The Image segmentation is referred to as one of the most important processes of image processing. Image segmentation is the technique of dividing or partitioning an image into parts, called segments. It is mostly useful for applications like image compression or object recognition, because for these types of applications, it is inefficient to process the whole image. So, image segmentation is used to segment the parts from image for further processing. There exist several image segmentation techniques, which partition the image into several parts based on certain image features like pixel intensity value, color, texture, etc. These all techniques are categorized based on the segmentation method used. In this paper the various image segmentation techniques are reviewed, discussed and finally a comparison of their advantages and disadvantages is listed.

Various Image Segmentation Techniques: A Review

Compared with common deep learning methods (e.g., convolutional neural networks), transfer learning is characterized by simplicity, efficiency and its low training cost, breaking the curse of small datasets. Medical image analysis plays an indispensable role in both scientific research and clinical diagnosis. Common medical image acquisition methods include Computer Tomography (CT), Magnetic Resonance Imaging (MRI), Ultrasound (US), X-Ray, etc. Although these medical imaging methods can be applied for non-invasive qualitative and quantitative analysis of patients—compared with image datasets in other computer vision fields such like faces—medical images, especially its labeling, is still scarce and insufficient. Therefore, more and more researchers adopted transfer learning for medical image processing. In this study, after reviewing one hundred representative papers from IEEE, Elsevier, Google Scholar, Web of Science and various sources published from 2000 to 2020, a comprehensive review is presented, including (i) structure of CNN, (ii) background knowledge of transfer learning, (iii) different types of strategies performing transfer learning, (iv) application of transfer learning in various sub-fields of medical image analysis, and (v) discussion on the future prospect of transfer learning in the field of medical image analysis. Through this review paper, beginners could receive an overall and systematic knowledge of transfer learning application in medical image analysis. And policymaker of related realm will benefit from the summary of the trend of transfer learning in medical imaging field and may be encouraged to make policy positive to the future development of transfer learning in the field of medical image analysis.

A Review of Deep Learning on Medical Image Analysis

The technology of image segmentation is widely used in medical image processing, face recognition pedestrian detection, etc. The current image segmentation techniques include region-based segmentation, edge detection segmentation, segmentation based on clustering, segmentation based on weakly-supervised learning in CNN, etc. This paper analyzes and summarizes these algorithms of image segmentation, and compares the advantages and disadvantages of different algorithms. Finally, we make a prediction of the development trend of image segmentation with the combination of these algorithms.

Image Segmentation Algorithms Overview.

Today, most of the people are affected by lung cancer, mainly because of the genetic changes of the tissues in the lungs. Other factors such as smoking, alcohol, and exposure to dangerous gases can also be considered the contributory causes of lung cancer. Due to the serious consequences of lung cancer, the medical associations have been striving to diagnose cancer in its early stage of growth by applying the computer-aided diagnosis process. Although the CAD system at healthcare centers is able to diagnose lung cancer during its early stage of growth, the accuracy of cancer detection is difficult to achieve, mainly because of the overfitting of lung cancer features and the dimensionality of the feature set. Thus, this paper introduces the effective and optimized neural computing and soft computing techniques to minimize the difficulties and issues in the feature set. Initially, lung biomedical data were collected from the ELVIRA Biomedical Data Set Repository. The noise present in the data was eliminated by applying the bin smoothing normalization process. The minimum repetition and Wolf heuristic features were subsequently selected to minimize the dimensionality and complexity of the features. The selected lung features were analyzed using discrete AdaBoost optimized ensemble learning generalized neural networks, which successfully analyzed the biomedical lung data and classified the normal and abnormal features with great effectiveness. The efficiency of the system was then evaluated using MATLAB experimental setup in terms of error rate, precision, recall, G-mean, F-measure, and prediction rate.

Automatic detection of lung cancer from biomedical data set using discrete AdaBoost optimized ensemble learning generalized neural networks

Timely prediction of cardiovascular diseases with the help of a computer-aided diagnosis system minimizes the mortality rate of cardiac disease patients. Cardiac arrhythmia detection is one of the most challenging tasks, because the variations of electrocardiogram(ECG) signal are very small, which cannot be detected by human eyes. In this study, an 11-layer deep convolutional neural network model is proposed for classification of the MIT-BIH arrhythmia database into five classes according to the ANSI-AAMI standards. In this CNN model, we designed a complete end-to-end structure of the classification method and applied without the denoising process of the database. The major advantage of the new methodology proposed is that the number of classifications will reduce and also the need to detect, and segment the QRS complexes, obviated. This MIT-BIH database has been artificially oversampled to handle the minority classes, class imbalance problem using SMOTE technique. This new CNN model was trained on the augmented ECG database and tested on the real dataset. The experimental results portray that the developed CNN model has better performance in terms of precision, recall, F-score, and overall accuracy as compared to the work mentioned in the literatures. These results also indicate that the best performance accuracy of 98.30% is obtained in the 70:30 train-test data set.

Automatic detection of arrhythmia from imbalanced ECG database using CNN model with SMOTE

Medical image segmentation is the most important process to assist in the visualization of the structure of importance in medical images. Malignant melanoma is the most frequent type of skin cancer but it is treatable, if diagnosed at an early stage. Dermoscopy is a non-invasive, diagnostic tool having great possibility in the early diagnosis of malignant melanoma, but their interpretation is time consuming. In this work, a new image segmentation algorithm, for the early diagnosis of the skin cancer, is proposed where the dermoscopic images are segmented using a threshold. This threshold selection is based on the Genetic Algorithm (GA) for region growth, followed by region merging procedure. The obtained segmented image is then compared with the ground truth image using various parameters such as False Positive Error (FPE), False Negative Error (FNE) Coefficient of similarity, spatial overlap. Genetic algorithm is a class of probabilistic optimization algorithms, powerful in finding optimal feature vectors. Identification of approximate global optimal region in Genetic Algorithm is a quick process. To merge regions with similar characteristics, we have used grey level and texture. The segmentation done through Genetic Algorithm is efficient when compared to the image segmentation done by conventional algorithms.

Image segmentation based on genetic algorithm for region growth and region merging

The heart is one of the crucial parts of a human being. The heart produces electrical signals and these cycles of electrical signals are called as cardiac cycles. The graphical recording of the cardiac cycle produced by an Electrocardiograph is called as Electro cardio gram (ECG) signal. The Electrocardiogram signal is used to diagnose the irregularity in heart beat. Automatic classification of ECG signals has applications in human-computer interaction, as well as in clinical application such as detection of key indicators of the onset of the certain illness. In this work an algorithm has been develop to detect the five abnormal beat signals includes Left bundle branch block beat (LBBB), Right bundle branch block beat (RBBB), Premature Ventricular Contraction (PVC), Atrial Premature Beat (APB) and Nodal (junction) Premature Beat (NPB) along with the normal beat. In order to prepare an appropriate input vector for the neural classifier several pre processing stages have been applied. Harr Wavelet Transform (HWT) is used in order to extract features from the ECG signal. Preprocessing and the classification of ECG signals is done using Forward Feed Neural Network Finally, the MIT-BIH [10] database is used to evaluate the proposed algorithm.

ECG signal feature extraction and classification using Harr Wavelet Transform and Neural Network

The heart is one of the crucial parts of a human being. The heart produces electrical signals and these cycles of electrical signals are normally called as cardiac cycles. The graphical recording of the cardiac cycles is known as Electrocardiogram (ECG) signal. The Electrocardiogram signal is used to diagnose the irregularity in heart beat which in turn can be used to identify heart problems. Automatic classification of ECG signals has applications in human-computer interaction, as well as in clinical application such as detection of key indicators of the onset of the certain illness. In this work, we have developed an algorithm to detect the five abnormal beat [2, 3] signals includes Left bundle branch block beat (LBBB), Right bundle branch block beat (RBBB), Premature Ventricular Contraction (PVC), Atrial Premature Beat (APB) and Nodal (junction) Premature Beat (NPB) along with the normal beat. In order to prepare an appropriate input vector for the neural classifier several pre processing stages have been applied. For efficient feature extraction we use hybrid feature extractor. The hybrid feature extraction is done in three steps, (i) Morphological based feature extraction (ii) Haar wavelet based feature extraction (iii) Tri-spectrum based feature extraction. The classification is done using Forward Feed Neural Network. Finally, the MIT-BIH [4] database is used to evaluate the proposed algorithm. The beat classification hybrid system (Hybrid + FFBN) based gives an accuracy is achieved 78 %, (Morp + FFBN) is achieved 62 %, (wavelet +FFBN) is achieved 65 %, (Spect + FFBN) is achieved 70 %, (Morp + Wavelet + FFBN) is achieved 62 %, (Morp + spect + FFBN) is achieved only 73 %.

Classification of ECG Signal Using Hybrid Feature Extraction and Neural Network Classifier

Problem statement: Malignant melanoma is the most frequent type of sk in cancer. Its incidence has been rapidly increasing over the last few decades. Medical image segmentation is the most essential and crucial process in order to faci litate the characterization and visualization of th e structure of interest in medical images. Approach: This study explains the task of segmenting skin lesions in Dermoscopy images based on intelligent s ystems such as Fuzzy and Neural Networks clustering techniques for the early diagnosis of Ma lignant Melanoma . The various intelligent systems based clustering techniques used were Fuzzy C Means Algorithm (FCM), Possibilistic C Means Algorithm (PCM), Hierarchical C Means Algorithm (HCM); C-mean based Fuzzy Hopfield Neural Network, Adaline Neural Network and Regression Neural Network. Results: The segmented images were compared with the ground truth image using various parameters such as False Positive Error (FPE), False Negative Error (FNE) Coefficient of si milarity, spatial overlap and their performance was evaluated. Conclusion: The experimental results show that Hierarchical C M eans algorithm( Fuzzy) provides better segmentation than other (Fuzzy C Means, Possibilistic C Means, Adaline Neural Network, FHNN and GRNN) clustering algorithms. Hierarchical C Means approach can handle uncertainties that exist in the data efficiently an d useful for the lesion segmentation in a computer aided diagnosis system to assist the clinical diagn osis of dermatologists.

/pdf/dermoscopic-image-segmentation-using-machine-learning-4dinhpjy1y.pdf

Dermoscopic Image Segmentation using Machine Learning Algorithm

One of the key challenges of face recognition is varying illumination. Many feature-based methods were developed in the recent years to detect the illuminant invariant facial features. This paper mainly focuses on the effect of various edge detection algorithms on illuminant variant images. The conventional sobel and adaboost are applied. Along with that, the proposed NSCT integrated ant colony optimization (ACO) approach is also used. The proposed method comprises a normal shrink filter in NSCT domain which produces illuminant invariant for the given image. Then, to capture the important geometrical structures and to reduce the feature dimensionality, ACO algorithm is performed. This combined approach fairly detects the edges with improved quality. Finally, for recognition, a graph matching algorithm is employed. This algorithm utilizes a group of feature points to explore their geometrical relationship in a graph arrangement. While applying the three methods to the YaleB database, experimental results show that the proposed work yields the better recognition.

S. H. Krishna Veni

Papers

Image segmentation based on genetic algorithm for region growth and region merging

ECG signal feature extraction and classification using Harr Wavelet Transform and Neural Network

Classification of ECG Signal Using Hybrid Feature Extraction and Neural Network Classifier

Dermoscopic Image Segmentation using Machine Learning Algorithm

An Analysis of Various Edge Detection Techniques on Illuminant Variant Images