Project Goal: Develop a benchtop demonstration system to show the feasibility of measuring volume dispensed for tissue staining systems using a capacitive sensor.

A key factor in tissue staining quality is the volume of reagent dispensed. A sensor platform that measures the volume dispensed allows the staining environment to be monitored, understood and controlled in real time throughout the staining process. This project presents a capacitive volume sensor to automate on-slide fluid volume measurements in real time.

This design uses a carefully defined sensing pattern on a printed circuit board to detect fluid on the surface of an adhered glass slide. It does so by measuring the change in capacitance of the system. A custom printed circuit and a capacitance-to-digital converter communicate with a custom graphical user interface on a host computer to report the volume sensed. Using a curve fit of experimental data, the capacitance is converted to volume and displayed as a graph on the graphical user interface.

Project Goal: Design, build and verify a system that uses light sensor data to store and calculate the volume of cancer-identifying reagents being dispensed onto a tissue sample, enabling technicians to confirm reagent volumes and avoid false diagnoses.

Roche Tissue Diagnostics’ automated tissue staining systems dispense set amounts of dye, or reagent, onto patient tissue samples to diagnose conditions such as cancer. To deliver the most precise and accurate diagnoses, the system needs to dispense the proper quantity of reagent on every sample. The team developed a new system that uses a neural network to calculate an accurate volume for each drop.

The system consists of a photoelectric sensor which is set in the path of the reagent dropper and sends a stream of analog input to a microcontroller. When a drop is detected, the microcontroller isolates the drop instance and performs preprocessing on it to prepare the analog input for the algorithm. The microcontroller then uses a pre-trained recurrent neural network to calculate an accurate volume for the drop. The volume calculation is saved with other bookkeeping data, then converted to analog and output to the Roche tissue staining system to verify the drop volume.

This system, built on advanced neural networks and made to integrate with the existing Roche Tissue Diagnostics tissue staining system, helps to verify the integrity of the tissue staining system at each drop, preventing patients from receiving false diagnoses.

Project Goal: Create a contactless liquid level sensor that can measure levels of tissue staining reagent with high accuracy, resolution and precision

The HE 600 machine automates tissue staining to diagnose diseases such as cancer. It dispenses reagents on top of tissue samples, which are fixed to microscope slides. The reagents are housed in 1-liter reservoirs, which use mechanical signal floats to determine liquid level. The floats are in direct contact with the reagents and so break down easily, putting excess strain on system pumps that need constant replacements. In addition, the floats also often get stuck or corroded, and thus do not accurately measure reagent levels.

This design uses contactless capacitive sensing to constantly measure the liquid level inside the reservoir. The system has a flexible printed circuit board (the sensor) and a small metal box containing the capacitive sensor chip and microcontroller. The entire system is adhered to the outside of the reservoir so that none of the components are in contact with the reagents.

The system provides real-time liquid volume levels via USB connection to a computer station with a graphical user interface. This design is meant to last the lifetime of the HE 600,creating a cost-effective and long-lasting volume measuring system.

Project Goal: Make the multifunctional displays, or MFDs, in a pilot's cockpit easier to read using course predictive software.

Commercial aircrafts are equipped with multifunctional displays, or MFDs, that use sensory equipment to collect the position of surrounding aircrafts and display them on a screen. The MFDs that pilots currently use are often cluttered with irrelevant aircraft information, making it difficult for the pilot to extract necessary traffic data. The team was tasked with developing machine learning algorithms to remove irrelevant traffic data, thereby improving pilots’ situational awareness.

The team developed three Predictive Aircraft Navigation, or PAN, algorithms, each of which includes a unique method for predicting aircraft flight paths: Convolutional Neural Network Long Short-Term Memory, Naive, and Encoder Decoder models.

Each of these models combine current and past flight data to predict future flight paths. The team then incorporated a comparison algorithm to rate the relevance of surrounding aircrafts according to their relation to the pilot. The results show a decluttered MFD so a pilot can extract critical information and make vital decisions in dense airspace.

Project Goal: Create a remote-controlled vertical farm to optimize hydroponic leafy green production.

Controlled environment agriculture aims to reduce inputs and increase outputs of food production to feed a rapidly increasing population. The V-Hive Greenbox is a modular and scalable vertical farming system designed to fill the volume of a shipping container. This project focuses on adding automated technology to the smallest fully functional scale to help reduce labor requirements.

The system has a metal frame of lighting boards with horticultural LEDs and growing boards with hydroponic growing channels. The boards are alternated in the frame for optimal light distribution to the plants. Their orientation is remotely adjustable in two dimensions to accommodate the light and space requirements of different plant species and growing phases, as well as harvesting and maintenance. This is accomplished using a belt and pulley rail system above the main frame, allowing the user to extend and retract boards out of the frame and to change the lateral spacing between lights and plants.

A Raspberry Pi can be accessed on-site using the mounted touch screen panel or remotely using Virtual Network Computing. The user can remotely control board movement, the nutrient pump, the air pump and lights. Data from environmental sensors and a camera feed can also be displayed and monitored from off site.

Project Goal: Improve the precision refrigerant charging station used for a portable medical refrigerator through the design and manufacture of an automated recharging station.

Refrigerant maintains the desired temperature of vaccines, food and living spaces. The less refrigerant a system requires, the more precision in refrigerant charging it requires. Providing an accurate quantity of refrigerant to any cooling system is critical. Most known methods of refrigerant delivery are done manually, introducing the potential for human error in the refrigerant delivery process. Improved methods of supplying refrigerant will ensure proper operation and efficiency.

The team designed a refrigerant recharging process system using electronically controlled pumps, valves and sensors. The system minimizes human involvement and maintains an accuracy of 40g ±0.10g of refrigerant during delivery. The monitoring system consists of load-cell-based weighing stations that continuously collect data on the mass of the refrigerant tanks and medical refrigerator. The system alerts operators of events, such as the need to switch out empty refrigerant tanks or full recovery tanks of excess refrigerant that are ready to be recycled.

This design improves refrigerant delivery accuracy by automating the refrigerant recharge process and monitoring refrigerant mass.

Project Goal: Create an interface for doctors to view and analyze metagenomic next-generation sequencing, or mNGS, data, allowing for rapid diagnosis and treatment of secondary infections for patients on ventilators.

During the COVID-19 pandemic, many patients are being hospitalized and intubated. Intubation increases the risk of developing secondary infections, which can be fatal in patients who are already very ill. Currently, it takes several days to diagnose secondary infections in intubated patients.

Metagenomic next-generation sequencing, or mNGS, can identify infections more quickly than current methods. However, it is not widely used in clinical settings, largely because clinicians are not yet comfortable with the technology. This project strives to push mNGS technology from "bench to bedside" by making the technology accessible and convenient within current infrastructure.

To make mNGS more accessible, the team created an interface for doctors to view and evaluate data. Lab personnel can upload data to the RShiny-based web interface, and doctors can choose how they’d like to view the data from among several formats. One option is to view the data in a format similar to existing reports, which the doctors are already familiar with. In addition, the app contains several unique analytical tools, including some to monitor the relative size of multiple infections over time, which is not currently possible.

Project Goal: Prevent the formation of pressure ulcers in patients via an autonomous sense and response system.

Pressure ulcers, also known as bed sores, are a recurring problem for immobile patients and often lead to infection, necrosis and other complications, which are resource-and cost-intensive to address. This project aims to prevent the formation of bed sores.

A dynamic air pressure mattress system has an embedded sensing device that sends pressure data to a central computer. This computer reads the pressure data, identifies potential areas of risk and determines the appropriate redistribution in pressure for the bladders in the mattress. The computer sends these commands to a microcontroller, which distributes the signals to specific solenoid valves that control and distribute the pressure as commanded.

The system identifies high-pressure locations on the patient's body and actively redistributes the pressure before these locations cause pressure ulcers. There is an LCD screen with user controls to manually adjust certain bed parameters, monitor the state of the system and reset the air mattress system.

Project Goal: Improve upon current market-standard indwelling urinary catheters by using electric stimulus and attenuate bacteria to reduce infection risk.

Attenuating bacteria and, in turn, reducing the risk of catheter-associated urinary tract infections, or CAUTI, are the main focuses of the Bacteria Attenuating Catheter. Current catheters have sought to address CAUTI with strategies ranging from silver nanoparticle coatings to antibiotic coatings. However, these have drawbacks. Most significantly, current catheters lead to an increase in the antibiotic resistance of bacteria, which can ultimately result in CAUTI.

A new system uses micro-amperage electrical stimulation to attenuate bacteria. Adding this to current market standard catheters reduces the risk of CAUTI by halting the aggregation of planktonic bacteria into biofilm. Analysis and testing of bacterial accumulation on the surface of various test catheters and the media (human pooled urine) yielded the most effective design.

The Bacteria Attenuating Catheter could decrease medical costs of treating CAUTI, which can cost up to $13,000 per incident, and reduce use of the 1.5 million single-use catheters used daily in the United States. It can also address the patient's needs while functioning as expected.

Project Goal: Develop an alternative to GPS that images and identifies star patterns during the day and night in order to determine orientation with respect to the celestial reference frame.

The prevalence of GPS use in the defense and civilian sectors has made it the target of disruption efforts. An alternative to GPS for determining global positioning in celestial navigation via astrophotography must be able to operate day or night.

The team used star imaging and star pattern recognition to create a custom-designed algorithm to locate global position. Daytime imaging can occur by combining the custom-designed optical system with image stacking to improve signal-to-noise ratio. They also developed a radiometric model to optimize design decisions and predict how the optical system would perform. A unique single, dual-axis gyroscope provided data to determine possible random changes in angle between successive images. These changes were used to calculate the necessary pixel adjustments during the image stacking process.

These additions made the system more robust, which allowed it to determine its orientation under shaky or unstable conditions during both day and night.