Dimensional Medical Images Research Articles

Vessel segmentation in MRI using a variational image subtraction approach

Vessel segmentation is important for many clinical applications, such as the diagnosis of vascular diseases, the planning of surgery, or the monitoring of the progress of disease. Although various approaches have been proposed to segment vessel structures from 3-dimensional medical images, to the best of our knowledge, there has been no known technique that uses magnetic resonance imaging (MRI) as prior information within the vessel segmentation of magnetic resonance angiography (MRA) or magnetic resonance venography (MRV) images. In this study, we propose a novel method that uses MRI images as an atlas, assuming that the patient has an MRI image in addition to MRA/MRV images. The proposed approach intends to increase vessel segmentation accuracy by using the available MRI image as prior information. We use a rigid mutual information registration of the MRA/MRV to the MRI, which provides subvoxel accurate multimodal image registration. On the other hand, vessel segmentation methods tend to mostly suer from imaging artifacts, such as Rician noise, radio frequency (RF) inhomogeneity, or partial volume eects that are generated by imaging devices. Therefore, this proposed method aims to extract all of the vascular structures from MRA/MRI or MRV/MRI pairs at the same time, while minimizing the combined eects of noise and RF inhomogeneity. Our method is validated both quantitatively and visually using BrainWeb phantom images and clinical MRI, MRA, and MRV images. Comparison and observer studies are also realized using the BrainWeb database and clinical images. The computation time is markedly reduced by developing a parallel implementation using the Nvidia compute unied device architecture and OpenMP frameworks in order to allow the use of the method in clinical settings.

A New Robust Zero-watermarking Algorithm for Medical Volume Data

Aiming at medical image information security problem, this paper proposes a new robust zero-watermarking algorithm for medical volume data based on three-dimensional discrete wavelet transform, three-dimensional Fourier transform and hermite chaotic neural network . Firstly, it employs a novel hermite chaotic neural network to generate the pseudo-random chaotic sequence for scrambling. Secondly, three dimensional medical image is transformed by three-dimensional discrete wavelet transform and three-dimensional discrete Fourier transform. Then, select the transformed low and intermediate frequency coefficients symbol as medical volume data characteristics to structure zero-watermarking. The algorithm integrates hermite chaotic neural network and zero-watermarking technology, which is not confined to artificially selected region of interest. The watermarking extraction does not need the original image. And its security depends on the chaotic sequence complexity and unpredictability, solving the watermark embedding, extraction safety and efficiency. The simulation results show that the algorithm is simple to implement, with good robustness, security and invisibility.

Imaging for Preintervention Planning

Although all preoperative imaging can be considered surgical planning (SP), it is defined in this article as the act of using preoperative data to simulate the surgical procedure or the result of the procedure. It is the combination, all or in part, of 3-dimensional (3D) medical imaging, applied computer vision, computer-aided design, and computational fluid dynamic (CFD) modeling to mimic and provide visual guidance for surgical procedures. This simulation is generally performed with multiple anatomic and physiological states to determine the robustness of the procedure. There are multiple advantages to this approach such as assessing standard interventions and creation of new surgical strategies without risking the patient’s health; this can potentially have both clinical and economic benefits. Since its first mention ≈30 years ago, SP is now a routine part of interventions in fields such as neurosurgery and orthopedics.1 Translating this paradigm to cardiovascular interventions provides not only enhanced 3D visualization but also the potential to interface with physics-driven computational solvers (eg, CFD) to predict the hemodynamic outcomes. Considering the complexity of fluid-solid interactions and the highly time-varying component of the cardiovascular system, these efforts are largely lagging behind those in the neurological and orthopedic communities, but recent advances appear promising.2 The Fontan operation for single ventricle (SV) patients, in which a conduit (the total cavopulmonary connection [TCPC]) is placed to channel systemic venous return passively into the pulmonary arteries (PAs), is the paradigm for this approach. This category of lesions is the leading cause of morbidity and mortality in congenital heart disease, and although it is generally associated with acceptable short-term outcomes, Fontan failure remains problematic. Progressive ventricular dysfunction, protein-losing enteropathy, poor exercise tolerance, pulmonary arteriovenous malformations (PAVMs), and liver dysfunction are some of the most commonly cited complications. These morbidities are multifactorial, and the underlying causes …

3D Medical Images Registration Based on GPU Parallel Computing

Real time 3D medical image registration method is key technology of medical image processing, especially in surgical operation navigation. However, current 3D medical image registration methods are time-consuming, which can’t meet the real time requirement of clinical application. To solve this problem, this paper presented a high performance computational method based on CUDA ( Compute Unified Device Architecture), which took full advantage of GPU parallel computing under CUDA architecture combined with image multiple scale and maximum mutual information to make fast registration of three dimensional medical image. Experiments showed that this algorithm can greatly accelerate the computational speed of registration of three dimensional medical image, and meet the real time requirement of clinical application.

3차원 무릎 자기공명영상 내에서 영역화와 정합 기법을 반복적으로 이용한 다중 해상도 기반의 뼈 영역화 기법

Recently, medical equipments are developed and used for diagnosis or studies. In addition, demand of techniques which automatically deal with three dimensional medical images obtained from the medical equipments is growing. One of the techniques is automatic bone segmentation which is expected to enhance the diagnosis efficiency of osteoporosis, fracture, and other bone diseases. Although various researches have been proposed to solve it, they are unable to be used in practice since a size of the medical data is large and there are many low contrast boundaries with other tissues. In this paper, we present a fast and accurate automatic framework for bone segmentation based on multi-resolutions. On a low resolution step, a position of the bone is roughly detected using constrained branch and mincut which find the optimal template from the training set. Then, the segmentation and the registration are iteratively conducted on the multiple resolutions. To evaluate the performance of the proposed method, we make an experiment with femur and tibia from 50 test knee magnetic resonance images using 100 training set. The proposed method outperformed the constrained branch and mincut in aspect of segmentation accuracy and implementation time.

Fast two dimensional to three dimensional registration of fluoroscopy and CT-scans using Octrees on segmentation maps

We introduce a computationally efficient approach to the generation of Digital Reconstructed Radiographs (DRRs) needed to perform three dimensional to two dimensional medical image registration and apply this algorithm to virtual surgery. The DRR generation process is the bottleneck of any three dimensional to two dimensional registration system, since its computational complexity scales with the number of voxels in the Computed Tomography Data, which can be of the order of tens to hundreds of millions. Our approach originates from the segmentation of the volumetric data into multiple regions, which allows a compact representation via Octree Data Structures. This, in turn, yields efficient storage and access of the attenuation indexes of the volumetric cells, required in the projection procedure that generates the DRR. A functional based on Mutual Information is then maximized to obtain the alignment of the DRR with the two dimensional X-ray fluoroscopy scans acquired during the operation. Promising experimental results on real data are presented.

Free Stuff for Your Computer

Free Stuff for Your Computer

Three-dimensional medical image analysis of the heart by the revised GMDH-type neural network self-selecting optimum neural network architecture

In this study, a revised group method of data handling (GMDH)-type neural network algorithm which self-selects the optimum neural network architecture is applied to 3-dimensional medical image analysis of the heart. The GMDH-type neural network can automatically organize the neural network architecture by using the heuristic self-organization method, which is the basic theory of the GMDH algorism. The heuristic self-organization method is a kind of evolutionary computation method. In this revised GMDH-type neural network algorithm, the optimum neural network architecture was automatically organized using the polynomial and sigmoid function neurons. Furthermore, the structural parameters, such as the number of layers, the number of neurons in the hidden layers, and the useful input variables, are selected automatically in order to minimize the prediction error criterion, defined as the prediction sum of squares (PSS).

Revised GMDH-type neural network algorithm for medical image recognition and its application to 3-dimensional medical image analysis of the heart

Revised GMDH-type neural network algorithm for medical image recognition and its application to 3-dimensional medical image analysis of the heart

EFFICIENT ALGORITHMS FOR THE OPTIMAL-RATIO REGION DETECTION PROBLEMS IN DISCRETE GEOMETRY WITH APPLICATIONS

In this paper, we study several interesting optimal-ratio region detection (ORD) problems in d- D (d ≥ 3) discrete geometric spaces, which arise in high dimensional medical image segmentation. Given a d- D voxel grid of n cells, two classes of geometric regions that are enclosed by a single or two coupled smooth heighfield surfaces defined on the entire grid domain are considered. The objective functions are normalized by a function of the desired regions, which avoids a bias to produce an overly large or small region resulting from data noise. The normalization functions that we employ are used in real medical image segmentation. To our best knowledge, no previous results on these problems in high dimensions are known. We develop a unified algorithmic framework based on a careful characterization of the intrinsic geometric structures and a nontrivial graph transformation scheme, yielding efficient polynomial time algorithms for solving these ORD problems. Our main ideas include the following. We observe that the optimal solution to the ORD problems can be obtained via the construction of a convex hull for a set of O(n) unknown 2-D points using the hand probing technique. The probing oracles are implemented by computing a minimum s-t cut in a weighted directed graph. The ORD problems are then solved by O(n) calls to the minimum s-t cut algorithm. For the class of regions bounded by a single heighfield surface, our further investigation shows that the O(n) calls to the minimum s-t cut algorithm are on a monotone parametric flow network, which enables to detect the optimal-ratio region in the complexity of computing a single maximum flow.

Differential convolution for medical diagnosis

This paper describes differential convolution and its application to high-resolution three dimensional medical images to distinguish between normal and diseased tissue. Underlying image processing principles are presented focusing on texture analysis and prototype application of three dimensional convolution to identify normal, benign and malignant tissue.

A Fast Medical 3D Image Registration Method Based on a Parametric Representation

Three dimensional medical image registration is a fundamental technique which applied various medical treatments such as image diagnosis, treatment planning, image guided surgery, etc. In radiation therapy for cancer treatment, quick alignment method around target lesion is required. To align CT images, similarity measurement like normalized cross correlation or mutual information should be calculated. The computational cost must reduce for achieving the online image registration.In this paper, we propose a novel quick rigid registration method which makes it possible to align the three dimensional images using the parametric eigenspace method. By projecting each CT slice image into the eigenspace as a set of low dimensional vectors, image similarity can be calculated very rapidly. The experiments using CT images of the same patient, it is found that the alignment accuracy is almost the same as the method using normalized cross correlation, and the computation time is less than one second.

On the Efficient Compression of fMRI Data Series of Brain

Four dimensional medical images, namely fMRI, require a large volume of memory to store the data. Compression techniques are therefore used for the storage and transmission of these medical images. This paper proposes a coding scheme for volumetric images which in the first stage recognizes significant images and in the second stage compresses those images using a JPEG-LS coding scheme. An example implementation for 4D fMRI data series of brain stored in ANALYZE file format is illustrated in this paper. The proposed scheme provides efficient compression for 4D fMRI medical images.

MITK and 3DMed : An Integrated Platform Applicable for the Development of Computer Assisted Intervention Systems

This paper introduces an integrated 3D medical image processing and analyzing software platform which is open interface and freely available. The platform consists of the Medical Imaging Toolkit (MITK) and the 3-Dimensional Medical Image Processing and Analyzing System (3DMed). MITK is an algorithm toolkit for research and software development, while 3DMed is a MITK based application system with a plug-in framework. The overall architecture and main capabilities of the platform are described in detail. Presented evaluations demonstrate that the platform can benefit the development of computer assisted intervention systems.

Recognition of 3-dimensional medical images of the lungs, pulmonary vessels and bronchial trees by artificial neural networks

Recognition of 3-dimensional medical images of the lungs, pulmonary vessels and bronchial trees by artificial neural networks

Recognition of 3-dimensional medical images of the head by radial basis function (RBF) neural network

Recognition of 3-dimensional medical images of the head by radial basis function (RBF) neural network

THE LAYERED NET SURFACE PROBLEMS IN DISCRETE GEOMETRY AND MEDICAL IMAGE SEGMENTATION

Efficient detection of multiple inter-related surfaces representing the boundaries of objects of interest in d-D images (d >/= 3) is important and remains challenging in many medical image analysis applications. In this paper, we study several layered net surface (LNS) problems captured by an interesting type of geometric graphs called ordered multi-column graphs in the d-D discrete space (d >/= 3 is any constant integer). The LNS problems model the simultaneous detection of multiple mutually related surfaces in three or higher dimensional medical images. Although we prove that the d-D LNS problem (d >/= 3) on a general ordered multi-column graph is NP-hard, the (special) ordered multi-column graphs that model medical image segmentation have the self-closure structures and thus admit polynomial time exact algorithms for solving the LNS problems. Our techniques also solve the related net surface volume (NSV) problems of computing well-shaped geometric regions of an optimal total volume in a d-D weighted voxel grid. The NSV problems find applications in medical image segmentation and data mining. Our techniques yield the first polynomial time exact algorithms for several high dimensional medical image segmentation problems. Experiments and comparisons based on real medical data showed that our LNS algorithms and software are computationally efficient and produce highly accurate and consistent segmentation results.

Recursive tracking of vascular tree axes in 3D medical images

Object This article describes a method for automated extraction of branching structures in three dimensional (3D) medical images.

Multilayered GMDH-Type Neural Network with Radial Basis Functions and its Application to 3-Dimensional Medical Image Recognition of the Liver

In this paper, a Group Method of Data Handling (GMDH)-type neural network algorithm with radial basis functions (RBF) is proposed. The proposed algorithm generates optimum RBF network architectures fitting the complexity of nonlinear systems using heuristic self-organization. The number of hidden layers, the number of neurons in hidden layers and relevant input variables are selected by minimizing prediction error defined as Akaike’s Information Criterion (AIC). Various nonlinear combinations of variables are initially generated in each layer and only relevant combinations are selected based on AIC. Hence, the optimum RBF network architecture fitting the complexity of the nonlinear system is obtained. We apply the GMDH-type neural network algorithm with RBF to 3-dimensional medical image recognition of the liver, showing that this algorithm is very easy and useful in 3-dimensional medical image recognition of the liver because the neural network architecture is automatically organized to minimize prediction error based on AIC.

GeoInterp: Contour Interpolation with Geodesic Snakes Geodesic Snakes

The process of outlining organs on three dimensional medical images is extremely time-consuming. In this document, we describe a new tool that can interpolate contours and also make them smooth. The tool uses geodesic snakes to perform these functions. The implementation uses The Insight Toolkit and the implementation of geodesic snakes that comes with SNAP.