Fabrication and Optimization of Micro-ECoG Electrodes and Microfluidic Actuated Neural Electrodes

Mentor: Flavia Vitale
Advisor: Brian Litt

A third of all epilepsy patients who experience continuous seizures are treated with surgery to remove epileptogenic brain tissue. Currently, surgeons determine disectable tissue using electrocorticography (ECoG), a technique that detects action potentials, electrical potential difference within a cell, from neurons in the brain. Current ECoG electrodes are not flexible enough to conform to the curved surface of the brain, while depth electrodes cause permanent damage to the brain. Furthermore, the resolution of these sensors is not accurate enough to precisely locate the epileptogenic areas increasing the potential to leave behind epileptic tissue or to extend the resection to healthy tissue. In this project, we want to develop and optimize a microarray of sensors made of materials that are flexible enough, and sensitive enough to increase temporal and spatial resolution. Additionally, a depth electrode delivery device will be developed to mitigate the damage done to neuronal tissue. The sensors are optimized to be used in rat models, which will determine the feasibility and safety of the device for clinical applications.

Project report

Design and Performance of a Multi-Port Robotic Suction Gripper

Mentor: Jun Seo
Advisor: Mark Yim

This paper presents the mechanical-workings and gripping performance characteristics of a vacuum-based robotic gripper. The gripper uses multiple ports with attached commercial suction cups to grasp objects from a variety of directions using suction power. The ports’ valves remain closed until opened by a hinge-based mechanism, in order to limit the effects of vacuum leakage. The ability for the valves to be actuated in a mechanically passive manner implies that this gripper could be used on a robotic limb without adding heavy, complex devices.

Project report

Wearable Haptic Feedback Actuators for Training in Robotic Surgery

Mentor: Jeremy Brown
Advisor: Katherne Kuchenbecker

The Intuitive Surgical da Vinci robot allows surgeons to perform minimally invasive surgery easily, but it does not provide any touch feedback. This lack of haptic cues is especially challenging to those who are first learning to use the da Vinci, because they must rely solely on their vision. Inexperienced users often unwittingly apply forces that injure the patient’s tissue or break sutures. The solution analyzed in this project focuses on a haptic feedback system for training. It places a three-axis force sensor beneath the task materials and a custom servo-driven haptic actuator on each of the user’s wrists. Electrical conductivity is used to determine whether each tool is touching the task. The left and right actuators squeeze the user’s wrists in proportion to the magnitude of the force applied by the respective instrument. Initial tests were conducted to determine whether this system is beneficial. One expert robotic surgeon and several lay people used the developed system to complete a task called ring rollercoaster, in which one guides a ring along a curved bar. All participants immediately understood how to interpret the provided haptic cues and remarked on their utility, and the expert surgeon was particularly enthusiastic about the system’s potential as a training tool. Ongoing work centers on designing a human-subject experiment to quantify the impact of this system on the learning curve of surgeons in training.

Project report

Different Parameters Affect the Piezoelectric Properties of Polyvinylidene Nanofibers

Advisor: Jorge Santiago

Many cases of hearing loss arise from damage that occurs to hair cells located inside of the cochlea. The use of the biocompatible and piezoelectric material polyvinylidene fluoride (PVDF) is being studied as a possible substitute for the damaged hair cells. More specifically, PVDF in the beta phase is of main concern because this is the phase in which it shows signs of being piezoelectric. A maximum amount of piezoelectric response is desired, therefore the process that makes PVDF piezoelectric must be optimized. In order to possibly confirm the passing of the PVDF nanofibers through the inner ear membrane, a light-sensitive dye (Rhodamine B) is included in the preparation of the solution. The amount added must be taken into consideration on the premise that it could affect the level of piezoelectricity developed. The aforementioned process that PVDF must go through is called poling, where the created sample is heated and exposed to an electric field. In doing so, the individual dipole moments of the molecule will be aligned to face similar directions, which makes the piezoelectric characteristics emerge. The point of heightened induced piezoelectricity in our poling setup is found by varying the voltages at which the sample is poled at. In doing so, the optimal poling setup for inducing piezoelectricity in PVDF is found. In this study, Piezoresponse force microscopy was used to measure levels of piezoelectricity in the samples and atomic force microscopy was used to analyze the physical properties of the samples. These results have facilitated the action of attaining higher piezoelectric levels in PVDF.

Project report

Comparing Apples and Oranges: Hyperspectral Imaging Techniques for Fruit Detection and Determination of Plant Health

Advisor: CJ Taylor

Remote sensing has grown to be an attractive strategy for monitoring agriculture in orchards and farms because it is a quick, simple, and non-destructive method of data acquisition. Our aim is to use hyperspectral imaging techniques to estimate crop yield and determine plant health based on data taken from the vegetation canopy. The canopy consists of what is visible on the outer layer of vegetation on a tree. To determine the viability of this method, an Ocean Optics USB 2000+ Vis-NIR spectrometer was used to collect preliminary data on leaves and some fruits that may be encountered. Reflectance measurements were taken for wavelengths ranging from 400-1000nm. To analyze the data, strategies involve using vegetation indices such as NDVI, PRI, and MCARI and the application of machine learning techniques to determine how to efficiently and accurately represent remote sensed data. Using a quadratic Support Vector Machine (SVM) on a subset of reflectance data from fruit a detection accuracy of over 80% and it was determined that PRI and NPCI vegetation indices can be used for fruit detection.

Project report

On Applications of Hydrogel-coated Nanorod Films as Soil Humidity Sensors

Advisor: Cherie Kagan

Hydrogels are a well-studied class of materials composed of polymer chains, known primarily for absorbing water. Our current multi-step curing process allows us to attach a substrate containing an array of hundreds of nanorods to the surface of a thin hydrogel film. By harnessing the optical properties of gold plasmonic nanorods and the hydrophilic properties of hydrogels, we demonstrate a new hydrogel-based optical humidity sensor that can measure soil wetness. Our preliminary results suggest that this sensor can be modified and produced on a larger scale to eventually serve as a useful tool for agriculturalists interested in measuring their soil’s wetness.

Project report

Enhancing Device Sensitivity of Graphene Field Effect Transistor DNA Biosensors via Single Layer Boron Nitride

Advisor: Charlie Johnson

When it comes to nanoscale devices, graphene continues to demonstrate its great potential as an ideal material for various nanostructures. Graphene is seen as an ideal material for nanoscale devices due to its unique physical and electrical properties, which include high electrical conductivity, sensitivity, and flexibility. As such, graphene proves to be useful for developing highly-sensitive sensors and electronics of various kinds. Here we present our progress on developing highly-sensitive graphene field-effect transistor (GFET) sensors capable of detecting single-stranded DNA sequences and provide electrical readouts from these sequences based on their composition. Here we chose to examine the effects of adding monolayer BN to our GFET. To do this, we tested two variations of our GFET devices: a variation consisting of monolayer BN deposited on top of our graphene layer and a variation consisting of monolayer BN deposited underneath our graphene layer. Our devices made with an external monolayer of BN deposited on top of the graphene layer showed higher sensitivity and a lower Dirac point voltage. Furthermore, our fabrication method is scalable and reproducible. These highly-sensitive and accurate bio-sensors show great potential for medical sensing applications such as disease detection and DNA hybridization.

Project report

Grasped Object Orientation Detection for Robust Peg-In-Hole Tasks

Mentor: Alex Zhu
Advisor: Kostas Daniilidis

This paper approaches the problem of detecting the orientation of an object while the object is grasped by a robot. Manipulator robots are learning new and exciting ways to interact with their environment every day. Object recognition and grasping algorithms allow a robot to identify, target, and manipulate objects of importance. However, a common problem in grasping is maintaining knowledge of the orientation, of the object while it is being grasped or manipulated. We have developed an algorithm that estimates the orientation of painted plastic test tubes, without assistance from external markers. We use line fitting to localize the center axis of the tube which assists in the tube’s orientation detection. We then compare the orientation of the tube to the gripper’s approach to the test tube array. Based on the angle difference between the +Z-direction of the gripper and the center axis of the tube, we can infer the movements necessary for the robot to reposition its arm to properly approach the test tube array and replace the test tube in the specified index.

Project report

Mental State Transitions and the Role of Polysynaptic Pathways

Advisor: Danielle Bassett

Brain state transitions are crucial for high order mental processes, but the relationship between the white matter architecture of the brain and its implementation of these transitions remains under study. We address this question using network control theory by first defining a brain state as a pattern of activity across brain regions, and then delineating a model of brain dynamics. From this we can calculate the optimal input signals necessary to shift the brain into states of activity in different cognitive systems, and subsequently assess the contributions made by different brain regions. We can show that these contributions are correlated with regional weighted degrees of control efficiency and network communicability, a measure of connectedness both by direct and longer, indirect pathways between brain regions. Finally, we identify an optimal weighted calculation of communicability in order to improve this correlation, lending insight on the role of longer polysynaptic pathways in facilitating brain state transitions. This insight can subsequently lead to further knowledge of the biological basis by which these longer pathways were formed and maintained, despite an apparent evolutionary disadvantage. We found that scaling the communicability values by weaker functions than the original exponential, such as linear functions, led to an improvement in the correlation between control efficiency and communicability, indicating influences of longer transition pathways which have not been fully accounted for in previous definitions of communicability.

Project report

Test Automation Platform for Computer-aided Clinical Trials

Mentor: Marco Beccani
Advisor: Rahul Mangharam

The complexity of software in medical devices, such as pacemakers and implantable cardioverter defibrillators (ICDs), has created the challenging problem of making sure that the devices deliver therapy in only appropriate situations. Because of patient condition variability, safety recalls have affected over 600,000 devices between 1990 and 2000 [1]. Of these recalled devices, 200,000 were recalled due to firmware issues [1]. The current standard for closed-loop testing of these devices is a clinical trial. Closed-loop means that output of the system (ICD outputs) is fed back to the input of the system (the heart) in order to change the input. These trials test the new device against the current standard of care by comparing their performances on real patients. These trials can last for four to six years, cost millions of dollars, and put the patients who are receiving the new device at risk. Because of these risks, it is necessary to have in silico pre-clinical trials that use virtual heart models, rather than real patient hearts, to test ICD devices. To perform these pre-clinical trials, a test automation platform is needed. In this paper, we propose a test automation platform that is able to test real ICD devices over a wide range of heart conditions. The platform is able to take voltage samples that represent electrocardiograms (EGMs) from the user’s PC and recreate these EGM samples into electrical signals. These electrical signals are then processed so they can be fed to an ICD device. This paper will go into more detail about the design for the hardware and software for the platform, along with discussing results for the platform.

Project report

Functional MRI-Compatible Orthosis for Hand Grasp Assistance During Robot-Assisted Therapy for Stroke Survivors

Mentors: Kevin Bui and Roshan Rai
Advisor: Michelle Johnson

Overall stroke hospitalizations have been decreasing in the US, but the number of stroke incidents within the HIV population have been on the rise. This trend is being investigated in a study involving functional MRI (fMRI) scans, except many potential subjects are being excluded because of spasticity, a condition that prevents them from opening their hands. This exclusion would lead to less representative results, which calls for the need of an orthosis that will assist in hand opening. However, the orthosis must be MR-safe, among other requirements. After experimenting with the Aquahand (a device designed to help stroke survivors with spasticity) and studying hand anatomy, a new design modeled after the hand has been conceptualized and tested in its early stages. Initial results have been promising, with a few more modifications to be made. The design will later have to be tested for effectiveness and compatibility with an fMRI.

Project report