Deep Learning Applications in Medical Imaging

Machine learning is a technique of parsing data, learning from that data, and then applying what has been learned to make informed decisions. Deep learning is actually a subset of machine learning. It technically is machine learning and functions in the same way, but it has different capabilities. The main difference between deep and machine learning is, machine learning models become well progressively, but the model still needs some guidance. If a machine learning model returns an inaccurate prediction, then the programmer needs to fix that problem explicitly, but in the case of deep learning, the model does it by itself. Automatic car driving system is a good example of deep learning. On other hand, Artificial Intelligence is a different thing from machine learning and deep learning. Deep learning and machine learning both are the subsets of AI.

Deep Learning Applications in Medical Imaging

Deep learning (DL) and artificial intelligence (AI) are emerging tools in the healthcare sector for medical diagnostics. This chapter elaborates on general reasons for the popularity of computational techniques such as deep learning and machine learning (ML) applications in the medical image processing domain. The initial part of this chapter focuses on reviewing the fundamental concepts of DL algorithms, competence with machine learning, need in healthcare, applications, and challenges in medical image processing. Doing so allows understanding the reasons for the construction of all of them and offers a different view on various domains in the medical sector. The tools and technology required for DL, selection, implementation, optimization, and testing are discussed with respect to an application of cancer detection. Thus, this chapter gives an overall vision of deep learning concepts related to biomedical research.

A CONVblock for Convolutional Neural Networks

The reduction in the size of convolution filters has been shown to be effective in image classification models. They make it possible to reduce the calculation and the number of parameters used in the operations of the convolution layer while increasing the efficiency of the representation. The authors present a deep architecture for classification with improved performance. The main objective of this architecture is to improve the main performances of the network thanks to a new design based on CONVblock. The proposal is evaluated on a classification database: CIFAR-10 and MNIST. The experimental results demonstrate the effectiveness of the proposed method. This architecture offers an error of 1.4% on CIFAR-10 and 0.055% on MNIST.

Relevance of Machine Learning to Cardiovascular Imaging

Artificial intelligence (AI) broadly concerns analytical algorithms that iteratively learn from big datasets, allowing computers to find concealed insights. These encompass a range of operations comprising several terms, including machine learning(ML), cognitive learning, deep learning, and reinforcement learning-based methods that can be used to incorporate and comprehend complex biomedical and healthcare data in scenarios where traditional statistical approaches cannot be implemented. For cardiovascular imaging in particular, machine learning guarantees to be a transformative tool that can address many unmet needs for patient-specific management, accurate prediction of disease progression, and the tracking of identifiable biomarkers of disease processes. In this chapter, the authors discuss fundamentals of machine learning algorithms for image analysis in the cardiovascular system by evaluating the need for ML in this field and examining the potential obstacles and challenges of implementation in the context of three common imaging modalities used in cardiovascular medicine.

Deep Learning for Medical Image Segmentation

Pixel accurate 2-D, 3-D medical image segmentation to identify abnormalities for further analysis is on high demand for computer-aided medical imaging applications. Various segmentation algorithms have been studied and applied in medical imaging for many years, but the problem remains challenging due to growing a large number of variety of applications starting from lung disease diagnosis based on x-ray images, nucleus detection, and segmentation based on microscopic pictures to kidney tumour segmentation. The recent innovation in deep learning brought revolutionary advances in computer vision. Image segmentation is one such area where deep learning shows its capacity and improves the performance by a larger margin than its successor. This chapter overviews the most popular deep learning-based image segmentation techniques and discusses their capabilities and basic advantages and limitations in the domain of medical imaging.

Malaria Parasites Detection Using Deep Neural Network

Malaria is a dreadful infectious disease caused by the bite of female Anopheles mosquito, by the protozoan parasites of the genus Plasmodium. It's an epidemic disease and demands rapid and accurate diagnosis for proper intervention. Microscopic test on the thick and thin blood smear to detect the malaria and counts the infected cells is the gold standard for diagnosis of this disease. An automation process in the form of computer-aided diagnosis is much needed as it plays a vital role in fully or semi-automated diagnosis of diseases based on medical image information. Deep learning has vast ranging applications. This work is to build a convolutional neural network to expertly detect the presence of malaria parasitized cells in the thin blood smear. The authors construct the model as small and computationally efficient to obtain the highest level of accuracy possible.

Conventional and Non-Conventional ANNs in Medical Diagnostics

Computer technology advancements in recent days have offered professionals in different fields the ability to gather data, process information, store, and retrieve at a faster rate and make effective decisions. The large collection of data among all various applications including medical diagnosis has paved the need to employ advanced artificial neural networks (ANN). This chapter provides a detailed working view of ANN, covering its various architectures and design techniques in brief. A detailed analysis and summary of medical diagnostics applications using various ANN techniques will be leveraged. Imbalanced data is the major problem with medical data. This chapter briefs on the various methods to handle imbalanced data. Finally, future directions and potential current challenges are suggested for additional applications in neural networks.

Automatic Lung Tuberculosis Detection Model Using Thorax Radiography Image

Tuberculosis (TB) is a communal disease with high death and disease rates worldwide. The chest radiograph (CXR) is commonly used in diagnostic solutions for lung TB. Automatic computer-aided solutions to identify TB using CXRs and can advance the efficiency of the diagnostic of TB. In this chapter, an automatic TB detection model using CXR image is proposed. By identifying open issues include how detect the lung region automatically and what are the features, one can identify if a given CXR image is infected or normal using three public datasets such as Schengen, Montgomery Country (MC), and JSRT. The possible textural features of a lung object are obtained from the first-order and second-order gray level co-occurrence matrix (GLCM) statistical features. The performance of the proposed model was evaluated using accuracy, sensitivity, and specificity, and the model achieved AUC 91%, 62%, 71%, and 81% on Schengen, JSRT, MC, and combined datasets.

A Survey on Prematurity Detection of Diabetic Retinopathy Based on Fundus Images Using Deep Learning Techniques

Diabetic retinopathy (DR) is a disease related to eye correlated with long-standing diabetes. It is a leading cause of blindness among working adults. Detection of this condition in the early stage is critical for good prognosis. Present day detection of DR normally requires digital fundus image or images generated using optical coherence tomography (OCT). As OCT are high-priced, diagnosis of DR using fundus image will benefit for the patient and the ophthalmologists. Manual inspection of morphological changes in blood vessels, microaneurysms, exudates, hemorrhages, and macula are time consuming and tedious tasks. So, designing a computer-aided system helps in analyzing the morphological changes and identifying the DR. This chapter reviews the applications of machine learning and deep learning algorithms for detection of nonproliferative diabetic retinopathy by analyzing fundus images.

Machine Learning for Prediction of Lung Cancer

This work is focused on lung cancer prediction using machine learning technique. Lung cancer is one of the widespread diseases due to the growth of irregular cell in both the lungs as a result of which this irregular cell starts growing into tumour, and this tumour can be cancerous as well as non-cancerous. In the traditional approach CT scan images has been used based on the report image segmentation has been done to remove the noise so that a clear picture can be generated to detect the location of tumor. Once the location is known then classification or clustering approach can be used to predict the stage of cancer. Previously supervised machine learning algorithm has been used to predict lung cancer. In this work a prediction model is proposed that is based on the median filter, watershed segmentation, and then feature extraction has done like texture and region. And on the extracted feature classification technique was applied for prediction of cancer.