Narrow Results

Accurate and rapid detection of Alzheimer’s disease (AD) using magnetic resonance imaging (MRI) gained considerable attention among research workers because of an increased number of current researches being driven by deep learning (DL) methods that have accomplished outstanding outcomes in variety of domains involving medical image analysis. Especially, convolution neural network (CNN) is primarily applied for the analyses of image datasets according to the capability of handling massive unstructured datasets and automatically extracting significant features. Earlier detection is dominant to the success and development interferences, and neuroimaging characterizes the potential regions for earlier diagnosis of AD. The study presents and develops a novel Deep Learning-based Magnetic Resonance Image Segmentation and Classification for AD Diagnosis (DLMRISC-ADD) model. The presented DLMRISC-ADD model mainly focuses on the segmentation of MRI images to detect AD. To accomplish this, the presented DLMRISC-ADD model follows a two-stage process, namely, skull stripping and image segmentation. At the preliminary stage, the presented DLMRISC-ADD model employs U-Net-based skull stripping approach to remove skull regions from the input MRIs. Next, in the second stage, the DLMRISC-ADD model applies QuickNAT model for MRI image segmentation, which identifies distinct parts such as white matter, gray matter, hippocampus, amygdala, and ventricles. Moreover, densely connected network (DenseNet201) feature extractor with sparse autoencoder (SAE) classifier is used for AD detection process. A brief set of simulations is implemented on ADNI dataset to demonstrate the improved performance of the DLMRISC-ADD method, and the outcomes are examined extensively. The experimental results exhibit the effectual segmentation results of the DLMRISC-ADD technique.

As a method based on the deformable surface evolution, brain extraction tools (BET) is widely used for brain extraction on cranial 3D magnetic resonance (MR) images. BET iteratively applies an evolution model depending on local parameters to the deformable surface until it reaches the brain border. In this paper, we propose a modified BET based on the fast model introducing a global parameter, the global mean inter-vertex distance of deformable surface. The surface evolution consists of two steps driven by two models: pre-processing step by the fast model and major step by the original model. We demonstrated our method in the experiments with T₁-weighted and diffusion weighted MR images of 19 normal subjects. The modified BET converged within 300 iterations less than half of about 1000 iterations in the original BET, while reducing substantially the false negative regions and self-intersections. The proposed method may be efficient for high resolution images and large-scale studies.

A brain tumor is a gathering of abnormal cells in the brain, forming a mass. In 2018, approximately 80,000 new instances of primary brain tumors were documented. Certain types of imaging modalities may not offer comprehensive details about all abnormal tissues, given the diverse biological composition present in tumors. In such perspective, a novel Modified Deep Convolutional AlexNet with Transfer Learning (MDCATL) model is proposed which uses brain Magnetic Resonance Imaging (MRI) images to predict the brain tumor. This MDCATL model operates through four essential stages such as preprocessing, segmentation, feature extraction, and prediction. First, the CLAHE approach is used for preprocessing the input brain MRI images. After that, the Hybrid Loss-based U-Net (HL-UNet) model is used to segment the preprocessed image. Subsequently, features encompassing color, texture, and statistical aspects are extracted from the segmented image. The Modified Deep Convolutional AlexNet model then utilizes these features to predict brain tumors on the input brain MRI images. Consequently, the MDCATL model effectively predicts brain MRI images through Transfer Learning across these stages. Furthermore, the efficiency of the proposed MDCATL framework is evaluated through various metrics like accuracy, precision, FNR, F-measure, etc. Our proposed MDCATL model achieves 0.944 at a learning percentage of 90, while existing state-of-art models and traditional classifiers such as AlexNet, EfficientNet, Vision transformer, CNN, CNN-LSTM, LSTM, ResNet, and DenseNet exhibited lower accuracy values.

The traveling salesman problem (TSP) is a prototypical problem of combinatorial optimization and, as such, it has received considerable attention from neural-network researchers seeking quick, heuristic solutions. An early stage in many computer vision tasks is the extraction of object shape from an image consisting of noisy candidate edge points. Since the desired shape will often be a closed contour, this problem can be viewed as a version of the TSP in which we wish to link only a subset of the points/cities (i.e. the "noisefree" ones). None of the extant neural techniques for solving the TSP can deal directly with this case. In this paper, we present a simple but effective modification to the (analog) elastic net of Durbin and Willshaw which shifts emphasis from global to local behavior during convergence, so allowing the net to ignore some image points. Unlike the original elastic net, this semi-localized version is shown to tolerate considerable amounts of noise. As an example practical application, we describe the extraction of "pseudo-3D" human lung outlines from multiple preprocessed magnetic resonance images of the torso. An effectiveness measure (ideally zero) quantifies the difference between the extracted shape and some idealized shape exemplar. Our method produces average effectiveness scores of 0.06 for lung shapes extracted from initial semi-automatic segmentations which define the noisefree case. This deteriorates to 0.1 when extraction is from a noisy edge-point image obtained fully-automatically using a feedforward neural network.

Magnetic Resonance Imaging (MRI) is an important diagnostic technique for brain tumors due to its ability to generate images without tissue damage or skull artifacts. Therefore, MRI images are widely used to achieve the segmentation of brain tumors. This paper is the first attempt to discuss the use of optimization spiking neural P systems to improve the threshold segmentation of brain tumor images. To be specific, a threshold segmentation approach based on optimization numerical spiking neural P systems with adaptive multi-mutation operators (ONSNPSamos) is proposed to segment brain tumor images. More specifically, an ONSNPSamo with a multi-mutation strategy is introduced to balance exploration and exploitation abilities. At the same time, an approach combining the ONSNPSamo and connectivity algorithms is proposed to address the brain tumor segmentation problem. Our experimental results from CEC 2017 benchmarks (basic, shifted and rotated, hybrid, and composition function optimization problems) demonstrate that the ONSNPSamo is better than or close to 12 optimization algorithms. Furthermore, case studies from BraTS 2019 show that the approach combining the ONSNPSamo and connectivity algorithms can more effectively segment brain tumor images than most algorithms involved.

Breast cancer is one of the leading causes of death from cancer in Taiwan. In this paper, we propose a feature-based scheme composed of preprocessing, feature extraction and a fuzzy classifier for suspicious region detection and identification. In the preprocessing stage, we first extract regions of interest and then coarsely determine suspicious regions via candidate screening. Some features are extracted based on intra-slice, texture and inter-slice analysis techniques for suspicious region identification. Intra-slice analysis evaluates the intensity and size of suspicious regions. To find a precise region, we propose a region growing algorithm based on ellipse-based approximation. In texture analysis, some texture cues are extracted from spatial and wavelet domains and integrated as a combined texture feature by using a neural network. Inter-slice analysis is based on the continuity characteristic and consistency of a suspicious region's size; the objective is to verify the static behavior of suspicious regions. Several magnetic resonance imaging (MRI) cases are utilized to evaluate the performance of the proposed scheme. Experimental results demonstrate that our scheme can not only extract regions of interest but also identify tumors well from magnetic resonance images.

Compressive Sensing for Magnetic Resonance Imaging (CS-MRI) aims to reconstruct Magnetic Resonance (MR) images from under-sampled raw data. There are two challenges to improve CS-MRI methods, i.e. designing an under-sampling algorithm to achieve optimal sampling, as well as designing fast and small deep neural networks to obtain reconstructed MR images with superior quality. To improve the reconstruction quality of MR images, we propose a novel deep convolutional neural network architecture for CS-MRI named MRCSNet. The MRCSNet consists of three sub-networks, a compressive sensing sampling sub-network, an initial reconstruction sub-network, and a refined reconstruction sub-network. Experimental results demonstrate that MRCSNet generates high-quality reconstructed MR images at various under-sampling ratios, and also meets the requirements of real-time CS-MRI applications. Compared to state-of-the-art CS-MRI approaches, MRCSNet offers a significant improvement in reconstruction accuracies, such as Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity (SSIM). Besides, it reduces the reconstruction error evaluated by the Normalized Root-Mean-Square Error (NRMSE). The source codes are available at https://github.com/TaihuLight/MRCSNet.

This paper offers a definition of precise, comprehensive, robust and practical neuroanatomical segmentation in magnetic resonance brain images with the goal of performing quantitative morphometric analyses. The main types of difficulties experienced with such problems are described, including those relating to the classification of MR signal intensities and the fact that there is insufficient information in the 2D image. To illustrate the details of obtaining a morphometric description, a case study of semi-automated methods is presented for segmenting the lateral ventricles and caudate nucleus in T1 coronal MR image data. The most significant remaining difficulties are summarized and are offered as objectives for further research.

A procedure for estimating the joint probability density function (pdf) of T₁, T₂ and proton spin density (P_D) for gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) in the brain is presented. The pdf's have numerous applications, including the study of tissue parameter variability in pathology and across populations. The procedure requires a multispectral, spin echo magnetic resonance imaging (MRI) data set of the brain. It consists of five automated steps:

(i) preprocess the data to remove extracranial tissue using a sequence of image processing operators;

(ii) estimate T₁, T₂ and P_D by fitting the preprocessed data to an imaging equation;

(iii) perform a fuzzy c-means clustering on the same preprocessed data to obtain a spatial map representing the membership value of the three tissue classes at each pixel location;

(iv) reject estimates which are not from pure tissue or have poor fits in the parameter estimation, and classify the remaining estimates as either GM, WM or CSF;

(v) compute statistics on the classified estimates to obtain a probability mass function and a Gaussian joint pdf of the tissue parameters for each tissue class.

Some preliminary results are shown comparing computed pdf's of young, elderly and Alzheimer's subjects. Two brief examples applying the joint pdf's to pulse sequence optimization and generation of computational phantoms are also provided.

Compressive Sensing (CS) reconstructs high-quality images from very few measurements, which are far below Nyquist rate. CS proves to be very useful for acquiring high dimensional image sets like Magnetic Resonance Imaging (MRI). However, the efficiency of MR image reconstruction is affected due to slow acquisition of voluminous k-space data. To improve the quality of reconstructed image and increase the speed of the reconstruction, a novel algorithm namely Adaptive Sparse Reconstruction using Convolution Neural Network AsrCNN has been, proposed for MR Images. AsrCNN employs a CNN, which consists of four convolutional layers and one fully connected layer. The proposed algorithm reconstructs MR images with immense quality, as it is trained over a large dataset with adaptive gradient optimization. The training set consists of $32\times 32$ image patches, which is used to create the dictionary by adaptively updating the weights. Subsequently, the dictionary is employed for recovery of sparse MR images corrupted with Gaussian noise. Patch-based approach in AsrCNN enables MR images of varied sizes to be processed without resizing. Experimental results for AsrCNN show an improvement of 1–5 dB in PSNR over previous state-of-art algorithms. Training has been done on GPU using Convolutional Architecture for Fast Feature Embedding (CAFFE) framework as it reduces significant amount of time in reconstructing images.

Variational principles and partial differential equations have proved to be fundamental elements in the mathematical modeling of extended systems in physics and engineering. Of particular interest are the equations that arise from a free energy functional. Recently variational principles have begun to be used in Image Processing to perform basic tasks such as denoising, debluring, etc. Great improvements can be achieved by selecting the most appropriate form for the functional.

In this article we show how these ideas can be applied not just to scalar fields (i.e. grayscale images) but also to curved manifolds such as the space of orientations. This work is motivated by the denoising of images acquired with Magnetic Resonance scanners using diffusion-sensitized magnetic gradients.

Accurate segmentation of human brain image is an essential step for clinical study of magnetic resonance imaging (MRI) images. However, vagueness and other ambiguity present between the brain tissues boundaries can lead to improper segmentation. Possibilistic fuzzy c-means (PFCM) algorithm is the hybridization of fuzzy c-means (FCM) and possibilistic c-means (PCM) algorithms which overcomes the problem of noise in the FCM algorithm and coincident clusters problem in the PCM algorithm. A major challenge posed in the PFCM algorithm for segmentation of ill-defined MRI image with noise is to take into account the ambiguity in the final localization of the feature vectors due to lack of qualitative information. This may lead to improper assignment of membership (typicality) value to their desired cluster. In this paper, we have proposed the possibilistic intuitionistic fuzzy c-means (PIFCM) algorithm for Atanassov’s intuitionistic fuzzy sets (A-IFS) which includes the advantages of the PCM, FCM algorithms and A-IFS. Real and simulated MRI brain images are segmented to show the superiority of the proposed PIFCM algorithm. The experimental results demonstrate that the proposed algorithm yields better result.

Wavelet transform is widely used in feature extraction of magnetic resonance imaging. However, the traditional discrete wavelet transform (DWT) suffers from translation variant property, which may extract significantly different features from two images of the same subject with only slight movement. In order to solve this problem, this paper utilizes stationary wavelet transform (SWT) to extract features instead of DWT. Experiments on a normal brain MRI demonstrate that wavelet coefficients via SWT are superior to those via DWT, in terms of translation invariant property. In addition, we applied SWT to normal and abnormal brain classification. The results demonstrate that SWT-based classifier is more accurate than that of DWT.

Venous varix of the volar digit (VVVD) is a rare condition that is usually described as a firm, blue, and painful mass. However, the clinical features of VVVD are still unclear. We treated a patient who had a firm, normal-skin-colored, and painless mass on the volar aspect of a digit. The mass was removed and was diagnosed as VVVD by histological examination. Magnetic resonance imaging was useful for assisting with the pre-operative diagnosis. We also review 11 previously reported cases.

Carpal tunnel syndrome caused by a ganglion is a rare condition. We report a case which presented with a rapidly progressive onset of symptoms and subsequent thenar palsy.

We report a case of phenytoin extravasation complicated by eschar formation. Pre-operative MRI study showed a large non-enhancing area over the dorsum of the imaged right wrist and hand corresponding with the site of phenytoin extravasation and raising the suspicion of subcutaneous tissue necrosis. The MRI findings correlated well with the intra-operative findings. We believe that pre-operative MRI in drug extravasation cases can characterise the type of soft tissue injury and define the extent of injury. This helps the surgeon in the surgical approach and treatment options.

We investigated the serial changes in MR appearance of anterior cruciate ligament (ACL) grafts after reconstruction, and looked for a correlation between MRI and clinical results. Fifty-four patients underwent serial MRI examinations at 5, 11 and 24 months after arthroscopically-assisted ACL reconstruction. The MR appearance of the graft was categorized into three types depending on the signal intensity and continuity of the ligament: low signal type, intermediate type and high signal type. For the low signal type, a homogeneous low signal band with continuity was visualized over the entire course; for the intermediate type, the signal intensity increased and low signal band was visualized only in part of the graft; and for the high signal type, the graft was not identified through the joint cavity due to markedly increased signal. Forty-three of the 54 patients retained normal low signal type during the first two years of reconstruction. They were classified as Group A. Four of the remaining 11 patients showed intermediate type at five months and altered to high signal type after 12 months. The remaining seven patients showed high signal type at five months and persisted with the same MR type until 24 months. These 11 patients with increased signal intensity were classified as Group B. The mean injured-to-uninjured differences of KT-2000 arthrometer measurements were significantly greater in Group B than that in Group A. Moreover, the percentage of cases with a difference of 5 mm or more was significantly higher in Group B (54.5%) than in Group A (9.3%). It is concluded that the majority of the grafts showed no changes in signal intensity during the first two years of reconstruction. The increase in signal intensity observed in some patients may be a reflection of a deterioration in graft integrity following reconstruction.

Purpose: A minipig model was used to demonstrate MRI findings in the first three months after an annular disruption. Methods: An incision was made into one of the lumbar discs in each of eight minipigs in the outer and middle parts of the annulus. The remaining intact discs in the lumbar and thoracolumbar regions were used as controls. MR imaging was performed one month and three months after trauma using both a 1.0 T and a 1.5 T MR unit. The histologic analysis was also carried out to demonstrate morphological changes in disc. Results: Eighty-eight percent of the injured discs had a diminished area of bright signal in the nucleus pulposus, also in cases where no signs of trauma in the annulus could be detected in MRI. The degeneration process of the nucleus pulposus was shown to progress during follow-up. High intensity zones were detected in 50% of the injured discs and they tended to appear already after one month follow-up. Histological examination showed that the high-intensity zone contained clusters of nuclear cells originating in the nucleus pulposus. Conclusion: It is concluded that lesions producing high-intensity zones can be induced in an experimental animal model and it can already be detected one month after the trauma. Degeneration process of the nucleus is generally initiated after a peripheral annular lesion.

The present retrospective study reviewed and examined the prevalence of thoracic disc degeneration, end-plate lesions and osteophytes in the thoracic spine using T2-weighted sagittal magnetic resonance images (MRI). The sample comprised 216 thoracic spine cases (101 males and 115 females), aged from 1 to 85 years (mean age = 42±19.7 years). Nuclear and anular degeneration, end-plate lesions and osteophytes were graded using a 3-point scale.

The prevalence of degeneration was highest in the nucleus (86%) and lowest in the end-plates (63%). Males had a higher prevalence of degeneration in the nucleus, anulus and end-plates, and a lower prevalence of osteophytes compared to females. Increasing cranio-caudal trends in the prevalence of degeneration in the nucleus, anulus and end-plates were observed, and these trends were statistically significant (p<0.01). Vertebral body osteophytes were most prevalent in the mid thoracic region. Osteophytes and degenerative changes in the nucleus and anulus increased significantly with age (p<0.05). These regional and age-related degenerative trends may influence the interpretation of thoracic spine pathology from MRI investigations.

Objective: To present clinical experience on matrix-induced autologous chondrocyte implantation (MACI), we hereby reported treatment with MACI for 3 patients suffering from chondral lesion of the knee, each of them has been followed for a minimum of 10 months.

Methods: Ages of 3 patients were 25, 15 and 32 years old respectively. And the cartilage defect size ranged from 6cm²-10.5cm²). IKDC2000 score was used for knee functional evaluation. Magnetic resonance imaging (MRI) and arthroscopy were performed preoperatively and postoperatively. A biopsy of the regenerated cartilage from one patient was histologically evaluated 15 months after MACI.

Results: In the postoperative period, no associated complications were observed. Each patient showed improvements both in clinical and functional status after surgery. MRI and arthroscopy showed the presence of hyaline-like cartilage at the site of implantation. The cartilage biopsy on the first patient showed a high ratio of hyaline-like cartilage tissue to fibrocartilage tissue which was 2 to 1.

Conclusion: The clinical outcome and histological evaluation suggest that MACI is able to relieve pain and restore the function of the knee, also is capable of regenerating hyaline cartilage. In conclusion, MACI appears a promising method for the treatment of chondral defects of the knee.