Our main story here on Last Week Tonight is covering our team’s dismemberment of an optical coherence tomography (OCT) machine. Don’t worry, we didn’t know what that meant either.

 

 

 

 

 

What’s that? This is old news, you say? Well, pardon the temporary absence of our host, but she’s back now after religious obligations and battling a mystery illness.

Back to the machine. Our source at Cleveland Clinic informed us that OCT machines are typically used for noninvasive imaging of the eye. It works by bouncing infrared light off of your retina, creating high res cross-sectional images that can be used to diagnose common eye diseases like diabetes-related retinopathy and glaucoma.

Cool. So …how does it do that?

Our team of expert engineers (read: grad students with one (1) screwdriver each) sought to figure this out. We started with the aluminum housing. The front panel was easy to disassemble, and secured with Torx screws. The back and side panels gave resistance, and were a much more intricate affair. (These were likely the portions not intended to be readily accessed). After removing the aluminum housing, we found a whole lotta wires.

It was clear that a significant amount of heft to this machine came from its casing. Our later analysis found that about 18.6% (~ 3kg) of the total weight came from the exterior, including the handle.

Affixed to the top panel was what we believe to be a grounding wire. It was too short to work around, so we snipped it with wire cutters to continue our disassembly.

 

 

Carefully untangling the entrails, we found free floating printed circuit boards (PCBs). As in: these key components were not secured to the housing, but rather were free to rattle around. We carefully unplugged their connections, and found more PCBs that had been secured. These were easily tackled with some Phillips head screwdrivers.

 

 

On to the PCBs. Those secured were held in place with pin and socket connections. Some were stacked using standoffs, ensuring sufficient airflow between the parts. We sacrificed a few wires in the disassembly, as they were connected to the side panels and needed to be cut to get deeper into the machine. 

 

 

 

 

 

 

 

Finally, we reached the optical components for the OCT machine. These were secured to plastic housing, which was modular and allowed for quick adjustments to positioning.

We found many lenses and long stretches of thin optical fiber. We also found my favorite component: a custom L bracket. No, not custom in a necessary way (strange angles, mounting holes, or material). Custom in a gritty way. It took me back to high school, where we’d save money by making L brackets from stock instead of buying “fancy” premade ones. The sensor that was mounted only used one set of the bracket’s holes. It’s an exercise left to the observer to determine if the different holes were for different configurations, or if the machinist had made a mistake and refused to make a new piece.

 

 

 

 

 

 

We turned our focus to the optical circulator, seeing if it was possible to unravel the unimaginable lengths of coiled optical fiber. We did not see the end, but left carnage in our wake. Focusing back on the sensor  mounted to the bottom, the fasteners securing its lid each had a dollop of something. At first we thought it could be glue, but loctite and similar adhesives are usually placed in the threads. It’s possible this served the same function as torque seal, which provides an easy visual indicator if fasteners have loosened. However, this explanation wasn’t satisfying either. Torque seal is typically used in automobile applications, or anywhere that maintaining correct torque is high stakes. This sensor was difficult to access, meaning the visual indicator was not readily available, and (to our knowledge) this machine was not subjected to high speeds that could rattle loose any screws.

dollops of something. We couldn’t remove the screws without stripping them

 

Satisfied that we had sufficiently gutted the machine, we went back to fully disassemble the aluminum housing. Once the side panels were removed, we worked on the handle. This proved simple enough. Using a flathead screwdriver for leverage, we popped off the plastic cover. This revealed a locking mechanism. We noted this function prior to scrapping the machine. The mechanism allowed the handle to be held rigid, so that when carried the machine would not swing and potentially get damaged.

 

Think of a mop bucket vs a plastic watering can. The former container is unsecured and easy to swing, while the latter has a fixed handle and does not sway. When this mechanism was engaged, the OCT could be carried like a plastic watering can.

 

 

The core part of this locking mechanism is the aluminum gizmo (see image below). One side is smooth, which allows the handle to freely slide and move about. The other side is pitted, holding handle fixed at specific angles. All of this came apart with a Phillips screwdriver and an Allen wrench. Post disassembly, we could easily identify that the side panels holding the handle were made from extruded aluminum. Though strange in shape, this provides increased strength with relatively low weight (only 565g per panel). It’s also easy to purchase off-the-shelf.

Ultimately, we were surprised that such a complex machine had obvious touches of human involvement. It was strange to see precisely soldered PCBs placed in the device, unsecured. The custom L bracket and hot glue made this OCT machine seem like it was made in someone’s garage, not by medical technology manufacturer BioMedTech. This could hint at a few things. Given its assembly, it’s likely that this device was not mass produced, and at most only a few thousand units were made. However, these details also may suggest post-manufacturing fixes made on the fly by a technician, meaning the device as it was originally made was prone to issues.

Certain parts seemed to be designed to be easily accessed by technicians, like the front panel which gave access to the unsecured PCBs (likely to quickly swap them out). Other sections seemed off-limits, like the delicate optical components that were held in a specific configuration by the plastic guides. For the entire disassembly, we only needed tools that could be found in a basic toolkit. In fact, we only used eight tools:

• Standard (aka flathead) screwdriver 
• Phillips screwdriver
• Torx screwdriver
• Needle nose pliers
• Wire cutters 
• Socket wrench / ratchet
• Allen wrench
• Vice grips

 (and arguably, the vice grips weren’t needed if we used enough force)

It doesn’t seem that this OCT machine was designed to be recycled, though it would be easy to scrap for parts. The combination of off-the-shelf and custom parts makes it modular enough to deconstruct and extract anything useful or expensive. The non-electronic components are mostly plastic, aluminum, steel, or copper, which have existing streamlined processes for recycling.

It was very fun to reverse engineer medical imaging technology, especially given that we could be selectively destructive since we weren’t putting it back together. Though disorganized, there was clear hierarchy to the internal structure of the machine, and its tough exterior ensured the delicate parts were safely housed. I didn’t consider that advanced tech like this would use a combination of basic modular parts like extruded aluminum with (crudely made) custom ones. I also never thought that stuff like this wouldn’t be mass produced, or assembled for peak efficiency.

That’s all we have for this week’s show. We hope to catch you next time, at our regularly scheduled hour, for what’s next!

APPENDIX

OCT Disassembly: tool and materials documentation
(https://docs.google.com/spreadsheets/d/1S1DQjNwIqb35mSERDYDsS35tdK7ONSxG/edit?usp=sharing&ouid=118355057400499546845&rtpof=true&sd=true)

Equipment-Take-Apart-Lab-Worksheet (PDF)