Ch 4 Flashcards
Have previous papers used TFs for depth resolved monitoring?
Yes, Pisano 2019 show the first use of the TFs for monitoring fibre photometry signals where they assess dopamine signals in the striatum as mice perform a behaviour task.
To my knowledge, nobody else has publishable results using these.
What are the main findings of Pisanello 2014?
Pisanello 2017?
- gold plated TF with windows to deliver light
- establih what angles of light are required for multomode derlivery of light regions fat away from the tip
- light derluvery length increases with increased NA and decreased TA
- the smaller the angle of entre the light will come out the top of TF, the larger, it will reflect inside and propagate
- illustrated site illumination in FITC and brain slices
- shown site with multi wavelength in dual fluorescent solution
- checked if shaking/twisting patch effected the light
- in vivo in striatum with silicon probe showed site detective excitation
)_____-
to show it could be whole brain or region specific optogenetics
- TF with no fold plating
- did ray simulations to confirm light patterns
- monitored inflammationn
- see increased modulation over intentiy and depth compared to FF
- implemented the galvo mittor for continuous sweeping
site directed opto in striatum
What are the main findings of Pisano 2019?
- did raster scanning along tf in FITC and monitored response with two hotodetectors to determine signal collection field
- found unform and depth specific collection field in fitxc and human tissue
- equalised power density along TF
- monitored in vivo dopamine in striatum woth depth-resolved differences.
How exactly is the TF made?
A heat and pull process where the tip of the fibre is heated and stretched until the ti is ~500 nm,
They set this with a NA of 0.37 and taper of ~2.4
How can the TF record fluorescence?
There is a change in the refractice index (the light refeaction level) from tissue to TF which makes it couple into the fibre.
Then the modes of light pass the same way as they fo on the way down – the site of entry is correlated with the angle of light entering the TF.
How does the taper angle determine the active region?
the tpaer angle decreased will increase the length of th e active region.
Pisanello found that increasing this shortened it most likely because the internal reflectance was too hgih.
How were those images of the fibres acquired?
The fibres were placed into a solution of high concentration fluorescein in a cuvette.
Then light was delivered through the system.
A camera was set up focussed though the cuvette and images were taken.
Why on the images are there some spects of fluorescence?
It is most likley the fact these fibres have been previously used so there may be some residue on them.
What was the function of the galvo mirror?
As we have 1.8mm of TF to work with, we needed a way to modulate where the light was being propagated out each depth of the fibre instead of it being along the whole shank.
This mirror modulates the direction of the laser beam in order to change where the light enters and leaves the TF.
Why for this system did you need to use a laser?
Because we are changing the focus of light on a rapid time scale and lasers are known to have a more focussed beam and a faster operation speed.
Why have you now got an additional photodetector channel?
Because lasers are more unstable than LEDs and can fluctate due to head differences. We also found that perhaps the way the laser was being driven - through the DAQ was causing an unstable power voltage and therefore, we implemented a photodetector to monitor this read out.
Did you use the 405 nm read out for analysis?
No, as we had out control group we did not use this. Also the changes were minor so shouldn’t have effected this profiles.
Why did the original laser deterioate?
This was purchased from a cheaper supplier so it is thought that there was some electrical faults that were perhaps not picked up as they do not have as vigiourous quality control.
Also, this laser was not based on a heat-sink which is important for temperature control in order to prevent damage.
Why was a glass slide ok for focussing onto a detector and not a lens for 405 nm channel?
Because the signals on here were very high as it was direct from the laser so even capturing a small part of this signal would be high enough.
What is a neutral density filter and what type did you use?
Why did you use this?
This is a filter that across all wavelengths, will reduce the light power at equal amounts.
*** need to check system
We used this because this laser was very high power and produced alot of light. To reduce it to working and safety limit we used this.
Why didi you use an aspheric lens that was length 50 mm focal length?
To generate a focal spot of <500 um and the appropriate distance of the system.
Why are some of the focal length of lenses different throughout the system?
As some were being coupled into the fibre they needed to be smaller.
What mirrors were used for first reflection in the system?
Broadband mirrors. These will reflect all wavelengths with equal transmission. These allow direction of the light path.
How is the light directed towards the photodetectors?
The light passes back through the fibre before reaching a 425 short pass mirror where light below 425 will pass and light above will be reflected. This removes the excitation light from our emission channels.
Then the light passes a 525 nm short pass dichroic mirror where light above 525 nm will be reflected before passing a 550 nm bandpass filter which is focused onto the photodetector by an aspheric lens. Light below 525nm will pass through the mirror before a 440 nm bandoass filter and being focussed onto a PD.
How is the light seperated for each photodetector? Is this not a mirror instead of a filter?
Yes, the diagram is wrong.
The light is seperated using a 525 nm shortpass mirror.
This seperates the light pass for 440 and 550 nm.
Why did you have a 1.25 mm ferrule setting for this system?
This is the setting that the TF can be made in and it works better as they are lightweight for the chronically implanted headcap.
Why did you have a 470 laer in the first system?
Because at first we discussed the possibility of using this as the control as this system was deisgned before we established the system needed reconfigured.
Why do you need to focus light onto the galvo mirror with a lens?
To make the beam focussed so that the light path is modulated appropriately.
What does the heat sink do?
The heat sink works as a fan to remove heat by dissapating it into air.
Therefore, it prevents the overheating of the laser and damage.
Why did you not need as many mirrrs once the new laser was put in?
Because we did not need to combine the two light paths
Why was the laser grounded?
The laser was grounded because we were seeing low/high frequencu noise. We grounded the laser and the DAQ to remove any outside encirtonmental contributions to this noise.
What do you mean by saying that the galvo mirro is a single axis?
The galvo only moves across a sinve direction. Some have capabilities to move in more than one.
How big is the angle range that the galvo mirror can go?
Do not know the exact angle, but the voltage range is -5 to 5v. I think this was around 12.5 degrees.
Why will the TF form these uniform rings?
Because the light should be moving up the TF shaft in a uniform manner and therefore from the tip, the light cone should form a unifrom circle.
What is the pupose of looking at these illumination rings on white paper?
It illustrates if there is uniform light being propagated at the TF and is representative of how well the light is aligned.
If the circle is dim, damaged it can represent dust on the ferrule or that one of the axis is not aligned appropriately.
What light output from the patch cable were we looking for and how could we tell if the system alignment was good from these values?
~2mW at the highest driving voltage as this was our safety limit.
We would look for consistent values across GVs maintained across days.
How did you monitor the voltage read-out on the 405 nm photodetector?
We had a built in online function which processed the signals as the others are.
How exactly was this system recalibration completed?
The lightoutput at the end of the patch cable with and without the TF attached was measured at increasing gv and laser power. The laser power was kept within the ange used for in vivo experiments as this gave the best estimate.
How was a linear regression modle for each GV formed?
The light output measured for the patch cable was plotted against the laser voltage. Then a linear model was fitted and the slope and y-intercept were used for all recordings.
Why was an initial bleaching of >8 hours required?
Because this would remove the autofluorescent particicles from the systwsm to a baseline level which would only need topped up before each recording.
Why did the bleaching protocol progress over time?
Because we realised that the af varied across galvo voltages which is why we added on the galvo sweep.
Then, we noticed some inconcsistently of baseline fluorescence across day and thought this was perhaps due to the af state of the system being varied so we implemented the bleaching threshold to try make this as consistent as possible.
What was wrong with bleaching at one galvo voltage?
Because there was varied AF at different galvo voaltages and bleachigna t only one votlage meant some af was not being bleached.
Why was bleaching completed at different voltages for each laser?
Because these lasers had a different max votlage.
Why was the glavo mirror drieven at 1Hz for bleaching?
Because this sweeped through all galvoes over 1 second which didnt require a rapid effect and so wouldnt be taxxing on the system.
If this bleaching threshold changed overtime, why and would this effect the results and day-to-day consistency?
Because occasionally the baseline laser intensity would be increased which meant if we continued with the original bleaching threshold, this wouldbe lower than usual.
This wouldnt have affected results as we didnt change this throughout a particular recording apradigm. This was more common if the laser had not been used in a few days.
For bleaching, why was a sampling frequency of 1000Hz used?
Because we do not need rapid signal detection. Sampling every 1 s provides a good enough image of how the af state is.
What is meant by syaing analogue signals were digitised for analysis?
This means we used a data acquisition board such as the NI DAQ to convert the analogue signals that exist in 0-1 from the system over to digital so that data can be processed appropriately.
On the other hand, signals were converted to analogue to drive appropriate systems.
Why was an AF recovery experiment completed?
As we were tracking a time-resolved profile, we needed to see what profile AF was recoverign to see if this overlapped withour recoridng. As if so, this would make it difficult to confirm if our signal was truely m04.
This helped establish the time profile of AF across depth for further analysis.
How long was the bleach and AF for this experiment?
How was it completed?
Bleach was 30-mins through sweeping galvo at max voltage.
AF recv was every 30-s for 12-h.
Why for AF recv did you have only one illumination repetition?
Because we were just wanting a snapshot of the fluorescence at this point and wanted to limit the bleaching thorugh this protocol.
Why was a sampling freq of 5000 Hz used?
Because we were modulating the laser with on and of periods in a ms range so in order to capture alll vital data, we sampled at this frequency range.
Why did you need to implement light protocols?
Because the light profile from the tf was not equal alongt the shaft which would make it hard to infer what the fluorescence data means as the brain tissue would not be excited at the same powers.
Why did you need to know the coupling efficiency of the TF?
Because this is a measure of the light transfer of the tf and allows us to better estimate and modulate the light powerr to suit the particular fibre for a better estimate of the light power across mice.
Why do the alignment limits with TFs exist which cause the dips in light power?
As the light will internally reflect more at higher GVs, more of the light energy will be lost in the cladding.
Why did you choose light protocols for
- light power
- power density
Because light power is the most conventional way as this means the same tissue is being illuminated at equal levels so we can get a more reliable comparison.
Power density was done as the light cone of the tf is different across tissue so this equalises across tissue which we thought would be desirable to try.
How will the light protocols compare to when the TF is implanted in brain tissue?
We at first did not consider the changes this would infer. However, using 2p slice images we see that the light from tf scatters differently in the brain tissue which may mean the estimated light protocols may have some error.
Why was 60/80 uW used?
Because this was in the low range which provided a good m04 signal.
Why did you use fluorescein for the image protocol?
Because we needed a larger volume of liquid and this was cheaper. Also, this protocol had already been established for fluorescein.
How did you determine an optimal voltage range for the image light protocol?
We would usually use one of the lower limit of voltages used for in vivo experiments, but we would re run this protcol adjusting the voltage appropriately if it appeared to high or low.
We wanted to see fluorescence along the fibre but not a saturated signal as this would be out with the pizel intensity range of the camera.
Why did you sweep from -4.5-4.5 for image protocol when you never use these values?
Just to get a full emission profile but we do not use that data.
Why do you complete the image protocol at a reoslution of 0.1V galvo?
This matches what we sample at in vivo. It also over samples for the estimation of the req power ratio.
Why were the images for image light protocl taken every 0.5 s?
Because there was a lag between the system and camera so while the galvo changed at 1000 Hz, sampling the camera at 500 Hz ensured we acquried the appropriate image.
What type of camera was used for image protocol
This was a small USB powered optical camera that had a magnification lens.
How was the active reguon of the TF labelled for image light protocol? How was it visually inspected?
A line was manuall drawn along the TF shaft where light was propagated. The goal was to mark along the whole active region of the TF so we can measure the light intensity.
We created a video of all frames with this line drawn on and we accessed if this followed along all regions where light propagated.
What do you mean pixel intenstieis were calculated for image protocol?
Along the line drawn along the TF shaft, the fluorescence intensity was measured by determining the pixel strength at regions along this line.
What do you mean by saying that the light distribution is different along the TF?
At higher galvos, the light dissipates in a downwards manner which means the light can pass over more GVs.
How does the in vitro light protocol calculation work?
The fluroescence has been recorded at each depth.
Due to rules of fluorescence, this is a direct measure of light output.
Now we want to create a ratio scale which will represent the depths with the greatest and lowest power across depth.
This is done by calculating the reciprocal value of fluorescence and normalising this against the minimum.
How does the image light protocol calculation work?
This measures the power density at each depth through complex calculations that determine the area of beam based on tha taper angle - sypplied by our collaborators.
We then need to normalise the power density values to be standardised and comparible which is done by subtracting the minimum PD at each depth.
Then, the recirprocal power density is calculated to illustrate the ratio required to equalise it across depth.
For fibre calibration why did you use those powers?
I used powers that were within the range used for in vivo experiments as this provided the best estimated power.
Why do you think 3V galvo provided the most accurate power estimations?
This was the centre galvo so most likely took a good balance between the higher signals at lower galvos and lower signals at higher galvos.
Why were sync channels of the laser and galvo needed?
Because this provided us information for post-processing for when the laser was on and at what galvo voltage each measure was at.
Additionally, it can be used to ensure the correct voltages were applied.
Did you monitor the laser voltages you applied?
Yes, we also acquired a mat file that has the laser voltages at each galvo for each light protocol.
More recently they have been used to help towards acquiring correlation.
How did you choose your desired powers?
As we couldn’t choose irradiance due to the inability to determine the area of light of each taper, we were unable to directly compare with FF experiments.
Therefore, for first recordings I started with a wide range of powers and found that below 60 UW, the profile was very low and the signal was noisy.
How does the formulatas for calcularing the desired powers work?
In step 1 for all light protocols we are calculating the required power needed for each galvo. For original this involves using the coupling efficiency so account for differences with each TF. Here, we devide the coupling/100 by the desired power. For models, this involves using the power ratio divited by the desired power to
Next step involves transferring this required power over to voltage usits using the prevosusly established intercept and slope. For models the coupling effciciency is also used by dividing this by req power.
How is slope and y-intercept used?
These are used to transfer the req power over to req voltage so we know at what rate to drive the laser.
What are the two steps in the power calculations for?
step 1 is for establishing the requried power
step 2 is for establishng the required laser
What was the additional project for collaborators?
They took brain samples that were sectioned into 300 um thickness and they put them into a two photon set up where they implanted a fibre and imaged the profile. This was to see the scattering profile and what the fluorescence looked like in live tissue.
What was the implanting issue in FAD44?
The fibre broke upon implantation.
Why would you not know the bleaching protocol for APK?
i have not noted what specific protocol was done, rather just that bleaching was completed.
Why were the powers sand sampling intervals changed for APK?
Because i was struggling to get a reliable signal so various parameters were changed in order to boost this. For wxampke, powers were decreased to reduse bleaching or increased to get a better profile. Sampling intervals were increased to reduce bleaching.
Why is there such a big age range in mice used for head-cap surgery?
Because we wanted mice in the young age range for genus as this is where it seems to be more efficous.
Whereas we were using mice of various ages fro PK in order to acquire a strong signal
How does isoflurane work?
Isoflurane is a general inhalation anesthetic used for induction and maintenance of general anesthesia. It induces muscle relaxation and reduces pains sensitivity by altering tissue excitability. It does so by decreasing the extent of gap junction mediated cell-cell coupling and altering the activity of the channels that underlie the action potential.
MOA of naropin
Why is it used?
Block the generation and conduction of nerve impulses. Specifically, they block the sodium channel and decrease chances of depolarization and consequent action potentials.
Local anaesthetic used to knumb the area of surgery to ease the pain and promote recovery of the aniaml.
MOA of vetrogesic
Why is it useD?
In summary, buprenorphine is a potent, long-acting analgesic acting at opiatereceptors in the central nervous system.
It is used to reduce pain in mice.
MOA of rimadyl
Why is it used?
The mechanism of action of carprofen, like that of other NSAIDs, is believed to be associated with the inhibition of cyclooxygenase activity. Two unique cyclooxygenases have been described in mammals. The constitutive cyclooxygenase, COX-1, synthesizes prostaglandins necessary for normal gastrointestinal and renal function. The inducible cyclooxygenase, COX-2, generates prostaglandins involved in inflammation.
Another pain killer.
Why did you inject saline before chronic surgery?
To hydrate the mouse with fluids since it is losing some throughout the surgery. Helps with recovery.
Why did you choose the ear bar approach?
Because this is a really good way to use the stereotaxic to stabilise the mouses skull.
What is the purpose of betadine?
It is an anti-septic solution which will clear the skin of infectious particles to reduce the risk of infection.
Why did the peristeal membrane need to be removed?
As this is a sticky membrane acting as a protectve barrier to the skull. However, if this is not removed it hinders access for burrholes. Also, the dental cement will not stick to it as well and increases the risk of the headcap falling off.
What were the differences in your implant sites and why?
We moved more lateral as my implants appeared to be sitated more medial than desired so I was trying to offset my error.
I then moved it more lateral and posterior to increase chances of hitting my target.
What were the skull screws needed for?
To stabilise the headcap.
Why was a straw attached to the back of the headcap?
For headfixation. The mouse would be attached to a metal rod by this.
What is kwikcast?
silicon gel that allows the brain tissue to be protected from dental cement.
What is meant by 4-day summary analysis?
The full 4 day recording data.
How will having different implant sites differ the results?
Potentially one of our strong signals are offsetting by a weaker signal which reduces the intensity of our collected signals.
Why are some bleaching in PK exps at 3.5V?
Perhaps when laser was dying or when I was trying to establish appropriate bleaching protocol.
Why did you increase the number of reps?
Because the laser stabi;lity varied so to get a better median signal.
Why did you increase the sampling interval?
Becaue the dynamics of the plaques was very slow so we did not need rapid sampling. And to reduce bleaching.
What is meant by:
- investigative time samples
- inaccurate power estimation
- laser dying
- needed further habituation
1) We just started the recordings so were unsure how long we would want to record to see the fluorescent profile. So we did various types like 8 hours seperated bu a cap and so on.
2) due to the inaccurate y-intercpet/slope, the powere applied was not estimated correctly.
3) the first laser had an electrical fault and began to deteriorate in power throughout recordings.
4) mouse was not comfortable being head fixed or being in the open chamber so needed to be habituated to this process more.
Why was a red bulbed light used?
Because its wavelength is out of the mouses spectra so they cant see it and because it means I can still see them and the system without the light itnerfering with the recorderd signal.
How was the head-fixation process?
Using the straw to attach the mouses head to a metal rod.
However, later just scruffed the mouse.
How did mice behave in this open recording chamber?
The bejaved reasonably normaly and slept, groomed, ate.
Why did you want to investigate the pharmacokinetic profile?
Because we wanted to see the time profile of the increase in fluorescence and how long it took to return to baseline when recording. Additionally we wanted to see if the system was able to detect changes in the concentration of m04 as a sign if it would be able to detect changes due to plaque modifying effects.
For chronic recordings why did you have a day 0 baseline?
Because this was a fluorescence signal of the brain tissue without m04 so could be used as a baseline signal for a control and normalisation.
For pharmacokinetic, why the 30-min baseline?
This provides us with a baseline signal on the day for normalisation.
How did you decide the duration of recordings?
After several recordings of monitoring for 10+ hours, we seen that the maximum signal appeared around 2/3 hours and therefore minimised the recording protocol for ease and to increase reps possible.
What is meant by an incomplete recording?
That recordings were not completed across every day - sometimes happen due to unpredictable events like i got an appointment, i was sick, there waa a fire drill etc.
Why did you investigate the re-dose protocol?
Because our goal was to do longitudinal recordings so we needed to see if we could maintain a strong fluorescent signal.
Did all recordings follow the same recording schedule?
The majority did bay the first few animals as this was recordings where I was trying to see what the best lengths were.
What do you think the fluorecence would look like a week after this protocol?
After PK - I think the signal would be returned to baseline with potentially some bleaching of the AF of plaques.
After RD - I think the fluorescence would be maintained, but after longer peiods like a month this may be limited as binding sites would be limited.
Do you think less redosing could be completed?
If the length of recording was reduced yes as the binding seems to be stable over lengths of time as long as the signal is not bleached.
What do you think happens after these 4.5-h after injection?
It seems like the max signal has been reached as the fluorescence doesnt increase on the next day so it seems to stay stabley bound.
When this signal decreases, do you think its bleaching or clearance? How would it clear?
There is not much information on the clearance of m04, in fact other studies suggest that once it is bound it remains bound for several days +90 days.
So this to me suggests that this is bleaching of the signsal.
By redosing like this, how would this affect teh binding site? Could the binding become saturated?
Yes, this will eventually bind and prevent alongation of the fibril which will prevent plaque growth.
How long were mouse recovering?
We allowed 5 days for recovery but usually I wouldnt record for over a week after surgery.
How were mice habituated to this?
Allowed to explore the chamber for several days.
Habituated to handling and scruffing and fake injections. and touching of the head cap
Habituated to being tethered.
Why were the photodetectors recorded from continuously?
So we can see the change when the LED is on and off.
Why did you only go up to 4.5V galvo?
As we wanted to include some negative values to account for some variations in initialisation but wanted to maintain the same pulse duration to prevent bleaching.
Also, as this is only a small distance this was suitable and still was able to mornitro forom our desired brain region.
Why was the perfusion now completed with naropin?
Because our onsite vet would rather we use this as it is stronger and more effective than lidocaine.
Why was the brain sectioned into 50um slices instead of 100?
Because the TF is much thinner than the conventional fibres so this would make it better to find the implant on histological sections.
What is pentobarbital?
It is a barbiturate which is a sedative which when in given an overdose can kill the animall.
What is lidocaine?
Lidocaine is a block Na+ and K+ ion channels and regulate intracellular and extracellular calcium concentrations through other ligand-gated ion channels.
It is a local anaesthetic
Why were chronic sections stained with TF ans Iba1”?
Because the TF are really thin so sometimes it was not possible to find them on the sections.
Therefore, we had to use iba1 as a marker of micorglia to identidy the inflammation surrounding the TF.
TS was used as a backup incase there was an issue with M04 staining.
What is iba1?
onized calcium binding adaptor molecule 1 (Iba1) is a microglia/macrophage-specific calcium-binding protein
Therefore it is a marker for microglia
What is blocking solution and why does this need to be completed?
Blocking is a step in histology that is used to prevent non specific binding of antibody to the tissue.
The blocking solution is normally a serum against a specific type of animal, the same as the secondary antibody, as this contains antibodies that binds to non specific sites.
A critical factor, though, is to use serum from the source species for the secondary antibody as opposed to the source species for the primary antibody. This is because serum from the primary antibody species would bind to reactive sites, but the secondary antibody would recognize those nonspecifically-bound antibodies along with the specific antibodies bound to the target antigen.
Why is the primary antibody incubation completed at 4oC?
For convenience and also it means that the componenets of the solution will not evaporate.
What is normal goat serum?
Normal Goat Serum, Cat. No. S-1000, is essentially spun whole blood, filtered, heat inactivated and preserved with the addition of sodium azide
Why was the exposure time diffreent again? What was it here?
It was 500 ms as I as completing the imaging for the majority of samples and found this was better for these 50 um images.
If you took all three channels, why do you not show them?
It ended up the majoruty of sampels you can see the rough fibre track in the m04 sample. Also, in the m04 channel plaques aeevidence so there wa no need to show ts.
How were sections with the TF identified?
We would acan each section with the micriscpe to find sections that had samage. I would confirm this by looking at the Iba1 channel to see inflammation at this track. I would go through all sections to identify the trart and end of this track.
Why was their channel baseline added?
Because there was a grounding issue like before so this made all the signals at 0V so that they could be understand easier.
Why were the sync pulse detection values different for different lasers?
Because the max voltage was different so the pulse amps varied.
What is meant by sweep points across the TF? How was this calculated?
This was the number of measures along the TF shaft.
This was calculated by taking the length of time light was taken to sweep the TF which was 10 ms.
This was then multiplied by the sampling frequency (5000) to determine how many data points per second there was.
This was then divided by 1000 to change it from ms to samples where we got 50.
What are empty points? Why would there be initialisation delays of the mirror?
Like all of the system there was sometimes an initialisation delay where it took a couple of milliseconds to reach the appropriate voltage.
Therefore, we added 10 data points at the start of each pulse to account for these errors.
Why did computer lagging occur and how would this effect the recordings?
Sometimes because of memory or other due to storage.
Sometimes the codes running so much as well as the online matlab analysis was hard for the computer and caused lags. This would result in the data recorded not being saved.
Why was 5% chosen as the SD away from median file size?
As this is a standard way to determine a difference by taking a 95% confidence interval. For example, this would mean that the data im expecting to be normal will be in this 95% range, with these wrong files, in the 5% range.
What type of interpolation was used?
Using the built-in matlab interpolate function for linear trends.
Here we apply the time samples, the data (with the empty NaN samples) and then indexes of the data points that need interpolated.
This involves a formula where the points of the data before and after are used to predict the point in the middle.
How was the connection between system and mouse disrupted?
Sometimes the patch cable would become slightly deattached.
Why was 95% of the global median signal chosen?
As this means the signals that were correct 95% of the time were maintained, which removes the dodgy connections.
Why was Norm and Model used?
Because there are various components of the recording with can interfere with the collected signal. These analytical approaches allow the possibility to extract them to get a better signal.
For Norm, why did you calclculate the moving median before calculating the median for F0?
I done this always to reduce the noise but in this case it doesnt really matter since I take the median anyway.
What are the limitations of Norm?
We are normalising against the median of day 0, which of course will have variations across time that we are not accounting for.
There also can be variations from day to day which may skew the normalisation approach.
What are the limitations of Model?
Sometimes the sample size is low which means there are not many data points to generate the model, making the goodness of fit quite poor.
Differences across days.
What are the limitations of original?
The light ourput varies across the TF so we cannot account for diferences or compare across depths.
What are the limitations of
- raw
- inV
-im
Raw - it contains background signals which mask the m04 signal
InV - not done in brain tissue
Im - sometimes has massive power at higher galvos which can bleach m04
Why is the moving median over 50-mins calculated?
Because this is over several time points to reduce the noise and fluctuations across recordings.
How is Norm calculated?
Why did you determine relative change?
What does relative change mean?
fluorescence - F0 (median over day 0) / f0
This is completed for each parameter.
We chose relative change because it allows us to compare the difference from day 0 over time.
This means the difference between two values.
For Norm, why does it depend on power conditionally on light protocol?
Because the power will vary depending on the lgihr protocol since we modulate the laser power for each one.
How can the model be reflective of the AF in TF,patch,brain?
Because we use the day 0 recording which is the baseline signal with no m04. Therefore, we should see a the change in AF over time.
Why would you want to increase the number of k-folds?
Because it would help promote the fitting of the model.
How does the generation of the model occur? What is the step by step process?
We split the day 0 data intwo two subsamples by taking every second data point.
This data is plit into test and train data where a linear model is fitted and the coefficients, and goodness of fit parameters are completed by using the test data.
Finally all folds are combined by calculating the mean coefficients for the final model.
What is test and train data?
Test data is the data you will use to test the gootness of fit og the model We use different data than the model has been trained with to avoid overfitting and bias.
Train data is the data used to generate the model.
Why did you go for linear model? Did you test others that may fit better?
For simplicity we wanted a model that had only two coefficients. Therefore, we tested linear and exponential.
Exponential did not provide a good gfit and linear actually was the most reasonable.
How was the model fit across recordings?
Some were better than others, especially as the sampling interval was reduced this reduced the fit. But across models the rmse remained low abd the predicted profile looked similar.
How were the folds of the model combined?
We took the mean coefficients.
How was the goodness of fit parameters tested?
What is the:
- r2
- adjusted r2
- rmse
By applying the test data to the model and then running the built in matlab functions for the following paramters.
r2 - statistical measure of how well the regression line approximates the actual data.
r2 adjusted - takes into account both the goodness of fit of the model and the number of predictors in the model.
rmse - average difference between a statistical model’s predicted values and the actual values.
What is n in the modelling equation?
some noise that we will be unable to detect with this model
How is F(t) in the model?
The model is fluorescence as a function of time
What are residuals?
Residuals are the difference between observed and predicted data
In the example, are these r2 and rmse not bad?
you would want your r2 to be as close to 1 as possible, so this model shows average fitting. However, teh rmse is very low so you can see there is not much difference between the original and predicted data which suggests a good fit.
Why would the goodness of fit vary across depth?
We believe this is because the AF componeent occurs higher depths and therefore can have greater variance across days which may account for this variance.
What is the black line on raw data ?
This is an example of when there has been a disconnection between patch cable and mouse - this data was processed before I changed the protcol which is why it is still shown on the images.
In this example, why is there a sudden increase in F?
Because the m04 has entereed the brain and binding to m04
Why was data at 120/440 used? Why was galvo below 0v removed?
Because we wanted to use 440 as this was the m04 channel.
120 was the highest power that was used across all recordings which allowed us to have a more reliable and consitent analysis.
bleow 0v galvo was removed as this was extra data points which would just reflect what 0-1v gsalvo shows.
Why did you choose to show median fluorescence for APK?
Because the signal was quite noisy so this allowed us to get an average signal without the data being skewed due to outliers.
Also we do not have a baseline recording in this to account for the AF.
How did you calculate the change in fluorescence across galvos for APK?
We took the median fluorescence at 30-minutes and at 210-minutes. Then, the absolute change was determined by subtracting the 30-min fluorscence to see the increase from baseline.
Why was 550 and 405 channels not used for analysis? What was their purpose?
Because they do not contain the m04 signal and we found the 440 nm signal provided good signal without the requirement of 550.
Also, 405 was used to online confirm the stability of the laser. Potentially further analysis could be to use this to correct any changes throughout the recording.
How was the colour range determined for heatmaps?
The colour map was the min and max signal across all of the days recordings.
Why was the colour range varied across all analytical approaches and exmaples?
Because they are all normalised in different ways which resulted in different values,
The norm should be higher to represent the change if present.
The model values will be very low because it will be a much smaller range since the background signal will be subtracted.
How was the data extracted at each chosen time point for summary analysis?
What are the pros/cons of this?
In this particular analysis, the data was extracted only at that specific time point, meaning that we are not taking the average at a certain time.
The pros are this is a simple approach and this is present across all recordings and its reflective of that exact time.
Cons are this will not account for a noisy outlier.
Therefore, it would be better for future analysis to perhaps have an average over this time point.
Why did you choose those time points for summary analysis?
Because all recordings should contain these time points and they are reflective of the recording.
Every15-mins shows the gradual increase in fluorescence. However, I do think that if I decreased the sampling, the rise would be more obvioous and the statistical power may be higher.
For exclusion, why would the power not be calculated properly or the bleaching not done?
In the first recordings the power was sometimes not calculated properly due to the recalibration still being optimised.
Bleaching may not have been done fi the laser was dying.
What was the exclusion criteria for chronic recordings?
Not all days recorded
Powers inaccurate
Investigative time samples
System issues
How was 1.62 mm calculated as the active region of the TF?
As 5V is representative of 1.8 mm, we scaled 4.5V appropriately. So it is an estimate.
Was the iba1 stained image used for alignment? Why?
Yes, we used this to full identify the deepest region of the TF as this was sometimes difficult or impossible to see with the M04 channel.
Therefore we aligned the fibre length from the deeperst TF region on iba1 and usign his we could manually transfer over to mark on the GUI.
How did you manage to mark the TF track on m04 images? What is the purpose of iba1?
Using the estimated track from the iba1 and m04 images, we manually marked on the aligned images by copying.
How is the estimated length of TF length determined?
They took the most dorsal and ventral z coordinates and subtracted them.
Why is there some differences between estimated and actual TF length?
Mistakes in the alignment, warping etx.
Inaccurate marking of the fibre
Inaccutate regression model
In accurate implantation
shrinkage of brain tissue from cyroprotection
What is the goal of TF identification and how is it completed?
We want to get a matrix of coordinates of the TF track which have the same sampling interval as the photometry measure.
For this the scaling factor must be considered as there will be differences between the photometry resultion and histology resolution. This is calculated by dividing the estimated length by actual length.
Then the TF resultion is determined by dividing the estimated length by the number of measures and then multiplying by the scaling factor.
Then, from the most dorsal coordinate, equal TF resolution values funtil the most ventral coordinates are determined.
Lastly, a two step regression model is completed where the coorsinates are estimated using the AP and samplign depths and then ML and sampling depths.
How are the brain regions the fibre passes determined?
The amasine using the allen atlas list of brain regions. Each of these brain regions are given an id number relating to specific coordinates.
Therefore, the function can run through each of the coordinates estimated for the fibre track and extract teh brain region ID.
For correlation, how was the TF recording chose?
We used the last pharmacokinetic recording and took an average of the hour 2-3 signals.
This was chosen because this was the one experiemtn all animals went through where they would have an increase in fluorescence.
Limitations with this is that this was not the last recording for these mice so their plaque pathology may have changed since.
Why were mice excluded from correlation analysis?
Due to image quality - damage, missing sections, missing track, warping
Due to estimated track - if the depth is estimated to be way longer usually due to warping of the section
Recordings - if the recording had been excluded previously
Why is there a variation of days from PK to perfusion?
Because other mice went through other recordings after this.
Why was 200 um used for plaque quantification?
Arbuityry but consistent with the FF measurement.
If the TF track doesnt cause damage, why is the contralateral hemisphere required?
Because you would often have some damage at cortical regions where the fibre has entered either due to the thicker core or when remocing the headcap. Also the fluorescence tends to be higher in regions of AF which could affect the automated detection.
Why would data be missing for TF data in correlation analysis?
Previous exclusion - inaccurate power, incomplete recording etc.
Why was a KS test used?
Kolmogorov Smirnov
To check the normality of the data. We used this rather than shapiro wilk as we had a greater number of samples.
Why was a kruskal wallis test used?
to determine if there are statistically significant differences between two or more groups of an independent variable on a continuous or ordinal dependent variable.
We wanted to compare differences across time points at each depth.
This is non-parametric optic that can account for differences in sample size in different groups.
Why were more 5xFAD- mice used for APK?
Because we had trouble confirming why we were getting a signal from the fad- so we wanted to see if this was consistent and try determine what was causing it.
Why was this implant site chosen?
Because it passes a plaque dense and plaque sparse region
Why were mice recorded from several times?
To follow the three Rs - reduce.
We can complete several different recordings on the same mice to get the most use of out the mouse.
How could the AF be contributing to these in vivo signals?
The AF in the patch cable could be recovering on the same time profile as the injection of M04.
You mention that APK show a stronger increase around 3/4V but across all there is a similar increase… why?
In these examples, the increase appears to be more promonent at the higher galvo from baseline.
How is the AF recovery got a exponential and linear trend for bleaching and recovery?
Bleaching takes an exponential trend as it has a slow decline before reaching a plateau
Recovery is leanear as it continues to increase at a steady rate.
You mention that for AF recv the increase is higher at higher GVs… is it?
It is increases across all galvo voltages, with it looking most strong around 3V.
What type of normalisation is present for the APK recs? What are the units of dF?
For the heatmaps, this is raw median fluorescence.
For the galvo comparisons this is absolute fluorescence as its subtracted and therefore the units are acttually (V).
Was it always the case that light output decreased with increasing GV? Why would this vary?
Not always although this was the most common.
Sometimes at the tip of the fibre the power would be decreased.
This will be due to the alignment of the system and due to the variability across fibres.
Why does the light output decrease when you add the TF?
Because of the coupling. While all parameters are matched, there will be some light lost upon this connection of the ferrules.
How does the recovery of AF work?
The process is not fully understood but probably the molecules in the fibre cladding restore their original confirmation.
Did you apply the three analytical technieus to this data… why?
No, we did not for APK data as we did not have the baseline day
How is the data normalised in 4.3.2B/C?
Average at 30-mins and 210-mins is calculated. 30-mins is subtracted fro m 210.
Why did you miss your target when implanting the TF?
Some skulls are different sizes and manual error, also some angle of the skull may contribute.
Are you still expecting a signal in the site 1/2 implants?
Yeah, at these deeper regions there is an accumulation of plaques but they are smaller.
Why should the AF be minimised in the chronic recordings?
Because we implemented additional measures to reduce the likelihood of af in recordings.
This was introduction of a different patch cable, bleaching protocols.
Why would you not expect to see a rise in fluorescence in 5xFAD- mice? What happens to the M04?
Because they do not have beta sheets in amyloid fibrils. Therefore, the m04 should be cleared from the brain. According to the original paper the unbound m04 is cleared through the original ways like csf.
Why would Norm show a faster inclune?
Because we are normalising against baseline which will make the detection more sensitive for smaller signals.
Why does the signal decline?
Because there is no more m04 entering the brain and the laser is continuously being illuminated resulting in bleaching of the fluorophore.
Does m04 remain stabley bound? How long will the signal remain?
According to other papers such by tara speirs jones, they see that m04 can remain stably bound for 90+ days.
Why may the original profile not allow the depth-resolution?
Because the light profile is inaccurate.
What do the individual traces on the graph mean? Why is there some variance across these and across days?
These illustrate each recording profile.
There is variance because each mouse is a different age so will have a different plaque load, also they may metabolise the drug differently so a different max concentration may be reached for each mouse.
Across days can also be because of some variations in AF of the system or laser stability.
What do you think of pseudo replication and will this have skewed the data representation in this case?
In this case we done it to complete as many recordings from the mice as possible. Also, when we first got signals from a particular mouse, we repeated to ensure this was a positive signal.
In this case, no most of the repeat recoerdings have been excluded.
Why would you get significance in the first compairosn, but not in multicomparison test?
- sample sizes in multicomparisons not providing enough power
- bonferroni causing conservative approach
- false positve in the kruskal wallis
Why in pharmacokinetic recordings are you seeing a significant difference at 3/4V across appraoches?
Because I am getting a stronger decline in fluorescence at these galvo voltages. I believe this is possible because there is less/no plaques in this area and therefore the laser will be bleaching the surrounding autofluorescence causing the decline.
Why for original protocol may there be significance across galvos if you cant see much change - is this because of the scale?
- the kruksal wallis could have picked up a false positive that is not deterected between groups
- sample size is too low to gain statistical power
- the variance is large which is shown by the low profile that cannot be detected
- the scale may contribute to this hiding some trend but no dignidicance was seen and this was mirrored in all other groups, consistent with Norm showing stronger incteases and significance
You mention that comparisons not shown in the tables were not significant, but how non-significant?
P values were > 0.8 or 1.
Why are you getting alow of p-values = 1?
Because there was no substrantual change in distribution of the values.
How did you choose the example data?
By showing the strongest and depth resolved signal.
Why do you think you see this declining signal on day 0 for norm data? Why not in model?
Because this may represent bleaching of AF throughout the recording.
Whereas in the model data they have accounted for the AF and therfore we dont see the decline.
Could this small rise in fluorescence be due to AF
No because we do not see it in the 5xfad- mice and not at every galvo.
Why do you sometimes see a rise in fluorescence before the time of injection?
Becuase of the way I have sampled the data - I have taken only the sinfular time points, rather than calculating the average over a certian time bin which will not account for variations in noise.
Also, other contributing factors such as the laser and tissue AD can contribtue to temporary increases in F.
In model, why are some of the values in a negative range?
There may be a decline in fluorescence from day 0 which may be due to the differences in state across days.
Why is there alot of variation between recordings?
- differences in system state - level of AF before recording and recover
- laser stability -low freq noise could increase over days and recalibration cannot be completely fool proof
- differences in the age and sex of mice - different level of plaques will cause different level of increase
- injection and metabolism of drug vary across mice
- the implant is vary so some may show greater increases at other depths
Why in in vitro pharma model is there one outlier recording where the fluorescence decrease across day -0 and 1?
- differences in system calibration
- the laser was recalibratied each day to be optimal but some times its not possible to be perfect
- this suggests a recording where the power used was esimtated to be higher than desired but it is unclear where the issue was for exclusion
For image-based recording, why is the signal saturated on day 0?
This example shows that the F on D0 was greater on D1 due to calibrationw here the light output has been estimated differently.
The image model is much more sensitive to these small changes because of the large ratio required to equalise the power density.
Why is no change in fluorescence detected with the image model for raw/norm when it was for the original approach? Also, why do you not see a decline on days 2 and 3?
The intensity difference across depth is much greater for the image based model which means that the heatmap will have a much larger intensity scale, which makes it harder to see small changes in profile.
Why in this example for image-based modelling is no increase evident?
The intensity difference across depth is much greater for the image based model which means that the heatmap will have a much larger intensity scale, which makes it harder to see small changes in profile.
Why does there seem to be more outliers for image-based modelling?
Because for image based modelling the ratio requried to increase the power density is much higher than other approaches which means the intesntiy used will be much greater.
This ended up introducing greater variability in the maximum fluorescence.
Additionally as this is characterised for each TF, there may be some variations in the length and active region which will mean that different regions will be supplied greater or lower power.
Overall, it shows limitations with the image model.
Why do you get significance at 0v galvo if you see no change in the plot (image/pharma/model)?
fasle pos
samp size
Since the model data for pharma actually showed signfiicance at lower galvos for image and original, would you not think this is the better profile?
It could be argued that this is better as it shows a signal that is without AF and reflects the depths in histology better.
BUT, significance was alot poorer and the increase was not as strong as with InVNorm.
For 5xfad- recordings, why is there some fluctations in signals across days - as seen in the heatmaps?
AF fluctuations - system, displacement of TF in brain, blood
bleach of AF throughout recordings
laser - some noise
Why is the fluorescence values across 5xfad- experiments in all analytical approaches higher than 5xfad+ recordings?
For Raw - the changes in signal intensity with M04 are very low so there is not much noticible difference.
Norm/Model - across days you can see no change in - but an increase in +. But sometimes the baseline signal is greater in FAD- which may be due to
???
Why do you not see the decline in day 0 signals in 5xfad- mice?
Potentially this rapid recline is due to plaques as they emit AF in this wavelength.
Did you detect any increase in fluorescence adter injection in 5xfad- mice? Why?
No we didnt, hich suggests that M04 is being cleared out at a rapid rate that will mean tat M04 concentration within the brain will not reach a state which is sensitive enough for the system to detect.
Why in 5xfad-, norm, in vitro on the example heatmaps are there some depth-resolved increases on day2/3?
Most likely AF increases
Why are the signals on day 0 for the image-based not saturated for the other examples?
It is a different recording and so the calibration for these recording swere probably more consistent or did not need redone over the days of the recordring.
For image based, why are there several outlioaers which are really strong in power that are at higher galvos?
The image model works to equalise power density which often results in a really large powers. This could have resulted in a great power.
What have the previous re-dosing studies shown?
They show that injecting M04 24-h before each imaging session results in the appropriate staining.
Why are you suggesting norm, in vitro is better?
Because the signal intesntiy is stronger and the rise in fluorescence is seen quicker.
Additionally, the depth resolution is more evident with stronger significance.
Could you not argye that model is better as it is still showing the significance but most prominently at the galvos that actually match up with the plaque pathology?
The model has some cases of showing specific increases espceically as the significance matches with where the plaques are, but the significance is alot poorer.
Why do you think you can see the depth-resolved changes using the original light protocol for the re-dose experiments but not pharmcokinetic?
RD wont record until 24-h after. This means the M04 can reach its maximum concentration and signal without being bleached.
PK is completed in real time and so the M04 is being bleached as its entering the breain which will limiy yher maximul concentration reached.
Why do you think you get stronger signals/significance for re-dose experiments?
Because the injection was completed and no recording was completed until 24-h. This gives time for max fluorescence to be reached.
Whereas for PK experiments the injection is completed throughout the recording where there will be bleaching at the same time and therefore it may not reach the same maximimal concentration.
There seems to be some decline aross the recoridng, does the signal deteriotatee?
Yeah there may be some bleaching across the 2-hour recording.
This will represent some bleaching of M04, which is consistent with the PK findings.
However, this was boosted back to a similar baseline level with injection and no significance decreases were found.
According to PK recordings, 2-h causes bleaching…why would you not see this in RD?
We do, which is noted in the results, the declines from 0-60mins.
This is due to bleaching of M04, but it doesnt cause a significanct decrease and so the fluorescence is reasonably consistent, but most likley a longer recording would diminish signals.
Why does the signal appear to decline acoss days in this RD example?
How will M04 affect the growth of plaques? Could this redose limit the progression?
If m04 was continuously re-dosed, eventually it will begin to bind to the end of the fibrils and block further growth.
Additionally, all binding sites could be taken.
This could limit the growth, which was shown in a previous study after months of administration.
According to the histology the greater plaque load is more around the middle of the fibre – why would you describe the greater signal to be around 0-1V galvo?
- vary across mice due to differences in implant site, some may have greater at these values.
- inaccuraces identifying the end of the TF track and resulting histological quantification
- due to differences in the light profile when in brain tissue due to light scattering.
If this signal decreases across 1-h, how can longitudinal recordings be completed? Also, how could you minimise this?
As plaque dynamics are slow, a decreased sampling interval could be used.
Additionally, the full illumination protocol could be reduced to a single power and light protocol which will limit the bleaching of M04.
Why would you see the significant differernces from 60-mins on day 0 but not 0-mins?
Potentially as seen in the PK experiments we seen a decline in F over the D0 which we assume to be due to the AF of plaques.
This could mean the F signal is lower at the middle of d0 recording, causing a greater increase in signal than 0-nin.
This may be different for image-based protocols just due to variations in signal collection - the way I sampled the data can contribute to this as I did not time bin to remove noise.
Why do you think you no longer have the significant decline at higher galvos in RD?
Because this was after the PK recording when it had went through 2 days of 5-h recordings and 2-days of 2-h recordings so there was much more bleaching of surrounding tissue.
Why do you think you never see the signal in raw data?
Because the raw data includes all autofluorescence.
Additionally, we cannot compare against a baseline well enough to acquire the depth-resolution.
Both of this hides the signal.
You get some really large increases with injection of m04, what may be causing these variations across mice?
Age variations - greater plaque load
Female mice - they have greater laque pathology - in HIP
Metabolism of M04
injection success
Why do you think in this example, depth resolution is not identificable in the model image-based data?
As the image protocol is equalising the power density - this means a greater power is likely to be supplied to greater regiosn of the TF.
For this, a baseline normalisation is important to acquire depth resolution which may be why we fail to see such changes for the model data with this approach.
What are your reasons for concluding the in vitro, norm is ther best?
What are your reasons against the others?
In Vitro Norm –> strongest signal increases, greatest significance from baseline, quickest detection of increase after injection
In Vitro Model –> ignificant increased, but weaker and weaker increase, significance not evidence across all time points
Im Norm –>weaker increase, weaker signifiacnce, more variability with signal intesnity
Im Model –> no increase or depth resoluton detected, more variability with signal intesntiy
Original/Raw –> no increase or depth resolution detected
Can you claim that the system maintained a fluorescence signal when yous ee these variations across days?
We see variations across days but we did not acquire a signifiacnt difference which suggests it is not extreme.
Additionally, once the redose was completed this boosted signals again.
Also, if a baseline signalw as completed with control mice too, this inconsistency can be controlled for. Potentially a role for the 550 nm signal.
How could you complete longitudinal recordings by accounting for these insablty recordings across days?
The laser would need to be driven alternatively - with a switch controller or external power supply.
Use the 550 or 440 nm signal (or both) for a normalisation and signal correction protocol, as others do with an isosbestic signal.
Why is there a decline in fluorescence across days in FAD- RD? but not PK?
Potentially bleaching was completed in Day 0 but not seen,.
Why would we have seen signfiicance at 1v in fad-, image, rd?
Again, there is a trend for decreasing AF but the distribution and sample size may not be strong enough in this case.
Why might the system not show the varying fluorescence across cortical layers mirroring histological sections?
- quantifying contralateral hemisphere so plaque pathology will be varied slightly
- sensitivity of the system
- inaccuracies of labelling TF track.
Why may this depth shift in histolofical corrleation occur?
- inaccuracies marking TF track
- quantifying contralaeal side
- shrinkage in histology
- differences in scattering of lught in the brain
Why would the raw data be negatively correlated?
BC there is no depth resolution of the raw signals.
As plaque is at deepest region this causes neg correlation.
Why are the correlation inconsistent across mice?
- inaccuracies marking TF track
- quantifying contralaeal side
- shrinkage in histology
- differences in scattering of lught in the brain
Without this correlation, how can you be sure these signals are actually plaques?
Visually the light increases in regions of plaques.
The increases were only seen in FAD+ not FAD-.
Consistent actoss light protocols, and several mice.
Why do you have some fad- with slight corr?
The analytical approach where we are adding noise to the plaques for correlation. As FAD- photometry profile is higher at the base of fibre this can introduce fasle poaitives/negatives.
What is your cut off for saying something is positively correlated as it swings between genotypes?
Around 0.3+ for positively correlated
Above 0.5 for strongly.
Why is there this bimodal approach with the correaltion in some caseS?
Some may be well aligned due to
- consistent plaque in other hemisephere
- good alignement of sections so same depth of subicukum is quantified
- accurate estimation of TF
Some may not due to
- inconsistencies in TF track, quantification and alignement of sections meaning the plaque dense region is shifted.
In fad- why is the profile higher at higher fgalvo and decrease with depth - opposite to flat fibre?
In this particular example, the raw F is greater at higher galvo which repersents the compensatory mechanisms of InV and Im.
For Norm, Model, its lower at higher galvos as it may show some decline from baseline due to bleaching.
Why does model seem to provide the best corr/alignment?
AF removal improtant for correaltion
While Norm removes some, it will not account for really sensitive changes in AF.
This may suggest for coreraltion modelled data may be better.
Why stronger significance for image and original corr but not in vitro?
Image and Original will have a greater variation of power along the TF shaft which will mean that there is greater likelyhood of negative correlation as it will have increased the baseline fluorescence at that depth which wont directly reflect the signals –> in raw as not normalised.
Whereas, inV has equalised power so these overestimations in depth is not seen so this doesnt effect the raw signal as much.
Why did you compare the corr for each protocol in one graph? What is your conclusion from this?
We wanted to determine if there was an analytical approach that provided a stronger correlation compared to others to see if there was an optimal appraoch for correlating data.
However, none proved to have a signidicanctly stronger correlation between norm and model. Raw illustrates as not suitable due to these power baseline inconsistensies.
Why did you not do group analysis for APK data?
Because we only completed a couple of recordings and the results were the same.
It was clear that we were not acquiring any increase in appropriate signals and therfore, it seemed suitable to only show an example of what this profile looked liek for +/- mice.
How is the AF forming ?
The component inside the patch cables that contributes to autofluorescence has not yet been identified.
However, this is bleached and provides energy. It is likely this undergoes some conformational change abck to its original state.
What have thorlabs shown in terms of low-AF patch cables?
That they recovery AF at a much slower rate over several days instead of hours.
Why may the AF recovery be faster on our recordings compared to what thorlabs show?
Because we are not using a low AF cable.
We use 405 nm which is highly AF and absorbable.
TFs can provide strong AF.
How does M04 clear from the non plaque brains? How has previous studies shown this?
The complete mechanism has not been shown but it was described by completing radioactive injection of M04 into the brain and taking plasma samples at 2-mins and 60-mins to identify the clearance.
They found it to have a half life of around 45 mins.
Why have TF been shown to have greater AF - Bianco
Although the polymeric cladding in 0.39 NA fibers is not present along the taper [10], during the heat-and-pull the polymer can likely lead to small infiltrations within the core, generating spurious bulk photoactive compounds in the taper volume.
polymeric claddings strongly increase AF signal of TFs
Why would the AF be greater at higher GVs?
Greater TIR which means the light is hitting off the cladding at a greater rate and therefore increasing the energy transfer and generation of AF.
What did montinaro 2021 show about optical properties in the brain?
The cytoarchitecture of different brain regions is crucial in defining the shape and the size of the light collection volume and, therefore, the type and number of cells contributing to the recorded signal.
Indeed, in M1, S1 and HP the thickness of the different layers set constrains or favors the ability to mine signal below a depth of 300 µm (Fig. 3(A)-(C), Fig. 4). On the contrary, structures with more uniform cell distribution at the millimeters scale, which likely results in uniform scattering parameters
it emphasizes how the shape and the size of the collection volume is defined by the combination of: (i) the influence of the anatomy of the brain region of interest on the photometry field and (ii) the fluorescence distribution across the cell type of interest.
In AD models, the presence of amyloid-beta (Aβ) plaques modifies photons propagation beside conventional tissue scattering, as the plaques are associated to higher inhomogeneity of refractive index, higher scattering coefficient, higher birefringence and higher anisotropy of scattering [34–36]. In addition, AD is known to alter connectivity, cellular density and the volumes of specific brain regions
How does the 30-mins compare to previous findings? What does this mea?
This is variation of studies that have monitored the pharmacokinetics of drugs after ip injection through various approaches like 2P, plasma samples and spectrometry.
They tend to see an average of drugs reaching the brain at 30-mins.
Why is there the changes in max fluorescence compared to other approaches?
Landmark - 24-h
Condello - 48-h
Friesen - 3-h
Because we are bleaching and because our imaging approach is much more seneitive than the other approaches which allows the signals to be detected quicker.
Why do the changes in M04 occur - do other studies monitro the stable nature - what do they show and how?
Condello and Lio do longitudinal recordings of M04 over 90-days with no reinjection and find that the probe remains bound stably with maintained fluorescence.
This suggest the changes we see are due to bleaching.
What were the main issues with the laser and how could you fix these? How did they affect the data?
Low/high freq noise
- implement a power supply for the laser to reduce noise from the laser
- a feedback loop to adjust the power in real time
Affected the data as iy introduced instability across days and the requirement to recalibrate which sometimes resulted in different powers being applied across days and therefroe messes up the baseline normalisation.
Could this increase in laser power be causing these increAses?
No, we moniored the laser throughout recordings and changes were not substantial.
Also, we account for these changes by normalising against day 0 which will contain the same alterations.
Also, both 5xfad+/- will have experinced these changes and we still see a sig diff.
What do you think about plaque pathology being modulated by TF and microglia?
The immune response is signfiicanctly reduced for microglia and astrocytes compared to FF as shown by Pisanello and colleagues.
This meant there was less clustering around the TF.
In regions where the TF was off ratget, the microglia response did not result in plaque accumulating as we did not see plaques along the qhole TF shaft.
This suggests that while microglia may accumulate, it has not resulted in migration of plaques.
What are the limitations with the corr analysis? What are your suggestions on how to fix this?
TF track identifiacation
Light cone - try estimate the volume of light propagated by teh TF and quantify through that rather than a cylinder cone
Scatter - the light protocols were developed in solution not accounting for variations in scatter.
What are the limitations with the model? How could this be improved?
There is only 2-folds so teh model has not been trained and trsted alot
the data sample is very low due to a sampling intevral of 5 mins being seperated into two groups
A way to have a greater baseline recoridng
How can you compare data from the same depths?
Maximal signal and minimum signal
Aligning the brain regions signal increases rather than depths
Pros and Cons of light protocols
Original
Pros - no variability with models, easy to implement, still acquire signals with norm approaches
Cons - not consistent power across depth, bias for depth-resolved comparisons
InV
Pros - equal power allows reliable depth resolved comparisons, provides strongest signals, best depth resolution, some correlation, not easily vulnterable by system changes, doesnt alter F baseline across depth
Cons - scattering effect in brain may be different, under assumption that this equalises, slight variations across depth
Im
Pros - accounts for power density
Cons - vulnerable to daily cahnges, over estimation of higher powers, bleaching of signals, increases baseline signal
Pros and cons of analytical approaches
Raw
Pros -/
Cons - cannot see increase or depth resolution
Norm
Pros - greater increase, depth resolution, quickest M04 detection, good depth res-hist, compatible with all light protocols, easy to implement, reliable
Cons - includes some AF, not good for corr
Model
Pros - removes AF, provides depth resolbed increases, seems promising for corr
Cons - model not perfect, not compatible with original or im due to variataions in DP as not normalised, increases not as significant
correction by using laser stability measurement
This could potentially be done by scaling the laser stability from 0-1 and normalising against this.
Was out of the scope of this project. But yes, using 405 or 550 could be an option for further normalisaiton.
could tapered light profile differences mask differences in signal
In solution and simulations the light profile is shown to be a cone like structure with lobes which can therefore mean we may be monitoring more tissue with greater GVs.
However, in previous Pisanello studies and work they have done for us in AD brain tissue, they show that due to the scattering in the brain, the light doesnt spread in the same way and is restricred to just several hundred micromtres away.
