with three different types of keypads. It was easy and familiar for the operator to deal with the web control page, which has the same layout of the ventilator keypad. In addition, the operator could observe the patient and the device clearly and realistically. This system can be applied not only to mechanical ventilators but also to several other medical devices. Our system provides a wireless real-time remote control for many medical devices without any considerable delay.

Digitally reconstructed radiographs (DRRs) play a significant role in modern clinical radiation therapy. They are used to verify patient alignments during image guided therapies with 2D-3D image registration. The generation of DRRs can be implemented intuitively in O(N3) relying on direct volume rendering (DVR) methods, such as ray marching. This complexity imposes certain limitations on the rendering performance if high quality DRR images are needed. Those DRRs can be alternatively generated in the k-space using the central slice theorem in O(N2logN). Several rendering pipelines have been designed to create the DRRs in the k-space, but they were either limited to specific vendor or entail particular software requirements. We present a high performance implementation of a k-space-based DRR generation pipeline that is executable on various heterogeneous computing architectures using OpenCL. Our implementation generates a DRR for a 5123 CT volume in 6, 2.7 and 0.68 milli-seconds on a commodity CPU, mid-range and high-end GPUs respectively.

The growing importance of three-dimensional radiotherapy treatment has been associated with the active presence of advanced computational workflows that can simulate conventional x-ray films from computed tomography (CT) volumetric data to create digitally reconstructed radiographs (DRR). These simulated x-ray images are used to continuously verify the patient alignment in image-guided therapies with 2D-3D image registration. The present DRR rendering pipelines are quite limited to handle huge imaging stacks generated by recent state-of-the-art CT imaging modalities. We present a high performance x-ray rendering pipeline that is capable of generating high quality DRRs from large scale CT volumes. The pipeline is designed to harness the immense computing power of all the heterogeneous computing platforms that are connected to the system relying on OpenCL. Load-balancing optimization is also addressed to equalize the rendering load across the entire system. The performance benchmarks demonstrate the capability of our pipeline to generate high quality DRRs from relatively large CT volumes at interactive frame rates using cost-effective multi-GPU workstations. A 5122 DRR frame can be rendered from 1024 × 2048 × 2048 CT volumes at 85 frames per second.

Since its inception, ultrasound has been recognized to be one of the most significant technologies in clinical imaging. The rapid and continuous development of ultrasound scanning technologies has added another challenge of rendering relatively large scale 3D volumes with high resolution. Moreover, some applications have required particular high quality ultrasound rendering at interactive frame rate for improved interpretation of the data. Based on the vast computing power of the modern Graphics Processing Units (GPUs), we present an efficient ultrasound reconstruction pipeline that is capable of rendering high quality and high resolution ultrasound images at interactive frame rates for large scale volumes. The reconstruction results of our rendering pipeline have been applied on a real fetal ultrasound data. The performance benchmarks of our pipeline demonstrate its capability of rendering high quality ultrasound images for 10243 volumes with resolutions of 20482 at interactive frame rates using a recent commodity GPU.

X-ray rendering is recognized to be an important visualization technique in several scientific and engineering domains. It is capable of generating digital radiographs of volumetric data in the spatial domain using the X-ray transform with \(\mathscr {O}(N^3)\) complexity. Alternatively, these radiographs can be reconstructed in the k-space in \(\mathscr {O}(N^2 log N)\). This paper presents the architecture of an optimized X-ray volume rendering framework based on the Fourier slice theorem. The framework exploits the modern designs of Graphics Processing Units (GPUs). The rendering pipeline is designed to run entirely on the GPUs relying on the Compute Unified Device Architecture (CUDA) technology for computing all the data-parallel kernels and OpenGL for executing complementary geometrical operations. The interoperability between CUDA and OpenGL operations is addressed to optimize the workflow. The benchmarking results show that our framework is capable of rendering an X-ray projection of size \(512^2\) in 0.5 milli-seconds using a GeForce GTX 970 GPU.

cancer represents one of the greatest medical causes of mortality. The majority of Hepatocellular carcinoma arises from the accumulation of genetic abnormalities, and possibly induced by exterior etiological factors especially HCV and HBV infections. There is a need for new tools to analysis the large sum of data to present relevant genetic changes that may be critical for both understanding how cancers develop and determining how they could ultimately be treated. Gene expression profiling may lead to new biomarkers that may help develop diagnostic accuracy for detecting Hepatocellular carcinoma. In this work, statistical technique (discrete stationary wavelet transform) for detection of copy number alternations to analysis high-density single-nucleotide polymorphism array of 30 cell lines on specific chromosomes, which are frequently detected in Hepatocellular carcinoma have been proposed. The results demonstrate the feasibility of whole-genome fine mapping of copy number alternations via high-density single-nucleotide polymorphism genotyping, Results revealed that a novel altered chromosomal region is discovered; region amplification (4q22.1) have been detected in 22 out of 30-Hepatocellular carcinoma cell lines (73%). This region strike, AFF1 and DSPP, tumor suppressor genes. This finding has not previously reported to be involved in liver carcinogenesis; it can be used to discover a new HCC biomarker, which helps in a better understanding of hepatocellular carcinoma.

This paper features an advanced implementation of the X-ray rendering algorithm that harnesses the giant computing power of the current commodity graphics processors to accelerate the generation of high resolution digitally reconstructed radiographs (DRRs). The presented pipeline exploits the latest features of NVIDIA Graphics Processing Unit (GPU) architectures, mainly bindless texture objects and dynamic parallelism. The rendering throughput is substantially improved by exploiting the interoperability mechanisms between CUDA and OpenGL. The benchmarks of our optimized rendering pipeline reflect its capability of generating DRRs with resolutions of 2048(2) and 4096(2) at interactive and semi interactive frame-rates using an NVIDIA GeForce 970 GTX device.

Digitally Reconstructed Radiographs (DRRs) play a vital role in medical imaging procedures and radiotherapy applications. They allow the continuous monitoring of patient positioning during image guided therapies using multi-dimensional image registration. Conventional generation of DRRs using spatial domain algorithms such as ray casting is associated with computational complexity of O(N(3)). Fourier slice theorem is an alternative approach for generating the DRRs in the k-space with reduced time complexity. In this work, we present a high performance, scalable, and optimized DRR generation pipeline on the Graphics Processing Unit (GPU). The strong scaling performance of the presented pipeline is investigated and demonstrated using two contemporary GPUs. Our pipeline is capable of generating DRRs for 512(3) volumes in less than a milli-second.

In dynamic healthcare environments, caregivers and patients are constantly moving. To increase the healthcare quality when it is necessary, caregivers need the ability to reach each other and securely access medical information and services from wherever they happened to be. This paper presents an Interactive Telemedicine Solution (ITS) to facilitate and automate the communication within a healthcare facility via Voice over Internet Protocol (VOIP), regular mobile phones, and Wi-Fi connectivity. Our system has the capability to exchange/provide securely healthcare information/services across geographic barriers through 3G/4G wireless communication network. Our system assumes the availability of an Electronic Health Record (EHR) system locally in the healthcare organization and/or on the cloud network such as a nation-wide EHR system. This paper demonstrate the potential of our system to provide effectively and securely remote healthcare solution.

Volume rendering plays a significant role in medical imaging. It allows exploring the internal structures of volumetric data acquired by the different imaging modalities. This exploration allows accurate diagnosis and consequently effective treatment. Various volume rendering techniques exists. However, compared to other techniques, Fourier volume rendering has gained a wide acceptance by the radiologists for several reasons. This technique generates attenuation renderings similar to x-ray radiographs that are well interpreted by the physicians. Additionally, it works with time complexity of O(N2logN), and thus, it delivers interactive frame rates for large scale medical volumes in comparison with spatial-domain rendering techniques. The complexity associated with developing a basic rendering pipeline for this technique hinders medical imaging scientists from focusing their research on investigating new algorithms for improving the reconstruction quality of the resulting digital radiographs. In this paper, we present a flexible, extensible and semi-interactive high-level MATLAB-based framework for the Fourier volume rendering pipeline.

The use of remote diagnosis and troubleshooting are widely used nowadays to improve instruments' uptime and minimize workflow interruption. However, such feature is not used in many critical care units and/or laboratories either due to limited capabilities and access offered by manufacturers of such equipment or because of high cost for interface software packages especially for developing countries. The negative impact is more felt in laboratories with old instruments lacking remote connection ports but having an embedded maintenance and/or troubleshooting software. This paper describes a direct access remote diagnosis and troubleshooting for the input and output devices of a critical care and laboratory instrument using a flat data cable, an 8 cm × 10 cm built-in printed circuit board designed for this purpose, a standard parallel cable, and a control software. The main advantage of this approach is its low cost, compact design and simple implementation besides being convenient and suitable for other instruments possessing input/output devices with the same specifications. Implementation of the software and hardware related to accessing the instrument's membrane keypad and display using a PC has been successfully accomplished and tested leading to the use of all PC communication capabilities including remote connection and access. By integrating all hardware components, the prototype system was capable of remotely diagnosing and controlling the instrument on real-time basis.

The resolution of medical data sets acquired by state-of-the-art imaging modalities is growing rapidly. Providing high-end workstations to visualize such data sets might not be affordable in some cases. To resolve this issue, there should be alternative convenient software solutions to handle huge data sets either by out-of-core or offline rendering applications. This presented work features an offline rendering pipeline that is capable of rendering digital x-ray radiographs of very large scale volumetric data sets, which cannot fit into the memory of the running platform relying on spatial-domain decomposition and projection-slice theorem.

Volume rendering became a crucial and significant tool in the medical field. This was due to the rapid evolution of imaging modalities such as Magnetic Resonance Imaging (MRI) machines, Computed Tomography (CT) scanners. Volume rendering techniques vary from category to another depending basically on the employed modality and eventually on the target application. Although spatial-domain volume rendering techniques have gained a wide acceptance during the past decades, however, they suffer from performance bottlenecks due to their O(N3) complexity for a volume of size N3. Fourier Volume Rendering (FVR) is an alternative technique that works on the k-space representation of the volume with reduced complexity of O(N2logN) for generating projection images of the spatial volume that look like X-ray ones. In this work, a functional hybrid implementation of the FVR pipeline on Central Processing Units (CPUs) and Graphics Processing Units (GPUs) is presented.

Frequency domain analysis is one of the most common analysis techniques in signal and image processing. Fast Fourier Transform (FFT) is a well know tool used to perform such analysis by obtaining the frequency spectrum for time- or spatial-domain signals and vice versa. FFT-Shift is a subsequent operation used to handle the resulting arrays from this stage as it centers the DC component of the resulting array at the origin of the spectrum. The modern Graphics Processing Units (GPUs) can be easily exploited to efficiently execute this operation considering the Compute Unified Device Architecture (CUDA) technology that was released by NVIDIA. In this work, we present an efficient high performance implementation for two- and three-dimensional FFT-Shift on the GPU exploiting its highly parallel architecture relying on the CUDA platform. We use Fourier volume rendering as an example to demonstrate the significance of this proposed implementation. It achieves a speedup of 65X for the 2D case & 219X for the 3D case.

This paper describes a high precision navigation system that helps neurosurgeons identify more easily the neuroendoscope's tip location in the patient's brain that can significantly reduce the length and increase the accuracy of minimally invasive procedures. It is an image-guided neurosurgery system that provides integrated information of preoperative 3-D medical volumes, intra-operative neuroendoscope 2-D images, position information of the endoscope and surgical tools. The system's components include data acquisition, segmentation, multi-modal registration, 3-D surface and volume reconstruction, 3-D visualization, surgical simulation and integration of tracking systems. An articulated arm (digitizer) is used for automated registration of the 3-D model to the patient, then it is integrated with the system to dynamically localize surgical instruments and the patient's head in the patient's coordinate system. The visualization subsystem shows the position and orientation of the neuroendoscope's tip in the 3-D model of preoperative Magnetic Resonance (MR) volume and in the three orthogonal (sagittal, cronoal and axial) MR slices. The system's components and clinical applications are described.

Presents a fast, six degrees of freedom, registration technique to accurately locate the position and orientation of medical volumes (e.g. CT, MRI) with respect to each other for the same patient. The technique uses surface registration and maximization of mutual information. We have developed a novel technique for surface registration which produces highly accurate results when registering two different volumes of the same individual generated from the same modality, such as preoperative MR and intraoperative MR volumes. In case surface registration is not able to accurately register the different volumes, the result is enhanced by multi-modal volume registration. The gain of this combination is to have an accurate alignment and to reduce time needed for registration. For the multi-modal volume registration, the maximization of mutual information (MI) as a matching criterion is used based on genetic algorithms (GA) as a search engine. Our results demonstrate that our registration technique allows for fast, accurate, robust and completely automatic registration of multimodality medical volumes.

Intra-operative MRI (iMRI) is a new technology that allows near real time updates of scans during a surgical procedure under the magnet. From an initial comprehensive scan, 3D volumes are obtained. During the surgery, only the portion of the organ under surgery needs to be re-scan (region of interest-ROI) in order to guide the progress of the surgery, and validate the procedure. We describe a registration approach which can be used for iMRI applications. Registration is performed through a hybrid of the surface point signature approach (SPS) and the maximization of the mutual information (MI) criterion. We demonstrate the approach on some real data.

This paper presents a 3-D visualization tool that manipulates and/or enhances by user input the segmented targets and other organs. A 3-D visualization tool is developed to create a precise and realistic 3-D model from CT/MR data set for manipulation in 3-D and permitting physician or planner to look through, around, and inside the various structures. The 3-D visualization tool is designed to assist and to evaluate the segmentation process. It can control the transparency of each 3-D object. It displays in one view a 2-D slice (axial, coronal, and/or sagittal)within a 3-D model of the segmented tumor or structures. This helps the radiotherapist or the operator to evaluate the adequacy of the generated target compared to the original 2-D slices. The graphical interface enables the operator to easily select a specific 2-D slice of the 3-D volume data set. The operator is enabled to manually override and adjust the automated segmentation results. After correction, the operator can see the 3-D model again and go back and forth till satisfactory segmentation is obtained. The novelty of this research work is in using state-of-the-art of image processing and 3-D visualization techniques to facilitate a process of a medical volume segmentation validation and assure the accuracy of the volume measurement of the structure of interest.

The growing scientific knowledge and rapid progress in medical imaging techniques has led to an increasing demand for better and more efficient methods of remote access to high-performance computer facilities. This paper introduces a web-based telemedicine project that provides interactive tools for surgical simulation and planning. The presented approach makes use of client-server architecture based on new internet technology where clients use an ordinary web browser to view, send, receive and manipulate patients' medical records while the server uses the supercomputer facility to generate online semi-automatic segmentation, 3D visualization, surgical simulation/planning and neuroendoscopic procedures navigation. The supercomputer (SGI ONYX 1000) is located at the Computer Vision and Image Processing Lab, University of Louisville, Kentucky. This system is under development in cooperation with the Department of Neurological Surgery, Alliant Health Systems, Louisville, Kentucky. The server is connected via a network to the Picture Archiving and Communication System at Alliant Health Systems through a DICOM standard interface that enables authorized clients to access patients' images from different medical modalities.© (1998) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.