Integrating Metaheuristics in Computer Vision for Real-World Optimization Problems (eBook)

Kapil Joshi, Shubham Mahajan, Amit Kant Pandit, Nitish Pathak (Herausgeber)

eBook Download: EPUB

2024
646 Seiten
Wiley (Verlag)
978-1-394-23093-8 (ISBN)

Infos

A comprehensive book providing high-quality research addressing challenges in theoretical and application aspects of soft computing and machine learning in image processing and computer vision.

Researchers are working to create new algorithms that combine the methods provided by CI approaches to solve the problems of image processing and computer vision such as image size, noise, illumination, and security. The 19 chapters in this book examine computational intelligence (CI) approaches as alternative solutions for automatic computer vision and image processing systems in a wide range of applications, using machine learning and soft computing.

Applications highlighted in the book include:

• diagnostic and therapeutic techniques for ischemic stroke, object detection, tracking face detection and recognition;
• computational-based strategies for drug repositioning and improving performance with feature selection, extraction, and learning;
• methods capable of retrieving photometric and geometric transformed images;
• concepts of trading the cryptocurrency market based on smart price action strategies; comparative evaluation and prediction of exoplanets using machine learning methods; the risk of using failure rate with the help of MTTF and MTBF to calculate reliability; a detailed description of various techniques using edge detection algorithms;
• machine learning in smart houses; the strengths and limitations of swarm intelligence and computation; how to use bidirectional LSTM for heart arrhythmia detection;
• a comprehensive study of content-based image-retrieval techniques for feature extraction;
• machine learning approaches to understanding angiogenesis;
• handwritten image enhancement based on neutroscopic-fuzzy.

Audience

The book has been designed for researchers, engineers, graduate, and post-graduate students wanting to learn more about the theoretical and application aspects of soft computing and machine learning in image processing and computer vision.

Kapil Joshi, PhD, is an assistant professor in the Computer Science & Engineering Department, Uttaranchal Institute of Technology in Dehradun, India. His doctorate was on image quality enhancement using fusion techniques. He has 8 years of academic experience and has published patents, research papers, and two books. In 2021, he was awarded the 'Best Young Researcher' Award in Global Education and Corporate Leadership received by Life Way Tech India Pvt. Ltd.

Shubham Mahajan, PhD, is an assistant professor in the School of Engineering at Ajeekya DY Patil University, Pune, Maharashtra, India. He has eight Indian, one Australian, and one German patent to his credit in artificial intelligence and image processing. He has authored/co-authored more than 50 publications including peer-reviewed journals and conferences. His main research interests include image processing, video compression, image segmentation, fuzzy entropy, nature-inspired computing methods with applications in optimization, data mining, machine learning, robotics, and optical communication.

Amit Kant Pandit, PhD, is an associate professor in the School of Electronics & Communication Engineering Shri Mata Vaishno Devi University, India. He has authored/co-authored more than 60 publications including peer-reviewed journals and conferences. He has two Indian and one Australian patent to his credit in artificial intelligence and image processing. His main research interests are image processing, video compression, image segmentation, fuzzy entropy, and nature-inspired computing methods with applications in optimization.

Nitish Pathak, PhD, is an associate professor in the Department of Information Technology, Bhagwan Parshuram Institute of Technology, New Delhi, India. He has 17 years of engineering education experience and has published more than 80 journal articles, in peer-reviewed journals as well as book chapters, patents, and conference papers. His research areas include intelligent computing techniques, empirical software engineering, and artificial intelligence.

Textauszug

1
Advancement in Diagnostic and Therapeutic Techniques for Ischemic Stroke

Mukul Jain1,2, Divya Patil2, Shubham Gupta3 and Shubham Mahajan4,5,6*

1Cell & Developmental Biology Lab, Research and Development Cell, Parul University, Vadodara, Gujarat, India

2Department of Life Sciences, Parul Institute of Applied Sciences, Parul University, Vadodara, Gujarat, India

3Department of Paramedical and Health Sciences, Faculty of Medicine, Parul University, Vadodara, Gujarat, India

4School of Engineering, Ajeenkya D Y Patil University, Pune, Maharashtra, (iNurture Education Solutions Pvt. Ltd., Bangalore), Pune, Maharashtra, India

5University Center for Research & Development (UCRD), Chandigarh University, Mohali, India

6Hourani Center for Applied Scientific Research, AI-Ahilyaya Amman University, Amman, Jordan

Abstract

Ischemic stroke is a cerebrovascular disease that is typically brought on by a disruption in the blood flow to the brain; it accounts for more than 80% of all strokes. 1.8 million neurons are thought to die per minute. The burden of stroke in people younger than 65 years has increased over the last decades, it has increased worldwide by 25% among adults aged 20 to 64 years. According to reports, both men and women can experience stroke symptoms, in this case women are more prone to have nontraditional symptoms of stroke like dizziness, lack of consciousness, slurred speech. In comparison to Western Europe, America, Australia, and similar to Eastern Europe, Asia has a higher stroke fatality rate. Acute management of stroke requires fast and efficient screening, imaging and modality. Better clinical outcomes can be attained through early detection and treatment of stroke. Stroke imaging techniques are computed tomography (CT) scans and magnetic resonance imaging (MRI). MRI is more preferred because it gives detailed and exact result of damage. Antithrombotic and neuroprotective therapies are the mainstays of conventional medicine, although their use is still constrained due to their poor safety. Utilizing nanomedicines is an additional option since they can boost therapeutic benefit and adverse effect reduction, achieve effective medicine collection at the target site, and improve pharmacokinetic behavior of pharmaceuticals in vivo. We have comprehensively described ischemic stroke, its epidemiology, advanced artificial intelligence–based diagnostic tools and its treatment therapies.

Keywords: Ischemic stroke, computed tomography, MRI, artificial intelligence, diagnostic tools, PET, EEG, nanomedicine

1.1 Introduction

Stroke is a fatal and disabling illness that affects more than 15 million individuals annually worldwide. It is an ageing disease that primarily affects persons over the age of 65 years [1, 2]. In the US, ischemic strokes, which are caused by a decrease in brain blood circulation, account for 87% of all cases [2]. According to a report released by WHO in 2016, strokes are the second most frequent cause of impairment in the population today worldwide (Figure 1.1). With the passing of each decade during the previous 40 years, it has been revealed that the number of stroke cases multiplied [11]. Both the incidence and prognosis of stroke are influenced by gender; nevertheless, overall, women have a higher stroke prevalence than men do due to aging-related increases in stroke risk and longer average lifespans for women [3]. Basic stroke classifications include hemorrhagic, transient ischemic attack, and ischemic strokes (Figure 1.2). Nearly 80% of strokes are ischemic, while various populations experience hemorrhagic and ischemic strokes more frequently in varied proportions [4]. The following categories—etiologic subtypes—have been used to further categorize ischemic strokes: cardioembolic, atherosclerosis, lacunar, other particular causes (dissections, vasculitis, specific hereditary illnesses, etc.), and strokes with no known etiology [5]. Prior to the main stroke episode, individuals always have ministrokes, commonly referred to as transient ischemic attacks (TIA). The majority of published research relies on MRI and CT scan pictures to categorize strokes, which is a costly method of early stroke detection. CT and MRI are employed in the majority of investigations that identify strokes. These days, noninvasive methods are becoming more and more common. One such method is the electroencephalograph (EEG), which is also a cheap or free method. The following issues have impeded the advancement of novel therapeutics for ischemic stroke and numerous other illnesses of the central nervous system (CNS): blood–brain barrier (BBB) prevents most CNS drugs from reaching the brain parenchyma effectively, which makes it difficult to treat or save areas of the brain that haven’t yet been completely affected; (2) many drugs have poor stability or toxicity after systemic and/or oral administration; (3) the physiopathology of the disease is not well understood; and (4) it is challenging to translate positive results from preclinical studies to the clinic [6]. One of the key problems in the treatment of stroke is the blood brain barrier. In order to transport enough medicine to the brain tissue, liposomes must be able to cross the BBB and it is ideal for them to stay in the bloodstream for a long time [7]. New stroke diagnostic methods have been created thanks to nanotechnology and artificial nanomaterials. Identifying the biochemical and pathophysiological causes of stroke may be made easier with the help of this technology [8]. Due to their distinct fluorescence characteristics, which enable high spatiotemporal accuracy at biologically relevant concentrations, optical CNT-based biosensors exhibit considerable potential [9, 10]. With the development of nanotechnology, it is now possible to carry drugs to the brain and, more precisely, the area that is ischemia, more efficiently [8]. The FDA has only approved one therapeutic drug—recombinant tissue plasminogen activator (tPA)—for the treatment of ischemic stroke, and its use is constrained by its limited therapeutic window, quick drug elimination, and potential for hemorrhagic transformation [138]. There is no particular cure in medical science to deal with strokes; therefore, early identification is the key to deal with strokes. The early identification can inhibit the disabilities, loss of deaths, and other brain-related severe problems [12].

Figure 1.1 The data represent country wise occurrence of stroke and mortality rate in male and female across the world. (a) Incidence of stroke worldwide, in which china has higher percentage rate that is 9% and lower rate in Saudi Arabia that is 1%. (b) Prevalence of ischemic stroke across the world covering 28 countries, the higher incidence of ischemic stroke is 10% and low rate is 2%. (c) The percentage of mortality rate of stroke in male across the globe is 4%, high mortality rate is reported in Madagascar that is 9%. (d) The percentage of mortality rate in female across the world, high mortality is noted in Madgascar, Afghanistan, and Indonesia that is 9%.

Figure 1.2 It demonstrates the different types of stroke occurrence and how it causes in the brain and disturbs the brain function.

1.2 Diagnostic Tools of Ischemic Stroke

1.2.1 Preimaging

Patients with acute neurological conditions require quick neuroimaging. American Stroke Association recommendations recommend that the only prior inquiry is an actual capillary blood glucose, which is retrieved from paramedics. What is an intravenous cannula for contrast or perfusion imaging is frequently necessary sequences, enabling the simultaneous collection of a blood panel. This often includes a coagulation screen, a renal function test, and an infection test if the patient uses anticoagulants. Though many departments of radiology demand a current renal function prior to providing contrast [10].

1.2.2 Imaging

1.2.2.1 Computed Tomography Scan

Neuroimaging in the setting of a hyperacute severe stroke is still primarily based on CT. A noncontrast CT scan of head is rapid, delicate, and cost-effective excluding cerebral bleeding, which is typically sufficient to make judgments on thrombolysis. Due to widespread availability and significantly shorter turnaround times, CT perfusion investigations are extraeffective in the accident scenario to detect and identify stroke patients who can profit from thrombolytic therapy and vascular blood flow [134, 135] CT perfusion sequences can evaluate a variety of brain perfusion is frequently performed using artificial intelligence, like MIstar (Apollo Medical Imaging Technology) or iSchemaView’s Rapid Processing of Perfusion and Diffusion (RAPID) CT perfusion. The contrast bolus-tracking approach known as CT perfusion can be used with almost any multidetector CT scanner that is now in use in emergency rooms all around the world [131, 132]. A crucial factor in determining the dangers and possible benefits of reperfusion treatments, ideal CT perfusion values and thresholds that characterize irreversible infarction have not received enough attention in research [133]. These technologies make interpretation easier by ensuring the use of validated thresholds and enhancing interobserver reproducibility [10]. The sensitivity of the CT signals is between 40% and 60% within the first 3 hours following the beginning of symptoms,...

Produkt-Details

• Erscheint lt. Verlag: 1.8.2024
• Sprache: englisch
• Themenwelt: Mathematik / Informatik ► Informatik ► Theorie / Studium
• Schlagworte: 3D images • Cluster • computer vision • Deep learning • feature extraction • Heuristic • Image • Image Processing • machine learning • medical image • Object detection • Open AI • Pattern • Sensor • Vision
• ISBN-10: 1-394-23093-1 / 1394230931
• ISBN-13: 978-1-394-23093-8 / 9781394230938