| dc.description.abstract | Ubiquitous capture, digitization and integration of health data remains a grand challenge, promising to transform how we understand, diagnose and treat disease. To realize its full potential, electronics must capture and transform increasing volumes of diverse data into highly processed information. Unfortunately, systems finely-integrating thousands of sensors with compute and memory are currently infeasible. Since the materials and technologies leveraged to fabricate conventional computing, memory, and sensing are often distinct, systems today rely on separately packaged chips for each of these functionalities, severely limiting data bandwidth between them. To enable radically new electronic systems for capturing and integrating ubiquitous data, new system architectures are required.
In Chapter One, I present benefits of emerging nanomaterials (in particular, carbon nanotubes, CNTs) for computing. I show how leveraging new nanomaterials, their novel properties and new fabrication capabilities enables radically new system architectures and applications. For field- effect transistors fabricated with carbon nanotubes (CNFETs), these benefits lead to the concept of monolithic 3D integration – the ability to integrate previously heterogeneous technologies within a single chip, achieving functionalities that exceed the sum of their parts.
Moving from theoretical promise to real-world implementation requires a CNFET technology that is compatible with commercial silicon-based semiconductor manufacturing without sacrificing performance benefits versus conventional silicon-based technologies. In Chapter Two, I outline the work that overcame this enduring challenge, achieving the first CNFET fabrication process in a commercial silicon-based semiconductor manufacturing facility and foundry.
While the first two chapters are dedicated to continued progress in the field of computing for medicine, the third chapter expands the lens of traditional computing to the field of medicine. In Chapter Three, I show how the classical engineering concept of up-sampling (increasing temporal and spatial sampling frequency) can be applied to the problem of clinical blood analysis, experimentally demonstrating the application of this new methodology, called Distributed Single Point Blood Analysis (D-SPAYSS), to the localization of blood clotting pathologies in vivo.
Overall, this thesis presents a vision for a new generation of electronic systems for computing in medicine, presenting conceptual and practical advances and laying the foundations for transforming these visions from concept into reality. | |
