Towards total mind control
This is a controversial title but it quite accurately describes what we actually do. We realize NCN grant called Symfonia 2013. Within this grant, we are building complex measurement and control system for optogenetics research. The title could be good for Hollywood pictures, but the goals of the project are very noble. We are building new tools that help biologists and brain scientists from Nencki Institute to more efficiently study brain-behavior on animal models thanks to automation. Automation means better accuracy, fewer errors, more reproducible results. As a result, fewer animals need to be used for research and the whole procedure lasts much less time. The main goal of this research is gaining knowledge of mechanisms that cause epilepsy and various forms of addiction. Why are we coping with optogenetics? It is quite a new research discipline. Less than 10 years old. And there is a lot to invent and make it more robust. What is optogenetics? It is a method in which animal (and also human) brain connections can be marked with special proteins that are sensitive to light. The power of this method is that only newly created brain connections are marked. This means that we know which neurons are responsible for certain memories, tasks, new knowledge. Once neurons are marked, they are such for a lifetime. We can then trigger such behavior, memories, fear just by illuminating the brain with the light of a certain wavelength. We can also do the opposite - disable temporarily certain neurons that hold certain emotion or behavior just by shining light with a specified wavelength. Early experiments require deep brain implants, but recent developments enable doing optogenetic research without opening the skull. Anyway, since the brain does not have any sensors, animals don't feel any pain.
There is plenty of literature on this topic, some example is here
So what do we want to achieve within this project? The Holy Grail of the optogenetics is the ability to freely control large groups of the animals, without human intervention that alters their behavior. We want to know where each animal is located, how is oriented, what is actually doing, what is its brain pattern, temperature, heart rate and even if it is sniffing some scent or not. Moreover, we want to remotely enable/disable brain implants to study how we altered animal behavior by this. We also want to make it in a conditional way, i.e. once an animal tastes the alcohol, it triggers optogenetic action that disables memories of alcohol addicts.
There are existing automated cages produced by specialised companies. Example is Intellicage.
Below is some photo that presents such cage. In the corners there are mechanisms that are dosing food, shacks, drinks, scents. There are also gateways that close the passage to certain compartments.
These cages are widely used by scientific community. However for certain types of research they are not really suitable. Some animals need to have separate chambers to recreate conditions that are close to naturally occurring. For this purpose, special cage called Eco-Hab was built by Nencki team with help of ELHEP group.
The cage consists of 4 compartments linked with plastic tubes equipped with RFID antennas. The antennas are connected to readout module developed by one of our students (Adam Flis).
This seems to be trivial. Just take 8 readers, connect them and ready. The real issue is the mutual interference generated by the readers. The tags are passive so need excitation energy. Standard RFID readers generate their field based on antenna resonant frequency. This ensures that they always operate in optimal conditions, even when antenna changes its frequency i,e, by proximity to some object. This is a reason why the early version of the readout system barely worked. Each reader had his own operating frequency that was a source of interference for nearby receivers. We had to redesign the readout system from scratch. All receivers were based on HTRC110 reader IC which has the ability to externally synchronize its operation. All 8 readers were connected to the common crystal oscillator and equipped with tiny AVR processor that decodes the asynchronous transmission. That was far easier than using a large CPU with independent decoding tasks due to very tight timings required to receive serial bit stream from each reader. 8 such reader circuits were attached to the master CPU that reads results once they are ready and transfers them to the control PC via USB interface. The project is quite successful, currently, we build more pieces for labs in Canada. We also got a patent on the automated cage. We are also optimizing it for mass production. The readout module was designed in 3U form-factor to make future expansion easier. The next step was to equip the cage with additional functionalities. The biologists wanted to detect if the certain mouse is sniffing the scent placed in a dedicated corner separated by the perforated wall. The question was how to detect if the animal is really touching the wall with its nose. Only nose matters. Legs, tail, fur is not the case. So after some brain-storm, we invented the optoelectronic detection system based on a laser curtain. The system was built by our student, Paweł Rasiński. The idea is quite simple - we have a cage separated by the wall and want to check which part of the body a certain animal is using to touch the wall. We needed to know the cross-section of the touching part. Quite an obvious idea was mentioned IR laser curtain coupled with CMOS camera equipped with IR filter.
Below is some video showing the algorithm in action.
The efficiency of the algorithm was better than 95%. After coupling with the animal tracking system, it will be a useful diagnostic tool. Another issue was time synchronization. The PC algorithm takes some time to process the data. This time varies from frame to frame. For this reason, we developed an IR display that is installed in the camera field of view.
The image processing algorithm decodes LED pattern arranged in Gray code.
As I wrote above, this is part of a larger system where information about animal position and sniffing can be combined with optogenetics.
We explored several ideas for the animal tracking system. We started with an array of RFID sensors installed in the fall but finally, we figured out that it is too complex and does not give sufficient detection speed nor precision. Below is a photo of the system capable of reading 56 RFID antennas. IR LED display is visible as well.
Each module is a complex device equipped with Cortex M-4 CPU, FPGA, coil drivers, tuning circuits and protection circuits. The detection array had 3 functions: implant position readout, delivery of power supply and micro power Bluetooth transceiver wakeup.
The system was too complex and too slow. Mice are hell fast creatures and electronics didn't manage to scan the whole array fast enough.
Our clever students (M.Sowiński, D.Krystkiewicz, and P.Rasiński) came with an interesting idea. The implantable electronics are already equipped with the 3D magnetic sensor. We decided to have a try and use it for animal tracking. The module is tiny. roughly 10x8mm. Even though it contains nearly 100 components.
The board is heavily packed, we have a lot of interesting stuff in it.
So, the idea is to explore ability of using array of Helmholz coils
The first experiments were very promising, It looks feasible to obtain the positioning accuracy of just a few cms. With proper calibration, this could be even less. We are currently building a full-scale cage with 4 coils attached to the rigid frame.
Another aspect is the supply of the implants. So far they are used with rats and the animals are large enough to fit the small LiOn battery. They are charged once a day which is sufficient. In the case of mice, we cannot really place even the smallest battery. Its energy density is not high enough. The only solution is the continuous delivery of power. We decided do explore the RF field as a source of energy. We need roughly 30-50mW of peak power for every implant. With help of our Ph.D. student, T.Filipek we managed to build RF supply chain. It consists of a DDS generator, a power preamplifier, a power amplifier and a protection circuit that takes care of coil energy transfer.
Together with coil below it forms the complete RF supply chain.
Due to high frequency, the coil was built using litz magnet wire.
The RF coil inductance is very fragile to temperature, the position of objects inside displacement of the frame. For this reason, we are currently developing automatic tuning and protection system based on microcontroller and DDS generator couple with RF power detector. The design is currently developed by our student, Karol Sucharski.
Once the biologists discovered that our implantable unit can be used to stream raw ADC data continuously, we had to switch from Bluetooth to Ant protocol to fully utilize the channel bandwidth. Since the single receiver could not handle the data rate of several transmitters, we had to build multi-channel Ant receiver hub that can be expanded to the desired number of channels. This work is maintained by our student, J.Jarosinski.
All modules have form of 3U board, so they can be easily mounted into standard 3U chassis.