Support vector machine based discrete wavelet transform for magnetic resonance imaging brain tumor classification

This study was to analyze the clinical application value of magnetic resonance imaging (MRI) image features based on intelligent algorithms in the diagnosis and treatment of breast cancer and to provide an effective reference assessment for breast cancer diagnosis. The MRI diagnosis model (ACO-MRI) based on the ant colony algorithm (ACO) was proposed, which was compared with the diagnosis methods based on support vector machine (SVM) and proximity (KNN) algorithm, and the proposed algorithm was applied to MRI images to diagnose breast cancer. The results showed that the accuracy, sensitivity, and specificity of the ACO-MRI model were greater than those of the KNN and SVM algorithm. Moreover, the specificity was statistically considerable compared with the two algorithms of KNN and SVM ( P < 0.05 ). By comparing 1/5 number of ants and the average gray path of the ACO-MRI model under 1/8 number of ants, it was found that the average gray path value of 1/8 number of ants was greatly higher than the average gray path value of 1/5 number of ants ( P < 0.05 ). The differences in the overall distribution of breast MRI imaging features among Luminal A, Luminal B, HER-2 overexpression, and TN were compared. There were considerable differences in the overall distribution of the three breast MRI imaging features of the boundaries, morphology, and enhancement methods among the four groups ( P < 0.05 ). In short, MRI image based on the intelligent algorithm ACO-MRI diagnosis model can effectively improve the diagnosis effect of breast cancer. Its image feature boundaries, morphology, and enhancement methods had good imaging features in the diagnosis of breast cancer.

Brain tumor is the phenomenal growth of abnormal tissues in the human brain. A malignant tumor is known as a cancerous cell, and it is the major cause of death among people across the globe. Nowadays, the detection and classification of brain tumors from Magnetic Resonance Images (MRI) is a very crucial task. This chapter addresses various computing methods such as Edge Detection (ED), feature extraction, and classification techniques for the detection and classification of brain tumor regions from MRI datasets. Initially, MRI images are collected, and then various pre-processing steps such as filtering and edge detection are applied. Then, different edge detection methods such as Sobel ED, Robert’s ED, Prewitt ED, Canny ED, Laplacian ED, and Laplacian of Gaussian (LoG) with sigma 3 are applied for the MRI datasets. Subsequently, segmentation techniques are applied for detecting tumor regions from MRI, and essential features are extracted using Discrete Wavelet Transform (DWT) method. Otsu’s thresholding and K-means clustering segmentation methods are used for the investigation. Further, the support vector machines (SVMs), Naïve Bayes (NB), K-nearest neighbors (KNN), Back Propagation Neural Network (BPNN), and feedforward neural networks (FFNN) classifiers are employed for the MRI classification purpose. The experiments are 146conducted using the brain tumor dataset in the MATLAB 2019a software environment. The experimental results are analyzed in multiple dimensions, and it shows that the SVM with Otsu’s thresholding method exhibits better performance with 86.11% accuracy during the classification.

Diabetic Retinopathy (DR) is a microangiopathic complication of diabetes mellitus that can cause visual impairment to permanent blindness. Early detection of DR is essential to prevent disease progression, but conventional methods require time, cost, and expertise that are not always available. This study aims to compare the performance of the Random Forest (RF) and Support Vector Machine (SVM) algorithms in DR classification based on texture features extracted from retinal fundus images. The dataset used consists of 3,000 retinal fundus images obtained from the Kaggle platform, divided into 2,400 training data and 600 test data. Image preprocessing includes conversion to grayscale, resizing to a resolution of 128×128 pixels, and normalization. Feature extraction is performed using a combination of Local Binary Pattern (LBP) and Gray Level Co-occurrence Matrix (GLCM) to produce a 14-dimensional feature vector. Performance evaluation uses accuracy, precision, recall, F1-score, ROC curve, and 5-fold cross-validation metrics. The results showed that Random Forest significantly outperformed SVM with an accuracy of 96% compared to 64%, an AUC value of 0.99 compared to 0.72, and an average cross-validation accuracy of 94.5% compared to 63.42%. Random Forest also showed balanced performance in both classes with precision, recall, and F1-score of 0.96, while SVM experienced classification imbalance especially in the disease class. This study proves that Random Forest is a more optimal algorithm for an automatic DR detection system based on fundus image texture features and can support increasing the accessibility of DR screening in areas with limited specialist medical personnel.

Efficient prediction and classification for cirrhosis disease using LBP, GLCM and SVM from MRI images

A new clinical diagnosis system for detecting brain tumor using integrated ResNet_Stacking with XGBoost

To identify tea categories more automatically and efficiently, we proposed an improved tea identification system based on Histogram Equalization (HE), Gray-Level Co-Occurrence Matrix (GLCM) and Support Vector Machine (SVM) algorithm. In our previous project, 25 images per class might be not enough to classify, and a small size of dataset will cause overfitting. Therefore, we collected 10 kinds of typical processed Chinese tea, photographed 300 images each category by Canon EOS 80D camera, and regarded them as a first-hand dataset. The dataset was randomly divided into training set and testing set, which both contain 1500 images. And we applied data augmentation methods to augment the training set to a 9000-image training set. All the images were resized to 256 * 256 pixels as the input of feature extraction process. We enhanced the image features through Histogram Equalization (HE) and extracted features from each image which were trained through Gray-Level Co-Occurrence Matrix (GLCM). The results show that the average accuracy reached 94.64%. The proposed method is effective for tea identification process.

Segmentation and classification of brain tumor are time-consuming and challenging chore in clinical image processing. Magnetic Resonance Imaging (MRI) offers more information related to human soft tissues that assists in diagnosing brain tumor. Precise segmentation of the MRI images is vital to diagnose brain tumor by means of computer-aided medical tools. Afterwards suitable segmentation of MRI brain tumor images, tumor classification is performed that is a hard chore owing to complications. Therefore, Gannet Aquila Optimization Algorithm_deep maxout network (GAOA_DMN) and GAOA_K-Net+speech enhancement generative adversarial network (GAOA_K-Net+Segan) is presented for classification and segmentation of brain tumor utilizing MRI images. Here, pre-processing phase performs noise removal from input image utilizing the Laplacian filter and also the region of interest (ROI) extraction is also carried out. Then, segmentation of brain tumor is conducted by K-Net+Segan, which is combined by Motyka similarity. However, K-Net+Segan for segmentation is trained by GAOA that is an amalgamation of Gannet Optimization Algorithm (GOA) and Aquila Optimizer (AO). From segmented image, features are extracted for performing classification phase. At last, brain tumor classification is conducted by DMN, which is tuned by GAOA and thus, output is obtained. Furthermore, GAOA_K-Net+Segan obtained better outcomes in terms of segmentation accuracy whereas devised GAOA_DMN achieved maximum accuracy, true negative rate (TNR) and true positive rate (TPR) of 92.7%, 94.5% and 91.5%.

Magnetic Resonance Imaging (MRI) might be a problematic assignment for tumor fluctuation and complexity because of brain image classification. This work presents the brain tumor classification system using Convolutional Gated Recurrent Neural Network (CGRNN) algorithm based on MRI images. The proposed tumor recognition framework comprises of four stages, to be specific preprocessing, feature extraction, segmentation and classification. Extraction of identified tumor framework features was accomplished utilizing Gray Level Co-occurrence Matrix (GLCM) strategy. At long last, the Convolutional Gated Recurrent Neural Network Classifier has been created to perceive various kinds of brain disease. The proposed framework can be effective in grouping these models and reacting to any variation from the abnormality. The entire framework is isolated into different types of phases: the Learning/Training Phase and the Recognition/Test Phase. A CGRNN classifier under the scholarly ideal separation measurements is utilized to decide the chance of every pixel having a place with the foreground (tumor) and the background. MATLAB software is used in the development of the simulation of the proposed system. The suggested method's simulation results show that the analysis of brain tumours is stable. It shows that the proposed brain tumor classifications are superior to those from brain MRIs than existing brain tumor classifications. The overall accuracy of the proposed system is 98.45%.

To determine the diagnostic accuracy of perfusion computed tomography (PCT) in the grading of cerebral glioma. Cross-sectional analytical study. Department of Radiology, Allied Hospital, Faisalabad, from January to June 2014. All the patients with untreated glioma had an initial non-contrast head CT and then PCT using 128 multidetector CTscanner. Perfusion maps of permeability surface (PS) and cerebral blood volume (CBV) were generated and measured. As control, a second volume of interest was placed in the contralateral healthy cortex. PCT parameters were compared with World Health Organization (WHO) glioma grades. Fifty patients of 30 - 70 years of age of both genders (mean 45.13 ±5.54), 31 (62%) males and 19 (38%) females were studied. These patients were classified as low-grade glioma group (22 patients) and high-grade glioma group (28 patients). PS showed the sensitivity of 95.45%, specificity of 92.86% and diagnostic accuracy of 94% in differentiating the low-grade and high-grade glioma by using a cut-off value of 3.6 ml/100 g/minute. By using a cut-off value of CBV of 2.08 (ml/100 g) among low-grade and high-grade glioma group, CBV showed the sensitivity of 77.3%, specificity of 89.3%, and diagnostic accuracy of 84%. The derived parameters (PS and CBV) correlate well with tumor histopathology, differentiating low-grade from high-grade gliomas. PS showed better accuracy for glioma grading.

—Segmenting tumor from MRI images is an essential but time consuming manual duty. Performing an automatic segmentation is a defying task since different forms of tumor tissue exist for diverse patients and in many cases the tumor is similar to the normal tissue. Various studies proposed earlier to handle the issue of precisely segmenting the tumor but they discard the degradations and their effect to the precision of the segmentation. This article provides a more precise segmentation process through the use of appropriate pre-processing algorithms. The authors studied many enhancement and restoration algorithms and selected the NL-means, Laplacian filter and histogram equalization to be used as preprocessing techniques. Experimental results showed that using a suitable preprocessing scheme would produce a better segmentation process. Index Terms — Medical image processing, Tumor segmentation, MRI images, Non-local Means, Histogram equalization, Laplacian filter. I. INTRODUCTION A tumor is a mass of tissue that grows out of control of the normal forces that regulates growth. Brain tumor has a higher mortality rate when compared to more common types of cancer. This underlines the need of accurate prognosis of brain tumor in health care industry [1]. MRI has become a widely-used method of high-quality medical imaging, especially in brain imaging [2]. Image segmentation refers to the process of partitioning a digital image into multiple segments, also known as super pixels. Image segmentation is an important process in computer vision and image processing, which divides an image into a number of disjoint regions such that the pixels have high similarity in order to use its different fields; science, medical, agriculture and industrial fields [3]. Besides, Magnetic Resonance Imaging (MRI) images are known to be corrupted by noise [4], blur [5] and low contrast artifact [6]. Therefore, this article discusses the issue of achieving a better segmentation through the use of proper image restoration and enhancement techniques. Moreover, the accurate and automatic segmentation of a brain tumor on MRI image is of great interest for assessing tumor growth and treatment responses, enhancing computer-assisted surgery, and constructing tumor growth models [7]. For those reasons, an automatic and reliable tumor segmentation method would be a useful tool since the medical image segmentation plays an instrumental role in clinical diagnosis [8]. The rest of the article sections are organized as the following: in section two, a brief elucidation about some related studies is presented as the used literature review. Moreover, detailed explanations about the employed algorithms are delivered in section three as a methodology. Also, the results of the proposed methodology are illustrated in section four as the experimental results. Likewise, a detailed discussion about the use of certain algorithms and its results is deliberated in section five. Finally, a brief closure about this article is provided in section six as a conclusion. II. LITERATURE REVIEW In this section, three recent studies are briefly described. The first study was presented by [9] which proposed a texture based analysis to identify irregularity in the brain and automatic segmentation using seeded region growing segmentation. The method of tumor detection based on the texture of the MRI is proposed to segment the abnormality from the regular adjacent, although small quantities misclassified, but the segmentation result is precise. Moreover, [10] suggested a strong technique for automatic detection of tumor for brain MRI images. He also proposed a new morphological technique which is used for the stripped skull region from the brain images. Detecting tumor precisely based on seeded region growing segmentation on an improved single seed point selection algorithm is

BackgroundAccurate preoperative glioma grading and isocitrate dehydrogenase (IDH) mutation status prediction via conventional magnetic resonance imaging (MRI) remain challenging. We aimed to evaluate the performance of longitudinal relaxation time in rotating frame (T1rho) mapping and amide proton transfer (APT)-weighted chemical exchange saturation transfer (CEST) for grading gliomas and predicting IDH mutation status using 5-T MRI.MethodsFifty patients with histopathologically confirmed glioma underwent 5-T MRI, including T2-weighted, T1-weighted, T1rho mapping, APT-weighted, and contrast-enhanced T1-weighted imaging. The T1rho and APT parameters were measured at the solid tumor portion, peritumoral edema, and contralateral normal-appearing white matter. In addition to absolute lesion values, the relative values (rT1rho and rAPT) were defined as the difference between the lesion and white matter. Differences between low-grade glioma (LGG) and high-grade glioma (HGG) and between IDH mutant-type and IDH wild-type were evaluated via the Mann-Whitney U test. Receiver operating characteristic (ROC) analyses were performed to assess the performance of each parameter and T1rho-plus-APT combination.ResultsAmong the 50 patients, 12 had LGG, 38 had HGG, 19 had the IDH mutant type, and 31 had IDH wild type. Compared with LGGs, HGGs exhibited significantly greater T1rho, rT1rho, APT, and rAPT values in both tumor and edema areas (P≤0.036). ROC analysis of individual parameters revealed that T1rho in edema achieved the highest area under the curve (AUC) in differentiating HGG from LGG (AUC, 0.974; sensitivity, 97.4%; specificity, 91.7%). With the combination of T1rho in edema and rAPT in tumor, the AUC further improved to 0.978 (sensitivity, 92.1%; specificity, 100%). Compared with the IDH mutant-type group, the IDH wild-type group had significantly greater T1rho and rT1rho values in tumor, T1rho and rT1rho values in edema, and APT and rAPT values in tumor (P≤0.003). T1rho in the edema achieved the highest AUC (0.883) in differentiating IDH wild type from IDH mutant type (sensitivity, 90.3%; specificity, 89.5%). Significant differences were noted between grade 2, 3, and 4 gliomas (P≤0.007).ConclusionsBoth T1rho mapping and APT-CEST can be used to grade gliomas and differentiate IDH mutational status. Combining these two techniques further aids in the differentiation of HGG from LGG.

Objective To investigate the value of computed tomography (CT)-magnetic resonance imaging (MRI) image fusion in target volume delineation of postoperative precise radiotherapy for gliomas. Methods Thirty-six patients newly diagnosed with gliomas were enrolled and received postoperative intensity-modulated radiotherapy. Landmark (L) combined with manual fusion (M) and maximization of mutual information (MI) was used for image fusion. Target volume and organs at risk were delineated based on CT images and fused images, respectively. Comparison of the volume was made by pairwise t test. The distance from a landmark on CT image to the corresponding one on MRI image was calculated. The volume method and geometric center method were used to calculate the degree of volume overlap and changes in central positions after image fusion. Results The L+ M+ MI method achieved a high registration accuracy in image fusion, with a registration error less than 2 mm. In patients with grade Ⅲ-IV gliomas, the CT images yielded significantly smaller gross tumor volume (GTV) and clinical target volume (CTV) than the MRI images (74.62±46.91 vs. 105.99±58.86 cm3,P=0.042; 304.03±130.05 vs. 387.94±150.12 cm3,P=0.040). After image fusion, the smallest change in central position occurred in the chiasma (1.32±1.42 mm), and the largest change occurred in CTV (7.99±11.06 mm), followed by GTV and the brain stem. Conclusions CT-MRI image fusion helps to reduce the uncertainty of target volume delineation in patients with gliomas, especially in those with edema and residual tumor after surgery. Key words: Tomography, X-ray computed; Magnetic resonance imaging; Image fusion; Brain neoplasms/radiotherapy

The segmentation, detection and extraction of malignant tumor regions from magnetic resonance (MR) images are challenging tasks in medical image analysis. Approaches based on machine and deep learning have been introduced, which performed better than traditional image processing methods. However, many approaches still show limited ability due to the complex dataset and image modalities. This study evaluated the deep learning approach's performance and traditional image processing algorithms for medical imaging segmentation, detection, and classification. The proposed system comprises multiple stages. The Median filters are used in the pre-processing step, and morphological operation and Otsu thresholding are used to segment MR images. Discrete Wavelet Transform (DWT) algorithm is considered in the extraction features, and their classification is executed by a convolutional neural network (CNN) and support vector machine (SVM) algorithms. The Mat lab has been used for simulation and experimental findings to evaluate the suggested method's performance on the brain's complex and highly 2D structures. The results show that the methodology is reliable and efficient, with 93.5% accuracy.

Objective: To investigate the perfusion characteristics of arterial spin labeling (ASL) in intracranial tumor and its application value in classification. Methods: The clinical, pathological and imaging data of 44 patients with gliomas confirmed by pathology were analyzed retrospectively, including 9 low grade gliomas, 15 high grade gliomas, 11 cases of meningiomas, 6 cases of neurilemmoma, 3 cases of metastatic tumors.Conventional plain scan, 3D- ASL and MRI dynamic enhanced imaging (DSC-MRI) were performed.The mean maximal cerebral blood flow (CBF) of the solid component of tumor was obtained based on the region of interest.Immunohistochemical staining was performed in 24 patients with glioma.The differences of cerebral blood flow map (CBF) and relative cerebral blood flow (rCBF) in 44 patients with intracranial tumors were compared. The results of paired t test between the tumor area and the contralateral mirror area were measured by the two methods. Results: Taken the normal control-lateral grey matter(GM) as reference to normalize the CBF of tumor, three normalized tumor blood flow (nTBF) acquired by ASL showed statistical difference between low grade and high grade gliomas respectively (P<0.05). While taken the mirror region (M) and normal control-lateral white matter (WM) as reference to normalize the CBF of tumor, it showed no statistical difference (P>0.05). There was no 1p deletion in the cases of ASL perfusion in low-grade glioma group.In the case of 1p deletion in high grade glioma group, ASL was low perfusion, and there was no 1p deletion in the cases of ASL perfusion. Conclusion: 3D ASL can be used to identify high-grade and low-grade gliomas which has important reference value in the qualitative diagnosis of brain tumors and preoperative grading of gliomas.A separate use of 3D-ASL might cause over-or underestimation of tumor diagnosis, therefore a comprehensive analysis is needed.

Brain tumor detection and segmentation from Magnetic Resonance Imaging (MRI) images is being one of the emerging fields in the biomedicine. A formidable undertaking in brain tumor surgery, medical care, treatment programme and quantitative assessment of MRI images is to precisely diagnose its location and extent. Recently, the convolutional neural network (CNN) based detection and segmentation method on brain tumor MRI images is being one of the emerging fields in the medical imaging as an automatic clinic treatment and evaluation solution. In this article, we put forward a brand new quadruplet loss in CNN framework, which achieves higher accuracy in brain tumor detection and segmentation than other pairwise loss and triplet loss methods. By applying the proposed quadruplet loss to the original L2Net CNN architecture leads to a more compact descriptor named QuadrupletNet. From our experiments, QuadrupletNet shows higher performance than other state-of-the-art loss functions e.g., the Triplet loss, as indicated in experiments on Multimodal Brain Tumor Image Segmentation (BRATS 2018) datasets, and on our own collected MRI brain tumor datasets (named MBTD).