Search

Alzheimer’s disease involves the death of neurons in the brain, which leads to the patient’s memory loss. It is an irreversible neurological disease which has no cure. Early prognosis is important to prevent progress of this disease and to treat the Alzheimer brain images of the patient are captured and classified based on the results. The images of the brain must be clear and it must differentiate the various parts of the brain clearly in order to find the type of the disease. suggestThis paper has suggested using a convolutional neural network to detect Alzheimer’s disease (CNN). Further, this study intends to achieve the best disease prediction by using machine learning techniquesfor the benefit of the medical team. It is one of the future scope techniques which make a machine everything possible which can be and cannot be done by humans. The Deep Learning technique has grown a lot in recent years. DL algorithms operate well with massive datasets, which is very useful in cases like AD. Also, the proposed research work has used the Convolutional Neural Network (CNN) technique to classify and identify diseases by using MRI images, with a total of 2560 normal images, 1792 MCI images, 717 early AD images, and 52 late AD portrayals.5121 photos were utilised to train the data and 1279 images were tested once the data had been trained. A significant accuracy of 94% is achieved, to cover a large dataset deep learning algorithm is more beneficial when compared with traditional techniques (machine learning).

Brain tumor can cause the creation of most aggressive cancer, with a much shorter life expectancy in most advanced stages, unless identified and treated accordingly. In earlier, radiologists have to manually identify the tumors from MRI images or other imaging types. That is both time consuming and threatening to the misclassification that could affect the recovery plan of a patient. Technological innovations and machine learning assist radiologists to detect tumors without invasive procedures. One of the machine learning algorithms that has been shown to be effective at image segmentation and classification is the convolutional neural network (CNN). In this proposed work, a novel CNN architecture was used on a publicly available figshare dataset to identify three brain tumor types. The proposed CNN architecture outperformed most state-of-the-art approaches, achieving a classification accuracy of 96.90 %. Precision, recall, and F1-score are some of the other evaluation metrics used in the study. In addition, the paper includes an in-depth analysis of misclassifications.

Main objective of this framework is to build a efficient deep learning model to detect the brain tumor. In this paper, the framework mainly focuses on the detection of brain tumor MRI images from the BraTS2020 dataset which is a part of the MICCAI BraTS2020 challenge, using U-Net architecture which is suitable for quick and accurate image classification and achieved a training accuracy of 98.485%. When compared to other architectures on BraTs2020 dataset, U-Net architecure with customization provides better results.

The tissues that are existing inside the skull becomes spontaneous called Brain tumor, and this tumor has gotten one of the significant worries in the medical care area prompting the passing of individual explicitly kids and men. This paper fundamentally outlines those processes that are vital to examine a patient's condition to ensure that brain tumors can be perceived as soon as possible alongside the certainty that the odds of errors are diminished as much as possible. The requirement for exactness and better determination can likewise assist tolerance with beating certain issues, for example, diminishing in cash, the danger of various types, for example, early demise, and so forth. In this paper, there is an explanation about the collection of datasets, improvement in the quality of image, image separation, feature mining, feature reduction as well as selection., detection of cancer cells, and performance estimation methodologies. This paper additionally gives an outline of the technique or methods needed to deal with the MRI (Magnetic Resonance Images) brain images in those above steps of processes.

Deep learning is a machine learning method that gains excellent attention from researchers in recent years. Deep Learning is considered a success in solving various complex problems in the medical image area. For example, brain tumour disease is a deadly disease that causes sufferers to have a low life expectancy after being diagnosed. Diagnosis from experts also takes much time, and experts’ diagnosis results depend on their experience. Therefore, we need an automatic classification that can help experts in making decisions. CNN has been paid much attention to the abilities of complex classification, so there are many state-of-the-art models from the improvement of CNN. Efficient Network is one of CNN models that proposes high accuracy and less computational. Accordingly, this study suggested using the Efficient Network architecture to classify the types of glioma, meningioma, and pituitary brain tumours. Efficient Network has eight levels of category, which are from EfficientNet-B0 to EfficientNet-B7. This study obtains accuracy the best results in EfficientNet-B1 and EfficientNet-B2, which achieved a high accuracy of 96%.

Medical image processing has gained a lot of relevance of late and is turning out to be a boon in different clinical aspects. One such emerging field is brain tumor detection from magnetic resonance images (MRI) of brain. Engineers are actively developing tools to process medical images and aid doctors in with their diagnosis. MRI images are clinical images which are obtained on a computer when a patient goes through MRI scanning done by a respective machine. A brain tumor is a cluster of an abnormal mass of tissue where cell growth is out of control due to an abnormality in the mechanisms that control normal cells. In this work, our main aim is to measure the size of a brain tumor in terms of its diameter and area from a patient's MRI imagery using MATLAB. The proposed method incorporates different layers of noise removal techniques to clean up the images as well as image segmentation and morphological operations for detection and extraction of tumors and determining their size. By measuring the size of the tumor in a patient's brain at regular intervals doctors can diagnose the severity of the tumor in a patient's brain and can administer suitable treatment before it gets malignant.

Deep learning becomes a key technology used in Computer Vision tasks due to its outstanding performances when compared to the traditional methods. It is a machine learning strategy that helps the computer system to learn by itself through examples. But to enhance the training process, deep learning models require rich dataset. In medical imaging applications, it is hard to get such large datasets due to patient privacy laws and the presence of data with incorrect labels. However, data augmentation methods have been adopted to expand the dataset to a large extend. This paper aims to discuss some of the recent approaches of data augmentation that can be used to generate artificial brain MRI images for the detection of Alzheimer’s Disease. To learn the useful aspects of the different augmentation methods, Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset is used. Various data augmentation strategies were employed and their impact on the capabilities of such deep learning methods was studied. Lastly, the most assuring research area of generative adversarial network in data augmentation that generates high-quality synthetic brain MRI images for the diagnosis purpose is identified.

A brain tumor is a mass or development of atypical cells inside the skull region of the brain. The growth of such malignancy in a confined space leads to a cohort of problems, like the malfunctioning of the brain. The tumor can be malignant or benign, and early detection might turn out to be a savior. For this purpose, computerized tomography (CT) scans and magnetic resonance imaging (MRI) scans are examined. In recent decades’ image processing, computer vision and deep learning approaches have gained substantial recognition. However, straightforward approaches using image enhancement techniques and morphological operations are also much efficient in this regard, such an image processing approach is compared to the state-of-the-art deep learning techniques in this paper for detecting a tumor in the MRI scans of the brain. The straightforward system is incorporated into four steps. First, the scan is pre-processed for adjustment of its quality. Second, the image is enhanced using image enhancement approaches. Third, edge detection approaches are applied to it. Fourth, image segmentation with morphological operators is applied to detect the tumor region. The findings are then compared with the results of previous deep learning techniques. The purpose of this study is to present that advanced deep learning algorithms can generate better results and perform multiple classifications of brain tumor detection in MRI images.

In this paper the definition of the main mechanisms of electroimpedance myography signal formation and the study of their contributions using physical modeling based on the finite element method is considered. The study used anatomical models of the forearm reconstructed from a series of MRI images of three volunteers corresponding to three wrist positions: rest, flexion and extension. The results obtained showed that the opposite effect of the skin-fat and muscle layers on the resulting electroimpedance myography signal determines the importance of selecting the electrode system design parameters, depending on the purpose of the measurement. An approach to the modeling of the processes that accompany muscle contraction during flexion and extension of the hand is proposed to assess the impact of changes in the thickness of the skin-fat layer, muscle morphology, size and location of the electrode systems on the generated signal, which allows individual selection of the optimal parameters of electroimpedance myography registration for the specific control task.

Brain tumor is formed by the uncontrolled growth of cancerous (malign) or non-cancerous (benign) unhealthy cells in the brain. In the present world brain tumor is a very dangerous disease and the main reason for many deaths. Magnetic Resonance Imaging (MRI) is mostly used to produce medical images for the brain tumor analysis. This paper has two objectives, first one is to identify if the MRI images has or not brain tumor using Convolutional Neural Network algorithm based on ResNet50 architecture. For second objective, image segmentation, we propose a generalized focal loss function based on Tversky index. Compared to the commonly used Dice loss, the proposed loss function achieves better precision and recall when training on small structures such as lesions.

In this paper, we propose applying Convolutional Auto-Encoder and Neural Networks (CAENN) in detection of cortical demantia based on MRI images in a challenging data set from National Brain Mapping Center of Iran. The proposed method is shown to have a higher precision compared to Convolutional Neural Network (CNN) and some other classification methods. The accuracy of the method was 0.95 for train data, 0.85 for test data and 0.78 based on the cross-validation.

Many algorithms and methods have been published regarding Corpus Callosum extraction from MRI images. However, all of these approaches use T1-weighted images or existing human brain atlas as their basis. We propose a new algorithm that can work with T2-FLAIR image. The basis of our method is from the observation that corpus callosum is located directly above a cavity inside the brain in almost every MRI images. We highlight the corpus callosum by first extracting the cavity inside the brain and then performing a morphological dilation upwards to get the rough location of the corpus callosum. Our evaluation shows that this method achieves an average of 76.90% in DSC when used to extract corpus callosum from 30 different images and 3 different subjects.

MRI and CT image fusion, which combines the features of MRI and CT images to obtain high quality images, is a significant domain. In this paper, a novel image fusion algorithm based on the weight difference between L1 and L2 norm and tight wavelet frame is presented. We first proposed the fusion model of MRI and CT images, and then proposed to apply alternate multiplier method to solve the model. Experiments show that both CT and T1-weighted MRI, CT and T2-weighted MRI, or CT and PD-weighted MRI, our proposed algorithm can achieve balance in the source images and obtain the fusion image with the best effect, which will be beneficial to help doctors make accurate diagnosis.

Segmentation and classification methods are very important in identifying brain diseases. It has been observed that brain tumor is the major remediable tumor. As we know that Brain tumor recognition is tedious and exorbitant work and if it is not identified on time, it can be converted into a hazardous situation. There is foremost requirement to build efficacious and favorable procedure for brain tumor segmentation as well as classification. MRI (magnetic resonance images) is considered here because it is non-invasive, safe and cost effective. That is why in imaging of the brain, it is widely used and considered. To tackle the brain tumor and to recognize its stages, experiments and researches are still going on. Through this study a structured and well mannered review of the methods of brain tumor segmentation and classification is presented. Tumor recognition through CAD is shown in this paper because manual detection is very difficult.

One of the most leading death causes in the world is brain tumor. Solving brain tumor segmentation and classification by relying mainly on classical medical image processing is a complex and challenging task. In fact, medical evidence shows that manual classification with human-assisted support can lead to improper prediction and diagnosis. This is mainly due to the variety and the similarity of tumors and normal tissues. Recently, deep learning techniques showed promising results towards improving accuracy of detection and classification of brain tumor from magnetic resonance imaging (MRI). In this paper, we propose a deep learning model for the classification of brain tumors from MRI images using convolutional neural network (CNN) based on transfer learning. The implemented system explores a number of CNN architectures, namely ResNet, Xception and MobilNet-V2. This latter achieved the best results with 98.24% and 98.42% in term of accuracy and F1-score, respectively.

A Machine Learning based software is developed using brain magnetic resonance which is used for segmentation and to analyze if the tumor is benign and malignant. In this paper, contrast improvement procedure, twofold thresholding is finished utilizing morphological tasks to upgrade the MRI picture and to eliminate undesirable non cerebral tissues from MR picture a cycle called the skull stripping is utilized most. This is done to fragment the cerebrum tumor and it is finished by utilizing a cut of Magnetic Resonance (MR). Picture Feature are determined utilizing GLCM is utilized to infer the picture highlights and precise division of the tumor district is done in less calculation time. This is finished via preparing these highlights in a specific strategy. Subsequently, in the proposed framework, utilizing triple procedure K-means, KNN and FCM highlights that are gotten from the GLCM, can be more precise and proficient. This Is finished utilizing triple method Means, FCM, and KNN so the exactness and blunder rate is determined for cerebrum MRI picture.

Magnetic resonance imaging (MRI) is the most effective and reliable imaging process to visualize the internal organs of the human body for clinical purposes and medical research. Although MRI does not use any radiation effect, it takes a long time to acquire an MRI image that makes discomfort to the patients, especially children and persons who have anxiety. Compressed sensing (CS) is a recently used technique for allowing faster acquisition of MRI. Since CS generates random under-sampling k-space measurements due to allowing faster acquisition, a non-linear algorithm is required for reconstructing a perfect MRI image. In this work, a comparative study is performed among single priors-, combined priors-, and deep learning-based non-linear algorithms for the reconstruction of MRI from CS measurements. Experiments are conducted to perform a comparative study that shows the significant aspects of the algorithms in terms of visual quality and quantitative measurements compared with each other. The results of the experiment proved the superiority of the combined priors-based method when compared with the single priors- and deep learning-based method.

Recent era witness a high mortality rate from brain cancer. Effective and early diagnosis using computer vision approach can be utilized for further therapeutics. The broad classifications of brain-related tumors identified so far are Meningiomas, Gliomas and Pituitary tumors. In this paper, an unsupervised computer vision approach is presented for effective extraction of Meningiomas tumors from Tl-weighted MRI images. The proposed approach utilizes a simple k-means clustering with thresholding approach followed by a set of morphological approaches in effective extraction of the tumor mass from the MRI images. The proposed approach is tested on a widely available benchmark dataset and compared with various state of art recent computer vision approaches using different performance measures for validation. The results show that the proposed approach archives 99.47% accuracy for the dataset indicating its superiority compared to others.

Femur bone marrow MRI images may provide more information for better characterizing hematologic malignancies and understanding their prognostics compared with blind bone marrow biopsies and aspirates. However, the interpretation of femur bone marrow MRI images is time-consuming and intervariable among different physicians. To develop a computer-aided diagnosis system for hematologic malignancies using femur bone marrow MRI images, we propose a fully automatic method for femur bone segmentation using deep learning. Five classical pretrained networks such as U-Net and Segnet based on a backbone of Resnet50 network are used for femoral bone segmentation. We conducted an experimental study on 200 cases with hematologic malignancies. Through training using 149 patients' T1-enhanced MRI images, we obtained an average Dice coefficient of 0.907 using Segnet based on a validation using 38 patients' T1-enhanced MRI images.

MRI is a kind of biological magnetic spin imaging technology. Making full use of MRI images is of great significance for improving the accuracy of medical image diagnosis process for the diagnosis of diseases. In the diagnosis of cerebral hemorrhage, the location and size of the hemorrhage area are meaningful indicators for diagnosing the severity of cerebral hemorrhage. In order to assist the doctor's diagnosis, we have proposed an automatic detection method based on the combination of morphology and threshold segmentation to calculate the location and size of the bleeding point according to the characteristics of the image of different sequences in the magnetic resonance image. : (1) Pre-process the MRI image; (2) Remove the skull part by morphology and leave only the brain soft tissue image; (3) Use Oust and morphology to extract the suspected lesion parts in the brain image; (4) Finally, The longitudinal relaxation time (T1 image) and transverse relaxation time (T2 images) were fused to locate the bleeding point. The experimental results show that the algorithm has high feasibility and accurate positioning of bleeding points.

Among neurological diseases, Multiple Sclerosis (MS) is the leading cause of disability in young adults. Several researches have been carried out to explore this disease and detect MS lesions in Magnetic Resonance (MR) images. In fact, lesions segmentation in MR images is very important for accurate diagnosis, adequate treatment and for monitoring the patient with MS. Spinal cord (SC) atrophy occurs at the onset of MS and there is a correlation between atrophy and disability development. Generally, Magnetic Resonance Imaging (MRI) is the most sensitive method, which allows the visualization of demyelination plaques and the quantification of spinal cord atrophy. Detection and quantification of SC atrophy in MRI images are the key to the assessment of the states of MS patients. In this paper, we present a comparative study between different spinal cord segmentation methods aiming to quantify spinal cord atrophy.

The emergence of PC innovation image handling strategies have turned out to be progressively vital in a wide assortment of uses. Image segmentation is a great subject in the field of image development and also it is a hotspot and focal point of image handling methods. For image segmentation, a few broadly useful calculations and systems weredesigned. Since there is no broad resolution to the issue of image segmentation, these methods must be regularly combined with area learning to adequately address an issue domain issue of image segmentation. The tumor in the portion of the edema is very hard to figure out the boundary. No one has provided an exact estimate of the tumor boundary for the edema. Computation of the Novelty segmentation that divides brain images into MR and CT images for tumor and edema. The location of deep tissues is done simultaneously with unfortunate tissues in the light of the fact that assessing change brought about by the spread of tumor as well as edema on strong tissues is significant in the arrangement of attention. By using Enhanced RANSAC algorithm for estimating region-of-interest in multimodal MRI images, and then extracting seed points from there for region-widening based on a new notion of “affinity.” Finally, a two-step approach is developed to optimize area-convergence glioma boundary and enhance distance regularization level setting process.

Recently, interest in MR-only treatment planning using synthetic CTs (synCTs) has grown rapidly in radiation therapy. However, developing class solutions for medical images that contain atypical anatomy remains a major limitation. In this paper, we propose a novel spatial attention-guided generative adversarial network (attention-GAN) model to generate accurate synCTs using T1-weighted MRI images as the input to address atypical anatomy. Experimental results on fifteen brain cancer patients show that attention-GAN outperformed existing synCT models and achieved an average MAE of 85.223±12.08, 232.41±60.86, 246.38±42.67 Hounsfield units between synCT and CT-SIM across the entire head, bone and air regions, respectively. Qualitative analysis shows that attention-GAN has the ability to use spatially focused areas to better handle outliers, areas with complex anatomy or post-surgical regions, and thus offer strong potential for supporting near real-time MR-only treatment planning.

The brain tumor is a disease that affects or harms the brain with unwanted tissues. This is very difficult to detect brain tumor tissue from whole brain. Early detection of tumor is very important to save patient's life. Detection or segmentation techniques are used to detect and segment the brain-tumor region from the MRI images of brain and it is very useful method in recent days. In medical, magnetic-resonance-imaging is a tough field in image processing because accuracy percentage must be very high so doctors could get proper idea about diseases to save patient's life. Some MRI images have been taken as inputs data. The brain-tumor segmentation process is performed for separating brain-tumor tissues from brain MRI images, The MRI images should be filtering such as with the median filtering technique and skull stripping should be done in pre-processing, the thresholding process is being done on the given MRI images with using the watershed segmentation method. Then at last the segmented tumor region is obtained. And then in other phase features extracted by GLCM methods using MATLAB software. Then, the some images have been classified using support vector machine (SVM), this system obtained with the average accuracy of 93.05%. Which is quite better than other conventional models.

With the emergence of sophisticated head-mounted displays (HMDs), virtual reality (VR) is gaining much interest in the field of medical science and diagnosis. There exist many software tools that support MRI imaging in 3D, however, limited attention has been paid to the VR domain. In this paper, we present a VR tool called VRvisu++ which attempts to bring the spatial advantage that VR has to offer to MRI imaging. This tool allows doctors and medical practitioners to directly interact with MRI images in a VR environment thereby supporting surgical training and clinical decision making.

Most deep learning based vertebral segmentation methods require laborious manual labelling tasks. We aim to establish an unsupervised deep learning pipeline for vertebral segmentation of MR images. We integrate the sub-optimal segmentation results produced by a rule-based method with a unique voting mechanism to provide supervision in the training process for the deep learning model. Preliminary validation shows a high segmentation accuracy achieved by our method without relying on any manual labelling.The clinical relevance of this study is that it provides an efficient vertebral segmentation method with high accuracy. Potential applications are in automated pathology detection and vertebral 3D reconstructions for biomechanical simulations and 3D printing, facilitating clinical decision making, surgical planning and tissue engineering.

Tumor indicates unfettered presence of a cluster of cells in a specific area of the body part. Brain tumor is considered one of the most common tumors for both men and women and can lead to high death risk if patients fail to obtain appropriate medical treatment. In order to diagnose brain tumors, electronic modalities are integrated and among them MRI is a popular one. For MRI brain tumor region analysis segmentation, detection and classification are considered as important steps in digital imaging pathology laboratory. Existing state-of-the-art approaches demand widespread amount of supervised training data from pathologists and may still accomplish poor results in images from unseen tissue types. A suitable framework has been presented in this study to identify brain tumor cells for MRI images. In this study for the first time in compare to all other existing accessible approaches morphological operations has been incorporated to eliminate undesirable regions and to assist segmentation and identification of region of interests. Compared with existing state-of the-art supervised models, our method generalizes considerably improved identified results on brain tumor cells deprived of training data. Even with training data, our approach attains the identical performance without supervision cost. This study results indicates an accuracy rate above 96.23% accuracy associated to existing works.

MRI scans have been very helpful in the study of the diagnosis or segmentation of brain tumors in recent years. The brain tumor may be detected due to MRI scans. The MRI image is shown in the nervous system to create abnormal tissue growth or blood blocks. The first step in diagnosing the brain tumor is to control the brain structure, which symmetrically and asymmetrically identifies abnormalities. The next step is segmentation based on morphological operations (Fuzzy transformation). In this post, we explored different methods for MRI image identification and brain tumor segmentation. Precise tumor removal is important for brain tumors because of the complex brain structure. Some parameters for extracting features such as configuration, form, dimensions and image position are considered. With respect to the results retrieved from extract features the process of tumor classification is performed. This paper offers a number of techniques for the prediction of brain tumors.

This paper presents a novel technique for predicting the possibility of being malignant brain tumor based on human physical symptoms using internet of things. Other state-of-the-art techniques for detecting the malignant brain tumor based on MRI images use X-ray light to detect malignant tumor, which is more expensive and harmful for health. On the other hand, the proposed device is designed in a portable way to monitor the real-time heart rate, blood pressure, body temperature and also more suitable for predicting the possibility of being malignant tumor compared to other existing techniques. This study also shows the relationship between symptoms and sub-symptoms of malignant brain tumor. The effectiveness of the proposed method for predicting the possibility of being malignant tumor from physical symptoms is testified by the result analysis.

Theoretical analysis of MRI image data different restorative materials in the oral cavity based on an intelligent system is conducted in this manuscript. The intelligence of the medical equipment not only enables doctors to obtain more detailed information on the curative effect but also makes it more convenient to operate. In this paper, we have the following novelties. (1) The long-term recording and the analysis of the operating center magnetic field frequency can steadily weaken the magnetic field. (2) The idea of super-resolution has achieved good results in the reconstruction of the multi-dimensional MRI data, the sparse model is applied to enhance the robustness of the framework. (3) The improved image will still show a decrease in MRI quality to some extent, affecting the accurate development of the surrounding important issues. Through the robustness testing and simulation on different materials, the performance of the model is validated.

MRI images provide a new pathway in the medical field for clinical analysis and diagnose process. As MRI images suffers from contrast degradations, these contrast degradations in MRI images can be solved by application of image contrast enhancement techniques making it more suitable for medical applications. The main objective of image enhancement is to obtain a high quality image after improving the characteristics of the input image. Contrast enhancement techniques are used in images to enhance the visibility quality of internal human body texture in magnetic resonance imaging (MRI) images. In this paper we have applied the Histogram Equalization (HE), Adaptive Histogram Equalization (AHE) and Contrast-Limited Adaptive Histogram Equalization (CLAHE) in order to improve the contrast of MRI medical images. Experimental results have shown good efficiency to improve the contrast of MRI medical images with very high quality. Comparison of various image quality measures like MSE, PSNR and SNR are taken into account for evaluation the quality of enhanced MRI medical images.

In order to accurately identify the uterine region of the patients with pernicious placenta previa, to extract the features of the uterine region, and to accurately predict the amount of bleeding in cesarean section before operation, this paper presents an automatic segmentation method of uterine region based on MRI images. This method uses DeepLab-V3+ network to segment the original MR abdominal cavity image and get the uterine region image. The experimental results show that the average recognition accuracy, sensitivity and specificity of the segmentation model used in this paper are 86.49%, 91.00% and 91.89% respectively on 82 sets of positive and 128 sets of negative MRI uterine images, which indicate that this method has potential clinical application prospects in recognition of uterine region for patients.

Brain tumor, a type of cancer, is caused by a genetic mutation of abnormal neuronal cells. However, it cannot be easily diagnosed and depends on the patient’s symptoms which range from hemialgia, seizure, irregular vision, mental shift and many more. The symptoms may vary depending on the region of the tumor. Currently, Magnetic Resonance Imaging (MRI) scan is the foremost means for tumor detection as well as identifying the position and extent of the tumor for surgical procedure. However, the MRIs are needed to be manually checked by a professional to determine the results. We propose a system which is an effective image processing algorithm for detecting and recognizing the tumor from MRI images to obtain the image segmentation of brain bleeding more accurately, making immediate medical treatment possible. In this system, the same problem with different procedures and methods is tested to find the best conjunction through trial and error. Through the combination, the effectiveness and error ratio of the results is engrossed on. This approach is for early detection of brain tumors with high correctness which gives an adequate result by detecting brain tumor properly.

Today image processing techniques plays a significant role in medical imaging. It is a very growing and challenging field. Medical imaging is advantageous in identification of the disease. Numerous people suffer from brain tumor; it is a serious and dangerous disease. Medical imaging provides appropriate diagnosis of brain tumor. There are many techniques to detect brain tumor from Magnetic Resonance Images (MRI) images. These techniques face many challenges like finding the location and size of the tumor, image segmentation is useful to detect the tumor from the brain image. Already, number of algorithms are developed and tested successfully for image segmentation. In this study paper we cover the basic concept and practices of brain tumor detection from MRI images; review of different brain tumor segmentation method is presented in this paper.

MRI-only radiotherapy is expected to be safer and more precise compared with conventional CT-based radiotherapy. But unlike CT, MR images is not related with electron density for radiotherapy planning. To use MR images for treatment planning, we proposed a residual learning based u-shaped deep neural network (RUN) to synthesize CT images from MR images. The RUN network is an encoding-decoding network consisted of 16 residual learning based units which are constructed with parameter-free shortcuts to alleviate the degradation problem of deep network and strengthen feature propagation and reuse. We input MR and corresponding CT images to the RUN network to learn a voxel-by-voxel mapping from MR to CT images. Synthetic CT images are generated for newly input MR images through the learned mapping model. Our datasets contain 35 patients with more than 5000 T1/T2-weighted MR and CT images pairs. We compare Hounsfield Unit (HU) discrepancies between synthetic CT and original CT images. The mean absolute error (MAE) was $66.12 \pm 5.95$ HU, peak signal noise ratio (PSNR) was $28.52 \pm 0.64$ dB, and structural similarity index (SSIM) was $0.97 \pm 0.005$ for T1-weighted MR. The comparison results of the same metrics for T2-weighted MR were $63.79 \pm 4.18$ HU, $28.8 \pm 0.55$ dB and $0.973 \pm 0.004$ respectively. Converting CT images from MR images for one patient takes about 3 seconds on average. Experimental results show that the proposed RUN network is accurate, robust, and efficient for predicting synthetic CT from MR images for MRI-only radiotherapy.

Recently, deep learning methods, in particular, convolutional neural networks (CNNs), have made a massive breakthrough in computer vision. And a big amount of annotated data is the essential cornerstone to reach this success. However, in the medical domain, it is usually difficult (and sometimes even impossible) to get sufficient data for some specific learning tasks. Consequently, in this work, a novel data augmentation solution, called semi-supervised attention-guided CycleGAN (SSA-CycleGAN) is proposed to resolve this problem. Specifically, a cycle-consistency GANs-based model is first proposed to generate synthetic tumor (resp., normal) images from normal (resp., tumor) images. Then, a semi-supervised attention module is further proposed to enhance the model's capability in learning the important details of the training images, which in turns help the generated synthetic images become more realistic. To verify its effectiveness, experimental studies are conducted on three medical image datasets with limited amounts of MRI images, and the proposed SSA-CycleGAN is applied to generate synthetic tumor and normal MRI images for data augmentation. Experimental results show that i) SSA-CycleGAN can add (resp., remove) tumor lesions on (resp., from) the original normal (resp., tumor) images and generate very realistic synthetic tumor (resp. normal) images; and ii) in the ResNet18-based MRI image classification tasks based on these datasets, data augmentation using SSA-CycleGAN achieves much better classification performances than the classic data augmentation methods.

This paper is aiming to review some machine learning tools and to see if support vector machines (SVM) are accurate in 3D MRI images. We intend to show that 3D images can be converted into 2D and by considering machine learning techniques and SVM tools. According many research and database, 3D images are more informative than 2D ones, however working with 3D images is time consuming and requires specific programs and coding. In this research work and next which case study is under progress, we try to see the benefits of having a 3D database but to use 2D vectors only for comparison and more accurate results. This case study helps to gain better outcome. Support vector machines (SVM) can be a useful tool during the diagnosis process, however to be able to make better assumptions, more tests are needed. The technology can be viewed as detailed because it is usually applied in the discrimination of the blocks found in the areas of MS lesions and the regions which are not affected by the lesions. Primarily, to correctly segregate the different areas the textural background plays a crucial part in elevating the effectiveness of the imaging concept. In this case, the study of the slice blocks should be done comprehensively as it aims in ensuring that the type of results provided showcases the exact situation of the individual suffering from the ailment.

Heart disease is highly prevalent in developed countries, causing 1 in 4 deaths. In this work we propose a method for a fully automated 4D reconstruction of the left ventricle of the heart. This can provide accurate information regarding the heart wall motion and in particular the hemodynamics of the ventricles. Such metrics are crucial for detecting heart function anomalies that can be an indication of heart disease. Our approach is fast, modular and extensible. In our testing, we found that generating the 4D reconstruction from a set of 250 MRI images takes less than a minute. The amount of time saved as a result of our work could greatly benefit physicians and cardiologist as they diagnose and treat patients.

Neural networks have progressively used at a surprising rate now a days. Deep neural networks are majorly used machine learning models in different areas, from analysis of image to natural language processing (NLP) and broadly conveyed in scholarly community and industry. These progresses to have enormous possibilities for medical imaging, analysis of medical data, medical diagnostics. This paper illustrates a thorough literature review of deep learning techniques. This paper illustrates brief discussion about current deep learning architectures used for medical imaging and magnetic resonance imaging (MRI) for image classification, detection, segmentation, registration, etc. This review mainly focuses on the applications of deep learning methods in medical diagnosis that uses MRI modality along with the recent developments and various challenges in deep learning related to analysis of medical images.

The accuracy level in diagnosing tumor type through MRI results is required to establish appropriate medical treatment. MRI results can be computationally examined using K-Nearest Neighbor method, a basic science application and classification technique of image processing. Tumor classification system is designed to detect tumor and edema in T1 and T2 images sequences, as well as to label and classify tumor type. Data interpretation of such system derives from Axial section of MRI results only, which is classified into three classes: Astrocytoma, Glioblastoma, and Oligodendroglioma. To detect tumor area, basic image processing technique is employed, comprising of image enhancement, image binarization, morphological image, and watershed. Tumor classification is applied after segmentation process of Shape Extration Feature is undertaken. The results of tumor classification obtained was 89.5 percent, which is able to provide information regarding tumor detection more clearly and specifically.

In todays world many numbers of people are affected with neurological disorder. Neurological disease is a brain abnormalities or brain dysfunction. It affects all age group of people. In past years, the detection of neurological disorder is a very difficult task but now, we are blessed with many medical technologies which are available such as MRI, CT scan and PET. Convolutional neural network is best classifier for image feature extraction without any image segment. In this paper, we detect neurological disorder for normal as well as abnormal brain using brain MRI dataset. To trace the MRI dataset, we select a convolutional neural network algorithm to classify the MRI data for normal and abnormal brain.

Detection anatomical structure for many tasks such and identification of the of the human vertebrae is crucial as diagnosis of slipping and herniated discs. Recently many deep learning methods have been proposed to find automatically structure of vertebrae. Since there are very little vertebral images available in this area, end to end systems not feasible. In this paper, sequential phase based deep learning system used to detect and identify human lumbar discs from MRI. Every disc and every two neighbor discs are trained like Hidden Markov Model (HMM). System can detect and identify every lumbar disc in 0.5 seconds, with %89 accuracy and average 1.1 mm localization error. These results shows that proposed method can he comparable with state of art.

Studying the gender differences of brains is important to understand the brain cognitive mechanism. With the increasing amounts of brain imaging data, researchers start to use deep learning for brain image processing. However, the interpretability of deep models is the big gap between them. At present, deep model visualization is the main method to solve the interpretability problem of deep networks. Class Activation Map (CAM) is a commonly-used deep model visualization method. But the resolution of CAMs are restricted since it only uses the last layer for visualization. In this paper, we proposed a novel convolutional network named Dense-CAM by combining DenseNet and CAM and realized the visualization of the whole network so that it can generate more accurate and more robust deep model visualization. The network was tested with the gender classification problem using more than 6000 samples and achieved an accuracy of 92.93%. Brain regions with significant differences between men and women are found with the proposed method, which can be used for future brain imaging studies.

More noteworthy test in brain tumor is arranging and huge assessment is assurance of the tumor degree. The non invasive magnetic resonance imaging (MRI) system has risen as a cutting edge analytic device for brain tumors without ionizing radiation. Cerebrum tumor degree division by manually from 3D MRI volumes is a tedious assignment and execution is exceptionally depended on administrator's involvement. In specific circumstance, dependable completely programmed division strategy for the brain tumor division is important for a productive estimation of tumor degree. To investigate this, we propose a completely programmed strategy for brain tumor division that is created utilizing U-Net based deep convolutional network.