LEE- Elektronik Mühendisliği-Doktora
Bu koleksiyon için kalıcı URI
Gözat
Son Başvurular
1 - 5 / 19
-
ÖgeMEMS sensor platform for vital monitoring under mri and intraocular pressure measurement(Graduate School, 2023-07-07)In part of this study, we aim to develop optical-based Microelectromechanical System (MEMS) sensors for minimally invasive and non-invasive medical devices used for Magnetic Resonance Imaging (MRI) interventions.The use of MEMS in medical and biological applications has been rising steadily because it allows multifunctional devices to be integrated on the same substrate with the help of miniaturization. These sensors will be used to track the location of medical devices in real-time as well as measure pressure and ambient temperature. The operator can see environmental factors like temperature and pressure as well as the locations of surgical tools like catheters with the aid of this cutting-edge technology. This improves the success of the procedure by enabling tracking and information gathering from the interventional device without degrading the quality of the collected images. This work aims to integrate MEMS and fiber optical components to be compatible with the imaging modality. It also develops a microfabrication sequence for MEMS sensor implementation. In the first approach, microsystems sensors are integrated on a fiber optics-based platform for use in therapies supported by magnetic resonance imaging. MEMS sensors are implemented on a platform, and single-mode fiber cables are integrated with this platform. During imaging-assisted surgical operations, the described platform provides real-time and in-situ pressure, temperature, and location feedback. A medical interventional device with an inner diameter of 1.8 mm may accommodate the platform. The platform has a three-dimensional printed polymer cap with perforation utilized for the circulation of blood in the vessel to allow correct monitoring of the temperature and pressure in real-time. At the fiber cable ends, Graded Index (GRIN) lenses were used to increase the effectiveness of optical signal collecting. Three laser beams illuminate the MEMS platform that contains temperature (T), pressure (P), and localisation (∆X) sensors. Each sensor used a separate light source with a different wavelength: a 637 nm laser for pressure, a 780 nm laser for localization, and an 875 nm LED (with 50 nm bandwidth) for temperature. A released metal-polymer-metal hybrid membrane changes the environment's pressure relative to the membrane's chamber pressure using an interferometer readout method based on diffraction gratings. To research and develop the best membrane for the intended blood pressure range (from 5 mmHg to 240 mmHg), we designed and made the membrane of the optical pressure sensor (over the platform) in multiple sizes between 200 and 400 um. Based on fluctuations in the energy bandgap with ambient temperature, temperature sensing is performed in semiconductors (such as GaAs) by changing the absorption and transmission. The incident light on the semiconductor (such as GaAs) at a certain wavelength are reflected with the temperature change signature. As an optical thermometer integrated on the platform and lighted by light-coupled fiber optic, we used a Gallium Arsenide die, where one surface of GaAs is coated with metal to operate as a mirror. The magneto-optical Kerr effect (MOKE), which describes the change in polarization on the reflected light beam caused by variable magnetization in a magnetic substance like Iron(III) Oxide, is used to determine the location of the medical device. Prisms are incorporated under the platform in a retro-reflector shape and covered with magnetic material to reflect lit polarized light in the direction of the fiber optic. The sensor chip's measurements of temperature precision (0.22 ◦C), pressure resolution (1 mmHg), and localization resolution (3 mm), all of which are pertinent to medical practice, were made. In a second study; the integrated MEMS pressure, temperature, and magneto-optical sensors are developed enabling the operator to get real-time data from all MRI-compatible fiber-based devices on a single platform. As a result, the operator will be able to undertake interventions with a solid collection of real-time information about the patient's condition during the procedure. By incorporating the sensors developed in this work, medical equipment like catheters and stents can open up new possibilities for interventional surgery. We developed our first proposed multi-sensor platform as a second approach, using one fiber optic and one light source. In order to do this, we describe a stacked temperature, pressure, and localization platform designed for magnetic resonance imaging-based minimally invasive surgical and diagnostic procedures. The platform includes a magnetized material on a double prism retro-reflector that uses the MOKE as a magnetic field sensor to provide localization feedback during magnetic resonance imaging, a Gallium Arsenide band-gap temperature sensor, and a titanium, parylene, and titanium three-layer membrane pressure sensor. In order to determine where the sensor and the interventional device, such as a catheter, ablation probe, etc. to which our platform is attached are located, we used the MOKE technique to assess the spatially changing magnetic field density. A single fiber optic connection may connect all sensors, and the gathered light is sent to a spectrometer and a polarimeter. To employ interferometry to measure the pressure, a microfabricated three-layer sealed membrane with embedded diffraction gratings is used. An analytical formulation that connects the pressure to optical intensity is developed for the three-layer microfabricated membrane sensor. The analytical conclusions are also supported by finite-element simulation results. The use of wavelength division multiplexing allows for simultaneous sensor addressing. A magnetic field sensor, a pressure sensor, and a temperature sensor each had proof-of-concept operations that revealed sensitivities of 25 mdeg/mG rotation of polarization, 1.5 nm/mmHg displacement in agreement with simulation results and analytical findings, and 0.36 nm/◦C bandgap wavelength shift, respectively. The suggested gadget can be modified for usage in clinical settings for magnetic resonance-assisted surgical operations with future development. Overall, new application areas, including those for RF ablation catheters, highly focused ultrasonic catheters, and laser ablation catheters, will be made possible by the successful demonstration of sensor functioning in the MRI modality. The results of this study could inspire the development of novel interventional medical systems and technologies. In a third study, we presented a novel implanted MEMS sensor-readout glasses pair for the real-time monitoring of intraocular pressure based on the design and manufacturing of several optical pressure sensors in varied membrane widths that are acceptable for human physiological pressure. The entire system consists of two components: (i) a diffraction grating interferometric MEMS sensor that can be implemented into the cornea or intraocular lens, and (ii) readout glasses embedded with a laser diode, miniaturized aspheric lenses, and a CMOS camera. The suggested intraocular pressure measuring device allows for an eye tilt tolerance of around ±8 degrees while being monitored by a camera because to the number of diffracted orders. Additionally, the sensor is protected from the effects of changing optical power (caused by eye movement or laser noise) by the use of one or more reference gratings nearby. The ray-tracing simulations of the readout glasses, the analytical modeling of the diffracted orders from the pressure sensor, and the FEM results showcasing the deflection versus pressure behavior of the MEMS sensor are all included in the detailed design of the proposed device. We demonstrated pressure measurement in the range of ∆ p = 40 mmHg using in-vitro tests, with an average deflection sensitivity of 4.06 nm/mmHg and a resolution of 2.5 mmHg. Overall, in this part of the study, we present the design, manufacturing, and characterization of the optical sensor-glasses pair for real-time monitoring of intraocular pressure. To track the diffracted order intensity as a function of IOP, the readout glasses are equipped with a light source, a telescope aspheric lens pair unit, a scattering plate, and an endoscopic camera. A reference grating is placed next to the sealed membrane-based pressure sensor in order to distinguish the recorded pressure from variations in optical laser power. With a Polydimethylsiloxane(PDMS)-based inflated eye phantom, the combined sensor-glasses platform was put to the test. We took measurements of the diffracted order intensities for eyeball rotations up to ±8 degrees with a potential increment of ±15 degrees. The glasses were 3D printed in a modular form using the Selective Laser Sintering (SLS) technique. According to ANSI laser safety guidelines, a 1 mW laser was used in the application. The last section of the thesis study includes the latest results on a non-contact, non-invasive device for measuring intraocular pressure. The pressure-monitoring device is a pair of wearing 3D printed glasses that includes a laser source, several miniature lenses and mirrors, a mask for structured corneal illumination, and a miniature camera. By measuring the radius of curvature of the grid-like pattern on the cornea, the pressure level may be determined. We use tests on an elastic eye phantom at various tilt angles, analytical modeling, ray tracing, finite element simulations, and experiments to support our concept. In addition, we have shown the innovative non-contact smart glass in proof-of-concept functioning. The findings show that a pressure measurement resolution of 2.4 mmHg between the 0-55 mmHg pressure range may be achieved. This is based on the change in the radius of curvature of the projected grid pattern on the eyeball with altering internal pressure. The laser diode (with a wavelength of 650 nm), the lenses, and the miniature camera all had their own cylindrical grooves. A PDMS-based eye phantom was utilized in the studies due to the elasticity property of a PDMS with a Young's modulus of roughly 1.5 MPa. The suggested device may be employed for individualized real-time intraocular pressure monitoring throughout the daylight with additional in-vivo testing and development.
-
ÖgeHigh speed data acquisition techniques for pipelined analog to digital converters in IHP SiGe BiCMOS 0.13 µm(Graduate School, 2024-04-29)The resolution of the analog-to-digital converter market may be categorized into 8-b, 10-b, 12-b, 14-b, 16-b, and other options. The incorporation of several resolutions arises from the demands of different applications. In 2018, the 12-b resolution lead the market for analog-to-digital converters. The 16-b type is expected to overtake and become the dominant force in the future. The use of 12-b for 5G connectivity presents a favorable opportunity for market expansion. Texas Instruments unveiled a groundbreaking ADC in May 2019, boasting the industry's largest bandwidth, lowest power consumption, and fastest sampling rate. This converter is anticipated to assist engineers in attaining optimal measurement precision for 5G testing, oscilloscopes, and direct X-band sampling in radar applications. In this work, a one way 11-b pipeline ADC, designed in a SiGe BiCMOS 0.13 μm, is presented. It has sampling frequencies up to 1.6 GS/s and can provide above 8-b ENOB for the low input signal frequency and 6.4-b ENOB for the highest input frequency of 799 MHz according to the simulation results obtained without calibration. For our ADC, sample-and-hold amplifier-less (SHA-less) architecture was preferred since the SHA was one of the most power-consuming sub-blocks, and brought inevitable noise and distortion. A composite ADC architecture, having 8x 1.5-b cascaded stages and a back-end 3-b flash ADC is designed to reach up to 11-b physical resolution. A novel MDAC is proposed to mitigate ISI. Moreover, a novel BiCMOS residue amplifier (RA), which performs 6.43 GHz UGB and 80 dB DC gain, is implemented. A non-overlapping clock generation architecture at 1.6 GHz is devised, incorporating a differential clock driver and clock level converters.The SNRjitter of the clock generation system is 57 dBc at a clock signal frequency of 1.6 GHz. The results of the measurements carried out at ITU VLSI are included in the thesis. SiGe BiCMOS 0.13 μm process is utilized to fabricate the complete ADC, which has a 1.6 V supply and a silicon area of 2.2 mm × 3 mm.
-
ÖgeMulti - capsule endoscopy: Demonstrations of inter - capsular control and (tactile) sensing(Graduate School, 2023-12-19)Wireless Capsule Endoscopy (WCE) is an emerging Gastrointestinal (GI) tract imaging and treatment method that is developed to be a better alternative to the traditional endoscopy/colonoscopy devices and procedures. Thanks to pill shaped design and dimensions (22-26 mm in length, 10-13 mm in diameter), patient can easily swallow a single WCE where the capsule travels through the GI tract, and leaves the body. The complete non-invasive monitoring of the small bowel is possible in this way which is not possible in conventional endoscopy/colonoscopy. Even tough the WCE has very important role in the GI diagnostics in the clinics, there are several limitations to its usage. First of all, it is still cannot take any biopsy sample or make therapeutic actions in the clinical applications. Nonetheless, clinical operator can not control locomotion of the capsule through GI tract, which affect the monitoring process negatively. Also, WCE cannot perform sunctioning, flushing etc. which can be possible with conventional endoscopy. Due to its size limitation because of the patient comfort, battery life is limited and this might result with an incomplete examination. There are exhaustive studies to find solution to these shortcomings in the literature which we will be also working with the similar aim. Commercial capsules have a CMOS camera integrated on both tips and takes video stream and/or photographs while travelling. A clinician interprets this visual data to diagnose any issues with GI tract. Besides imaging, numerous literary studies on capsule endoscopy have demonstrated drug delivery, navigation strategies, tactile sensing for tumor diagnosis, and biopsy to increase the functionality of the WCE. While each function can work individually, using these in conjunction is needed to achieve complex treatment methods without any invasive process. Yet, the size limitation due to patient comfort hampers the availability of multiple features within a single capsule. In the first two parts of our study, in an effort to increase the space and functionality, we propose the usage of multiple capsules in conjunction. For our method, capsules together form a capsule-train in GI tract, whose wagons are connected with magnetic push/pull forces without any contact to each other. We focused on contactless/wireless connection between capsuels due to possible tissue damage. By knowing the distance between capsules, several functions can work in junction, e.g. the first capsule uses camera to find a tumor and second capsule takes a biopsy sample accurately thanks to known distance between capsules. For the first section, we have used passive magnets to achieve a wireless force connection between two capsules. After trying several magnet arrangamets on the tips of the capsules, we finalized a design where two capsules has a balanced constant distance in-between by using both pulling and pushing forces. We have used two large ring magnets that pulls each other and two cylindrical magnets that pushes each other. Cylindrical magnets are placed on the tips of the capsules while the ring magnets are placed with more distance to each other. Here, pulling force is dominant at large distances due to ring magnets and pushing force is dominant at smaller distances due to cylindrical magnets. This arrangament achieves constant distance in-between capsules without any energy consumption. Distance value is determined by the magnet sizes and arrangement. However, this method only works in thight tube-like shapes where tips of the capsules can not misalign, which might result with a clinch between a tip capsule with a ring capsule. To test the passive connection, we have used straight plastic tubes where we move capsules together with a constant in-between distance. Here, we pulled one of the capsules with a stepper motor and monitored in between distance with a camera placed above the test setup where an image processing code is ran to monitor the distance. We have achieved the capsule train without any connection breaks for typical bowel movement speed. As the second section, we improved our capsule train model to be more applicable on more challenging real life environments. Since typical human bowel diameter is ~25 mm while WCE diameter is ~12 mm, two capsules will have an angle between their tips, which results with a difference in between force due to angle. Here, we demonstrate an active distance control model with a closed loop control via the placement of a sphere permanent magnet on one capsule and a solenoid on the other capsule. Hall Effect Sensors have employed to determine distance between capsules. A PID controller have been developed to achieve stabilized desired distance between capsules by manipulating solenoid current. Experiments were conducted by pulling the leading capsule at typical human peristalsis speed. An inter-capsule distance of 1.94 mm was achieved on the average for the desired distance as 2 mm on 3D-printed plastic phantoms, while 0.97 ± 0.28 mm of distance was observed for the ex-vivo bovine tissue, for a set distance of 1mm. By achieving successful demonstration of inter-capsule control, this work substantiates realizability of multi capsule endoscopy for future studies. As the third part of our study, we focused on to develop a novel diagnostic tactile sensing method to use in capsule endoscopy, which supports our multi-capsule approach by adding a new method to WCE functional palette. Since palpation is a widespread diagnostics method for clinicians, using this method inside GI tract will be an useful addition since some of the abnormalities such as inflammation of early stage tumors cannot be seen on visual imagery while having higher elasticity modulus than healthy tissue. Planned tactile sensing model measures the tissue elasticity modulus. In our fourth part this study, we will be presenting our tactile sensing mechanism that adopts the atomic force microscopy (AFM) methodology and fits it into a single WCE volume. AFM is used to achieve surface imaging up to nanoscale levels of resolution by scanning a moving cantilever through targeted area. On each discrete scanning position on the sample, cantilever base moves down to make cantilever tip interact with the surface. Tip deflection occurring due to interaction between the cantilever tip and the surface gets recorded. Mechanical surface properties such as topology, elasticity, adhesivity etc. can be deducted from this data by Hertz contact model to be used to inspect the tissue healthiness. Main difference of our model is having a different cantilever tip displacement measurement method. Conventional AFM uses a lazer beam reflecting from the cantilever tip and lands on a 4 part photodiode sensor where the sensor output is directly related to cantilever tip deflection. Since using such system in a single capsule is not possible with current hardware, we focused on using a different sensing method and decided to use a piezoelectric material attached onto cantilever. As cantilever tip bends, piezomaterial placed onto cantilever is also bends and generates charge. Amount of generated electric charge, therefore the current output, is directly related to piezo deformation. By using this mechanism, we can use the same elasticity modulus measurement procedure with AFM approach, which is widely used in practical applications. We planned to use a micro stepper motor in the capsule due to size limitations. We have 3D modeled the cantilever and inner-capsule holding parts around the typical WCE and motor dimensions. Since each manufactured cantilever and piezo material is unique, we needed to calibrate each assembly on a glass surface where no indentation occurs. Also, we needed a charge amplifier circuit to read piezo signal output since the used piezo sheets has ~10 mm3 area with a very small electrical charge generation. The test procedure calculates the stiffness of the cantilever by finding it's resonant frequency. We need this value to find force acting on the tip of the capsule. By knowing the piezo output on glass and the stiffness of the cantilever, we were able to start our test on different materials and tissues. We have prepared two jellies with different densities to experiment on. Also, we used chicken breast, bovine liver, sheep stomach, cow stomach, and human colon tissue as tissue tests. After all measurements, we were able to compare real elasticity values of jellies, chicken breast and bovine liver tissues to measured values by the cantilever. After all test, we have obtained 30% maximum error on chicken breast, 15% on bovine liver, %33 on stiffer jelly and %49 on the second jelly. Since the tissues with inflammation or tumors have 4x to 5x larger elasticity values than healthy tissues, there results would be usable to determine the state of the inspected tissue which is our main motivation for this section. As an additional study, which can be identified as the fourth section of this study is about 3D locomotion of a single WCE (with passive magnets inside) in stomach by an externally controlled electromagnet array used by an clinician. Thanks to this mobility, clinician can see any desired area in the stomach. Here, WCE position and angle is controlled by 5 electromagnet array placed under the patient in lying position. Capsule tip levitation is controlled by a single, larger electromagnet placed above the patient. As my study, I have simulated the angular position of WCE by alternating the electromagnet currents to achieve any desired distance. Later on, I conducted rotation, levitation and displacement experiments on 3D printed PLA bovine stomach model. Also, same experiments done in ex-vivo environment where bovine stomach tissue is paved into 3D printed PLA model. These experiments succesfully valiated the applicability and accuracy of 3D control with 6 magnet array arrangement.
-
ÖgeNovel fractional order calculus-based audio processing methods and their applications on neural networks for classification and synthesis problems(Graduate School, 2023-10-24)This thesis dissertation aims to explore the application of the Fractional Order Calculus (FOC) framework in addressing contemporary problems in audio signal processing. One crucial aspect of present audio signal processing approaches is their reliance on large amounts of data, which necessitates appropriate tools for increasing amount of data. Another important aspect is in relation to the methods used to produce inference models. The neural network approaches dominating the field often require optimization of a large number of parameters. As a result, digital signal processing (DSP) tools are being repurposed to reduce the parameters of neural models. The introductory chapter provides an overview of the dissertation's purpose, which is to investigate whether FOC can provide novel methods to solve problems in neural network-based audio signal classification and reconstruction. Chapter 2 introduces the FOC framework by explaining its capabilities and complexities. While the complexities of FOC have often caused it to be overlooked in engineering applications, its capabilities have attracted the interest of many researchers in various fields, including audio processing, time series estimation, and image enhancement. Providing examples of FOC based applications on audio signal processing, this chapter aims to provide fammiliarity to the FOC concept. The dissertation is structured such that each chapter focuses on a specific application of audio signal processing. Chapter 3 tackles the problem of audio classification, which is categorised by being speech, music or environmental sound signals. Due to the limited availability of data for environmental sound signals, data augmentation methods remain crucial for Environmental Sound Classifaciton (ESC) problems. The chapter presents three FOC based data augmentation methods: Fractional Order Mask, Fractional Order Frequency Scale, and Fractional Order Mel Scale. Fractional Order Mask and Fractional Order Mel Scale methods are applied to Mel Spectrogram and Log-Mel Spectrogram representations of envrionmental sound data. Experiments on ESC problem with neural architectures demonstrate their effectiveness as data augmentation tools in improving the accuracy of neural network models. The findings indicate that employing a data augmentation procedure in combination with the proposed methods can yield a boost of approximately 7.7% in performance for a 5-layer CNN when Log-Mel Spectrograms are used as input. Similarly, the augmented dataset resulted in a increase of over 9% in performance for a 18-layer ResNet. Chapter 4 delves into audio synthesis and its importance in reconstructing time domain representations of audio signals. The history of vocoding methods and their relation to signal reconstruction approaches are discussed. The chapter focuses on phase reconstruction with methods such as SPSI and spectral consistency based iterative methods such as the Griffin-Lim Algorithm (GLA) and its novel forms. In this chapter a FOC based method is proposed. The FOC based method models a signal's Power Spectral Density Function (PSDF) using Fractional Differential Equations (FDE), estimating the instantaneous frequency of a peak in a windowed audio spectrum. This method proves effective in phase reconstruction. The results show the usage of FOC framework provided up to 4% better quality than SPSI. The experiments also highlight the proposed method's effectiveness as an initial phase estimator for spectral consistency based iterative methods. Chapter 5 explores the contemporary research topic of Neural Audio Synthesis (NAS).
-
ÖgeFully simulated and model based power consumption estimation of internet of things devices(Graduate School, 2024-09-11)The rapid proliferation of Internet of Things (IoT) devices, driven by the demand for efficient computing units integrated into various networks, underscores the critical need for energy efficiency. As these devices often operate in mobile settings, optimizing energy consumption becomes paramount to minimize maintenance costs associated with battery replacement or recharge. Additionally, power efficiency directly impacts the portability of IoT devices, enhancing their usability and effectiveness in diverse environments. Software and hardware factors are influential on the energy consumption of IoT devices. Due challenges of battery replacement, IoT devices rely heavily on software-based power management for optimization. Thus, software updates playing a significant role in battery life expectancy. To plan maintenance processes effectively, manufacturers and service providers must accurately estimate energy consumption and battery lifetime, necessitating a holistic approach to power estimation. Traditional methods of energy consumption estimation, reliant on physical measurements, are impractical due to the extensive hardware and software design iterations required. Consequently, a fully simulated, model-based approach to power consumption estimation emerges as essential, especially considering the frequent update requirements of IoT devices. Such an approach enables accurate estimation throughout design changes and updates, facilitating efficient planning and management of power consumption across various scenarios. This thesis proposes a comprehensive energy consumption model tailored for IoT devices, complemented by fully simulated, model-based system-level power estimation approaches. By leveraging simulation environments like Open Virtual Platform (OVP), the proposed methodology achieves approximately \%97 accuracy in typical real-life scenarios. Notably, the methodology eliminates the need for completed hardware and software designs, enabling efficient power estimation throughout the development and operational phases of IoT devices. In conclusion, the study contributes a novel methodology for accurate power consumption estimation in IoT devices, addressing the challenges posed by evolving hardware and software requirements. By embracing simulation-based modeling and system-level approaches, the proposed methodology offers a practical and efficient solution for managing power consumption in IoT devices, ultimately enhancing their usability, reliability, and sustainability in diverse application domains. All these advantages have been validated through different applications, and the results have been shared within the scope of the thesis study.