Artificial Intelligence in Tissue and Organ Regeneration

By Chandra P. Sharma

Artificial Intelligence in Tissue and Organ Regeneration discusses the role of artificial intelligence as a highly sought-after technology in the area of organ and tissue regeneration. Certain groups have made significant progress in mass producing mini organs and organoids from stem cells utilizing such techniques. As time goes on, there will be a need to improve these procedures, protocols, regulatory guidelines, and their clinical implications.

• Integrates existing literature in a highly interdisciplinary area
• Presents comprehensive current and future perspectives, combining artificial intelligence and machine learning with organ and tissue regeneration
• Provides new and emerging technology that is useful in healthcare and the medical field

• Language: English
• Publisher: Academic Press
• Release date: Aug 18, 2023
• ISBN: 9780443184994

Book preview

Preface

Tissue engineering is one of the fastest progressing areas leading to organ regeneration. Tremendous growth gives rise to the availability of extensive data; this makes the optimization of future experiments with precision an extremely time consuming, complex, and difficult. The support of artificial intelligence (AI) is very useful to resolve these issues with advances in AI algorithms, data-driven biomaterials prediction and synthesis, machine leaning (ML), and AI-assisted tissue-scaffold design and imaging. AI-driven robotic manufacturing can significantly accelerate organ regeneration in the coming years. This book will address the basic concepts and the applications of artificial intelligence for organ regeneration comprehensively.

This unique book is an interdisciplinary attempt to address such critical issues and to bridge the gap between clinical practitioners and biomedical researchers.

This book contains 16 chapters divided in three sections such as follows: Section 1 deals with ML and AI, Section 2 deals with AI applications with organ regeneration, and Section 3 deals with AI approaches with biomanufacturing. The contributing authors are from a versatile group of experts with backgrounds in academia, basic sciences, clinical medicine, and industry. Furthermore, this book includes contributions from Spain, Switzerland, India, and North America, helping to broaden the views in a global perspective on the development and application of AI in tissue-engineered organ regeneration for clinical applications. All chapters have been very comprehensive with latest knowledge in the specialized field. This is an excellent book which fills the gap for bioengineering graduate students, young faculty, researchers, and industrial partners.

We thank all the contributors for their excellent contributions, Ms. Elizabeth A. Brown for encouraging the project development at its every stage and Ms. Susan Ikeda, Elsevier Publishing, for her effective coordination.

We take this time to express our gratitude and thanks to our family members for sustained support during the course of this project.

Chandra P Sharma would like to thank and very much appreciate his wife Aruna Sharma for her continuous support during this entire project and also to Rishi, Neil, Agastya, Shreya Maitrey, and Varun for their refreshing support time to time.

Thomas Chandy would like to express his thanks to his wife Rachel Thomas for her understanding, encouragement, and continuing support to complete this project.

Vinoy Thomas acknowledges the constant support and encouragement from his wife Bini Mathew, PhD, senior scientist at the Southern Research Institute, Birmingham, Alabama, throughout the course of this book project.

Chandra P Sharma

Thomas Chandy

Vinoy Thomas

Section 1

Machine learning and artificial intelligence concepts

Outline

Chapter 1. Artificial Intelligence in tissue and organ regeneration: An introduction

Chapter 2. Introduction to artificial intelligence and machine learning algorithms

Chapter 3. ML and AI approaches for design of tissue scaffolds

Chapter 4. Use of artificial intelligence in assistive devices

Chapter 1: Artificial Intelligence in tissue and organ regeneration: An introduction

Willi Paul, and Chandra P. Sharma     Biomedical Technology Wing, Sree Chitra Tirunal Institute for Medical Sciences & Technology, Thiruvananthapuram, Kerala, India

Abstract

AI is a scientific field founded way back in 1956 and is now used in daily life, such as cyber security, customer relation management, internet searches, voice assistants, face recognition, and machine learning-based financial fraud detection. It emphasizes the acquisition of information and the rules needed and using that information rules to reach definite or approximate conclusions. AI can also change the way data is analyzed, tasks and activities performed, and the endpoints achieved. In tissue engineering, AI-aided design can predict the final mechanical, chemical, and biological properties of intelligent scaffolds to improve tissue growth and organ regeneration. Machine learning and AI are leading the scientific developments in tissue engineering, such as 3D bio printing, image reconstruction, image enhancement, object detection, image segmentation, and diagnosis. AI can help with personalized and faster decision-making, analyzing high-complexity information from many sources, and understanding the metabolic functions of cell under different conditions. AI is being used to study disease mechanisms and provide information on drug development and novel insights in cardiovascular diseases. It has already achieved high performance in diagnosis, predicting prognosis, epidemic prediction, and drug & vaccine development. AI is also being used in biomaterials research to create medical devices that are optimal for the patient. AI and ML also help in generating and analyzing large-scale data of cell, blood or tissue biomaterial interactions, which can narrow the choice of suitable materials for device development and accelerate the development of more personalized medical devices.

Keywords

Artificial intelligence; Biomaterials; Healthcare; Organ regeneration; Tissue engineering; Tissue regeneration

Introduction

Artificial Intelligence (AI) term was christened by John McCarthy in a 2-month Dartmouth Summer Research Project, held at Dartmouth College in Hanover, New Hampshire in 1956. This summer workshop was the founding event for Artificial Intelligence as a scientific field (McCorduck, 2004). In the 1950s many discussions were going on in the field of thinking machines. Cybernetics, automata theory, and complex information processing were the basics for insentient objects coming to life as intelligent beings. Ancient Egyptians and Greeks practiced automaton, a control mechanism designed to automatically follow a sequence of operations. The story of the giant bronze Talos is considered one of the earliest conceptions of a robot (ancient Greek). Talos was automated to protect the island of Crete from invaders. It guarded the island by hurling boulders at any enemy ships that approach and marches around three times daily. Pandora is another mythical artificial being that was believed to be programmed to infiltrate the human world and release her jar of miseries. It is believed that in the third century BCE, Chinese engineers and mathematicians developed animated puppets that played real music. It was programmed and operated by water-driven wheels, underwater chains, ropes, or paddle wheels enabling inanimate musicians, singers, acrobats, and animals to move and make music. Artificial intelligence is now one of the trending concepts in computer science and it can solve many real-life problems. AI emphasizes the acquisition of information and the rules needed and using that information rules to reach definite or approximate conclusions. There will be continuous self-correction to ensure that they offer the most accurate results. AI is now used in daily life, in cyber security, customer relation management, internet searches, voice assistants, face recognition, and machine learning (ML)-based financial fraud detection. It can also comprehend and categorize large volumes of clinical documents providing remarkable insight into detailed understanding, and improved methodologies, with significantly beneficial patient outcomes. AI on the other hand can be called intelligent automation that adds dynamism to the way humans and machines interact. It can change the way how data is analyzed, the tasks and activities performed, and the endpoints achieved.

Tissue engineering concepts have been widely studied in cartilage, skin, bone, vascular and nerve tissue, and organ regeneration. In tissue engineering, the combination of materials, the cell-cell interactions, and the cell-matrix interactions are very important like the scaffold's 3D structure and its physical properties. The pore size and structural distribution for 3D printed scaffolds are very crucial for tissue growth and organ regeneration. Tissue repair or healing is a natural, complicated, and continuous dynamic process in any living organism i.e., restoration of tissue function and architecture after the injury. However, the complex structure of organs cannot be exactly duplicated as the traditional scaffold fabrication technique has its limitations. 3D bioprinting technique has emerged as a strategy to improve the regeneration of organs. However, poor biocompatibility, undesirable degradation products, and incorporation of suitable legends are major challenges and require a painstaking optimization procedure. AI-aided design can predict the final mechanical, chemical, and biological properties of intelligent scaffolds so that the different chemical combinations of materials can be studied and optimized for the development of intelligent and dynamic scaffolds in tissue and organ regeneration. As reported, ML and AI will be leading the scientific developments in tissue engineering as shown in Fig. 1.1, in delivering intelligent and efficient medical devices (Pearce & Mikos, 2021). Artificial intelligence has been utilized for the analysis of high spatial resolution medical imaging including that of 3D printing in tissue engineering, particularly in organoids development (Gao et al., 2022). AI helps in image reconstruction, image enhancement, object detection, image segmentation, diagnosis, and prediction which leads to rapid development in tissue engineering. Full clinical adaption, however, requires a thorough evaluation of AI algorithms for avoiding potential biases and variations, because it has only been demonstrated in a small section of datasets at single imaging sites. However, compared to the traditional large-deformation-based finite element solution methods, analysis run times are significantly reduced when neural network methods are adopted in complex organ-level simulations in clinically relevant time frames with accuracy (Sacks et al., 2022).

AI in organ regeneration

Organ transplantation is one of the most challenging and complex fields in medicine and the prime impediment is implant failure by organ rejection by the body. Several organs that have been successfully transplanted include the heart, kidneys, liver, lungs, pancreas, intestine, thymus, and uterus. However, finding the best possible match remains an important criterion in the process. Organ matching characteristics and timeframe during which the organ remains viable are the significant factors that need to be taken into consideration for the allocation of donor organs for optimal use and equal access. Limited time for decision making, after analyzing these important parameters is also a limiting factor. AI could significantly help in supporting personalized and faster decision-making (Clement & Maldonado, 2021). Choosing an optimal immunosuppressant regimen will be easy with AI, analyzing high-complexity information from many sources. There is a continuing shortage of organs for transplantation in patients. Hence, the application of AI is also equally important in the case of xenotransplantation, particularly for the analysis of biopsy images.

Understanding the metabolic functions of a cell under different conditions is helpful in the optimization of in vitro cell growth environments, or in predicting cellular behavior in vivo and in vitro particularly to model the disease in a dish. Induced pluripotent stem cells (iPSC) are differentiated into disease-causing cell types, including cardiomyocytes (Yuasa, 2022) which can express the patient phenotype. This along with AI helps to study disease mechanisms and provide information on drug development and novel insights into cardiovascular diseases. This will certainly help in future regenerative therapy or organ regeneration. This will also help to better characterization of the subgroups of heterogeneous diseases. The utility of AI in organ transplant selection and organ regeneration is described in Chapters 5, 7 and 8.

Artificial intelligence in healthcare

Artificial intelligence is sure to enhance significantly the system efficiency in the life sciences and healthcare industries. This era in the healthcare sector had already initiated with the AI achieving high performance in diagnosis, predicting prognosis, epidemic prediction, and drug and vaccine development (Wang et al., 2021). It significantly helped in the radiological imaging of the vasculature, infected by COVID-19. According to the Markets and Markets 2021 survey, the growth of AI in the healthcare market is projected to reach USD 67.4 billion in 2027, an increase from USD 6.9 billion in 2021 (Artificial Intelligence in Healthcare Market, 2021). The latest statistics projected the global market share of AI in healthcare as USD 95.65 billion by the year 2028 (Artificial Intelligence in Healthcare Market, 2022) primarily because of its rising potential in genomics and drug discovery. Google's DeepMind AI could achieve success in detecting eye disease from scans and is developing AI research and mobile tools for cancer radiotherapy treatment and to predict patient deterioration (Liu et al., 2021). It is also used to predict the structure of proteins (Hassabis, 2022).

An early cancer diagnosis is recognized as the key priority by the world health organization and focuses on detecting symptomatic patients as early as possible. AI algorithms will assist clinicians in screening asymptomatic patients, triaging symptomatic patients, and diagnosing cancer recurrence (Hunter et al., 2022). It also helps in identifying anticancer targets and discovering novel drugs utilizing multiomics technologies. The five aspects of these technologies are shown in Fig. 1.2 are discussed in a recent review (You et al., 2022) and may fuel further advancement in the field. The AI techniques like support vector machine (SVM), logistic regression (LR), bagging tree, boosting tree, and dual-channel bidirectional long short-term memory (DC-BiLSTM) network have been successfully studied for the classification of benign versus malignant diagnosis of prostate cancer (Kim et al., 2021). Integrating artificial intelligence in imaging modalities helped to make better clinical decisions in the diagnosis of pancreatic cancer (Hameed & Krishnan, 2022), oral cancer (Hegde et al., 2022), lung cancer (Chassagnon et al., 2022), hepatocellular carcinoma (Martinino et al., 2022), breast cancer (Madani et al., 2022), etc. Radiomics with AI also helped the extraction of specific features on CT and MRI images, such as intensity, shape, and surface texture of malignancy, which otherwise go unnoticed. The significant challenges in manually interpreting the MRI images in the characterization of ischemic stroke, like heterogeneity of stroke pathogenesis, morphology, image acquisition modalities, sequences, and intralesional tissue signal intensity, as well as lesion-to-normal tissue contrast, etc., can be overcome by using ML techniques and is reported to be able to detect the infarcts with an acceptable accuracy of 70%–90% (Subudhi et al., 2022). AI finds application in image-guided radiation therapy (Niu et al., 2022), enabling image reconstruction for the accurate assessment of the radiation dose delivered. AI-based models of breast images helped in predicting breast cancer risk, the likelihood of tumor malignancy, and tumor response to treatment (Jones et al., 2022) driven by quantitative patient-specific data compared to the present generic qualitative markers.

Figure 1.1  Utilizing machine learning to design novel tissue engineering medical devices. Reproduced from Pearce, H. A., & Mikos, A. G. (2021). Machine learning and medical devices: The next step for tissue engineering. Engineering, 7(1), 1704, under Creative Commons License.

Figure 1.2  Artificial intelligence to integrate multiomics data (e.g., epigenetics, genomics, proteomics, and metabolomics) for cancer therapeutic target identification. Reproduced from You, Y., Lai, X., Pan, Y., Zheng, H., Vera, J., Liu, S., et al. (2022). Artificial intelligence in cancer target identification and drug discovery. Signal Transduction and Targeted Therapy, 7, 156, under Creative Commons License.

Application of AI in tissue engineering

Artificial neural networks were used way back in 2005 in vascular tissue engineering to obtain the best model configuration for the prediction of the best tissue engineering strategy with a predictive accuracy of 94.24% (Xu, Ge, Zhou, Yan, et al., 2005). During the same time, it's also been reported developing a scheme engine utilizing AI methods to generate different tissue engineering schemes in better curing of experimental animals (Xu, Ge, Zhou, & Yang, 2005). Since artificial intelligence systems can study very large data sets and make more precise decisions, it is predicted that AI can make tissue engineering easily acceptable in the medical market with reduced risk. Swarm Intelligence-based artificial neural network was developed in 2009 to predict the outcome of success in tissue engineering applications like peripheral nerve regeneration (Conforth et al., 2009). Understanding and theoretical analysis of the biological neural networks paved the way for learning, remembering, and adapting to the situations and inputs, allowing it to solve problems or make decisions. This led to the development of bio-inspired artificial neural networks.

The application of 3D convolutional neural networks (CNN) in predicting different essential properties of tissue engineering scaffolds has been reviewed by Bermejillo Barrera et al. (2021). The results obtained validate an AI-based methodology for predicting the properties of complex structures, which may be applicable when the computational cost of other simulation methods results are unaffordable, particularly a powerful resource for predicting the mechanical properties of innovative scaffolds. Predicting suitable printing conditions (material composition and printing parameters) has been achieved through the use of ML by researchers at Rice University (Conev et al., 2020). The speed of printing was found to be more critical among the parameters like material composition, pressure, layer, and spacing in affecting print quality. Experimental designs in bioink composition, bioprinting platforms, and material deposition strategies for fabricating vascularized tissues for tissue-engineered products such as skin, cornea, and cartilage were possible with computer simulation and artificial intelligence (Salg et al., 2022). Data-driven AI-based approaches are emphasized in the clinical protocols of cranioplasty treatment (Jegadeesan et al., 2022). 3D printing and tissue engineering will accelerate the design and manufacture of patient-specific cranial implants when surgery is performed in conjunction with ML techniques. This will help in less manual intervention and clinical complications with a shorter postoperative time that significantly reduces the cost of treatment. Future translation of 3D-printed complex tissues can be achieved by integrating AI and robot-assisted apparatus by controlling the mechanical features of bioinks. Applications of AI in the fabrication of tissue scaffolds and 3D bioprinting are discussed in Chapters 3 and 4. ML algorithms in the formulation of bioinks are discussed in Chapter 9.

Mimicking tumor microenvironments as cancer models helps in the development of preclinical therapeutics. However, the complexity and dynamic nature of cancer tissues are the limiting factors in successful drug development. AI significantly helps in analyzing, predicting, and interpreting the complex information of tissue-tissue, tissue-cell, and cell-cell interactions, fluid flows, and biomechanical information that were inaccessible by traditional methods (Das & Fernandez, 2022). Lab on Chip technology in drug discovery, in modeling physiological situations, generates big data and requires AI methods that can be significantly advantageous in analyzing massive datasets (Harofte et al., 2022).

Organoids are mini 3D tissue cultures derived from stem cells. Such self-organized cultures can replicate the complexity of human organs. The right environment for the stem cells forms tiny structures that resemble miniature organs that resemble the brain, kidney, lung, intestine, stomach, and liver. These mini-organs immensely help in modeling disease conditions, drug discovery, and therapeutic tools. However, the complexity of such organoid cultures poses a significant challenge for miniaturization and automation. The latest study by Freedman and colleagues (Czerniecki et al., 2018) establishes a robotic pipeline to manufacture and analyze organoids from human Pluripotent Stem Cells (hPSCs) in microwell formats and are capable of modeling complex human differentiation and disease states. This high throughput method also exhibits inherent species specificity and provides an attractive starting point for screening approaches focusing on therapeutic discovery, toxicology, and regenerative medicine. Increased demand for AI-aided automated organoid platforms is anticipated (Louey et al., 2021). AI-based applications are used to study the differential expression of genes for human brain organoids (Badai et al., 2020). Brain organoids can mimic the dynamic spatiotemporal process of early brain development, model various human brain disorders, and serve as an effective preclinical platform to test and guide personalized treatment. Automated software solutions enable the quantification and patterning of data derived from brain organoid models. These applications extract information about the cell count, cell size, cell shape, cell density, and dynamics within the cerebral model. Within organoid cell culture tiny particles known as microfluidic droplets are used to encapsulate induced Pluripotent Stem Cells (iPSCs). Microfluidic droplets and bubbles can be easily quantified with the help of advanced microscopy image analysis software. Powerful AI algorithms provide invaluable tools for the automated, fast, precise, and reproducible analysis of data generated with brain organoid structures on an easily accessible user interface. Utilizing AI in stem cell therapy has been discussed by Srinivasan et al. (2021), concluding that it helps in determining the viability, functionality, biosafety, and inefficacy of stem cells, as well as appropriate patient selection, and also significantly increases precision and accuracy. Applications of AI in stem cell therapies and the development of multiscale scaffolds for cancer organoids are discussed in Chapters 10 and 11, respectively.

AI in biomaterials evolution

Although a systematic theoretical approach of factorial designs was being applied in biomaterials research, the Edisonian approach also continues to be a successful strategy. The constant evolution and innovation in biomaterial field, however, requires a more comprehensive approach as the human system is extremely dynamic. The conventional approaches in the development of biomaterials make the process very expensive and strenuous, making the translation of medical devices in clinical practice a slow process. Thus, biomaterials research needs a high-throughput data-based dynamic approach to successfully create medical devices that are optimal for the patient (Suwardi et al., 2022). The applications of ML and AI are being investigated now on a large scale in biomaterial science for materials and medical device development. In a recent study, stimuli-sensitive bottlebrush polymers were designed and synthesized utilizing CNN model (Joshi et al., 2022). The softness and flexibility of these materials make it challenging to study the stimuli-sensitive shape changes and control their properties. CNN was utilized to identify and predict similarities in shape and function in polymers that helped to control the shape changes. AI and ML also help in generating and analyzing large-scale data of cell, blood, or tissue biomaterial interactions. This can narrow done the choice of suitable materials for device development and also accelerate the development of more personalized medical devices. The water repulsion and nature of protein adsorbed on any surface decides the ultimate fate of any material when it comes in contact with blood. Even the protein adsorption affinity and the material's hydrophilicity or hydrophobicity can be predicted using AI, which helps in designing organic materials with desired functions (Kwaria et al., 2020). Polymers are widely used to manufacture medical devices and are used in drug delivery, tissue engineering, and medical diagnostics. It requires to exhibit specific properties based on its application. The varied chemical and morphological complexities of polymers are a big challenge in specific applications. The prediction of polymer properties and its uncertainties based on the similarity in the database of Polymer Genome Project has been achieved utilising ML (Kim et al., 2018). Polymer Genome is an informatics platform for polymer property prediction and design using machine learning (http://www.polymergenome.org). Polymers may be queried either using the drawing tool or by specifying SMILES string of repeat units (Tran et al., 2020). Advanced bio-manufacturing processes of AI applications are discussed in Chapter