Briteseed is a medical company focused on creating safer minimally invasive surgical tools. They initially achieved this in their laparoscopic 8mm Smart Sealer tool, which uses single wavelength sensor technology to identify and verify the location of blood vessels within the tissue encapsulated by the jaws of the tool.

Now Briteseed is looking towards improving their blood vessel detection technology by moving towards using hyperspectral imaging (HSI) in new iterations of their tools, which requires redesigning and testing the new HSI tool.

Original HSI Testing Bed

As the HSI tool currently stands, there were two major issues I was tasked with solving:

• Ambient light affecting data quality: The HSI tool has one jaw containing LEDs which are transmitted through the tissue and received by the sensor housed within the opposite jaw. When tests are conducted with the current HSI tool, the team at Briteseed suspected that ambient light from external sources was affecting the quality of data collected due to the design and placement of the sensor within the tool jaw.
• Jaws moving independent to the handle: As a surgical tool, Briteseed's technology needs to create precise movements that correspond to movement initiated by the surgeon. Unfortunately, the current design of the tool allows for undesired vertical and lateral movement that needs to be removed.

HSI tool jaws

How can we mask ambient light to improve data collection quality?

I approached this problem by first outlining the main user requirements and constraints:

User Requirements

• Solution needs to block out most if not all the ambient light
• Solution needs to be cleanable (as the surface will often come into contact with tissue samples)

Constraints

• Jaw size can't be modified (fixed size constraint)
• Sensor within the jaw can't be obstructed by the solution (shape constraint)

With those requirements and constraints in mind, I brainstormed and CAD modeled 3 different solutions:

• Sticker mask: Use an opaque sticker on the inner side of the clear acrylic currently protecting the sensor to block out ambient light
• Plastic sheet mask: Use an additional layer of opaque plastic that sits within the jaw to block out ambient light
• Plastic sheet and sticker mask: Use a combination of the first two ideas to mask out ambient light

Sticker mask (bottom view)

Plastic sheet mask (bottom view)

Plastic sheet and sticker mask (bottom view)

After weighing out the pros and cons of each option, I decided that the plastic sheet and sticker mask was the best option for the following reasons:

• The plastic sheet layer blocks out ambient light coming in from other angles around the sensor
• The sticker provides additional coverage over non-active areas of the sensor that aren't covered by the plastic sheet (e.g. adhesive bumps on the ends of the sensor)
• The solution doesn't add any additional geometry to the surface of the jaw, making it easy to clean
• The solution makes use of existing parts, materials, and processes to manufacture their tool jaws

Plastic sheet and sticker mask modeled in the sensor jaw

Plastic sheet and sticker mask in the sensor jaw (cross-section view)

I prototyped the solution and after validating the size and fit of the components, implemented it in the HSI tools currently being used for testing at Briteseed.

Plastic sheet and sticker mask components

Step 1 of mask assembly (sensor in the jaw)

Step 2 of mask assembly (opaque acrylic mask)

Step 3 of mask assembly (sticker and clear acrylic)

Initial testing has shown that the mask...

• Successfully removes the presence of ambient light and its interference during data collection
• Also removes some undesired artifacts caused by the sensors as the interior acrylic mask ensures firm adhesion at the contact point between the sensor and the flexible PCB strip

Plastic sheet and sticker mask components

How can we improve the mapping of jaw angle to handle movement and remove unwanted lateral jaw movement?

I approached this issue by:

• Interacting with and observing the way the HSI tool moves
• Using CAD model simulations of the tool to understand the actuation system

HSI Tool CAD Simulation: I used constraints to limit degrees of freedom and mimic the physical tool's movement

Which led me to realize that there were two main problems with the independent jaw movement:

Problems

• Tool jaws rotate vertically (independent to handle movement)
• Tool jaws move laterally (when pressure is applied)

Solutions

• Limit vertical movement in the end effector
• Fill lateral gaps within the end effector

Constraints

• Can't modify existing metal components
• Solution needs to be temporary and modifiable for researchers to remove, if needed

From there, I prototyped and tested the following solutions:

• Part modified: Pivot
• Goal: Reduce vertical movement
• Change made: Removed slot

Original pivot (with slot)

CAD model of new pivot design

New pivot implemented

• Part modified: Long tube
• Goal: Reduce vertical movement
• Change made: Added a vertical stopper
• Next steps: Creating a metal version of the stopper

CAD vertical stopper in the HSI tool

CAD model of vertical stopper

Vertical stopper implemented

• Part modified: Pushrod
• Goal: Reduce vertical movement
• Change made: Added stickers
• Next steps: Modify and  3D print a new metal version of the pushrod

Original pushrod

Pushrod stickers

Pushrod stickers implemented

• Part modified: Pushrod pin
• Goal: Reduce lateral movement
• Change made: Replaced with a PEEK pin (cut to length)
• Next steps: Standardize dimension units (diameter was in imperial units and length in metric units), decrease yoke inner wall distance

Original metal pushrod pin

New pushrod pin being fitted for size

• Part modified: Jaw flags
• Goal: Reduce lateral movement
• Change made: Added stickers
• Next steps: Modify and 3D print a new metal version of the jaws

Jaws (without stickers)

Example of jaw stickers

Jaw stickers implemented

To test their HSI technology, the team at Briteseed had created an automated test bed to conduct consistent ex-vivo tests with porcine tissue samples. My tasks involved updating and improving the test bed in the following ways:

• Allowing the tool to come into contact with samples: The first version of the test bed required samples to be enclosed in glass slides before they could be loaded into the system. The team wanted the sample holder to be improved so that the tool could come in contact with the samples directly and better mimic the tool's use conditions in operating rooms.
• Allowing for data collection with both the Energetiq and HSI tools: The first version of the test bed only had one tool holder that would accommodate the single wavelength Energetiq tool. With the addition of the 15 wavelength HSI tool, the team wanted tool holders for both tools to conduct tests simultaneously.
• Allowing for dynamic data collection: The first version of the test bed was static and the sample couldn't be moved when the tool was on. The team wanted to be able to move samples between tools during data collection for a more efficient and streamlined process.

Original test bed (back)

Original test bed (front)

How can we improve the sample holder for dynamic testing?

• Convert the sample holder so that the tissue samples don't need to be enclosed in glass
• Create an adapter for the z-axis motor so that the sample can move in the z-direction

I first established a list of requirements for designing the new sample holder, which are as follows:

• Holds the blood vessels (BVs) and connector tubes
• Holds the tissue surrounding the BVs
• Is adjustable to accommodate samples of different sizes
• Is easy to clean
• Is easy to swap out different samples
• Is secured to the test bed and doesn't move during active data collection

I also took inspiration from sample holders on Briteseed's other test beds and used the following process to create the sample holders:

• Note down important dimensions on the existing test bed
• Obtain or create simplified CAD files for existing components
• CAD model the new components
• Conduct a design review
• Prototype the new sample holder
• Test the new components and iterate based on changes that need to be made

Pictures of the CAD models and the prototypes of the new sample holder can be seen below:

CAD model of new sample holder (front)

Fit test for the sample holder

Implementing the new sample holder

CAD model of new sample holder (back)

Fit test for the sample holder

Implementing the new sample holder

How can we integrate the HSI tool into the test bed for dynamic testing?

• Create a new tool holder for the HSI tool
• Add a jaw angling mechanism to hold the jaws open at specific angles
• Add channels to protect the fragile flex PCBs

How can we remove the need for metal jaws within our tool holder?

• Create an alternative tool holder that doesn't require the metal jaws (has "fake" jaws)

Similar to the sample holder, I first established a list of requirements, which are as follows:

• Keeps the sensor jaw flat and stationary during testing
• Maintains the jaws' orientation in relation to each other
• Allows for movement and rotation only in the LED jaw

Briteseed also wanted two versions of the new HSI tool holder:

• Version A: requires the tool's metal jaws to function
• Version B: could house the LED and sensors without
  the metal jaws

I then used a similar process as the one used for the sample holder to create the new tool holders:

• Note down important dimensions on the existing test bed and HSI tool
• Obtain or create simplified CAD files for existing components
• CAD model the new components
• Conduct a design review
• Prototype the new tool holder
• Test the new components and iterate based on changes that need to be made

Pictures of the CAD models and the prototypes of the new tool holders can be seen below:

CAD model of tool holder A

CAD model of tool holder A

Sectional view of tool holder A

Final prototype of tool holder A

CAD model of tool holder B

CAD model of tool holder B

Sectional view of tool holder B

Final prototype of tool holder B

How can we test and achieve consistent results during dynamic testing?

I collaborated with some of the other interns to troubleshoot and improve some of the automation on the test beds, including:

• Fixing faulty hardware (broken motor drivers, miswired cables)
• Identifying software issues (non-responsive limit switches or GUI buttons, loud motor movement, unactivated soft limits)

The video below shows a successful homing calibration after the other interns and I troubleshooted the test bed issues.