Tuesday, June 25, 2013

Video Tracking Extra: Spiders at the Oxford Silk Group

First field alpha trial of the HVS Image 2014 Alpha

Well they are not "exactly" radial arm, but to us equally fascinating. So, a great trip to Beth Mortimer at the Oxford Silk Group on Monday to help them track spiders on webs moving from the old 1980s VP112s that they have to the HVS Image 2014 Alpha. This was essentially a performance and optical trial but we got lock on even on the smallest arachnid even without auto-thresholding and background subtraction and about five times the industry standard frame rate actually exceeding the old big-iron hardware boxes. Infra-red pick up from an LED source (peak 810nm) was impressive on the VX-800 camera.

In case you are wondering we'd expect 12 months testing from a first field Alpha to the release version.

Followed by lunch at the Eagle & Child after ("pulled beef" pudding and Brakspears) and cookies at Ben's Cookies - both highly recommended.

Friday, June 21, 2013

Neuroscience Extra: UCL Neuroscience Symposium 21 June 2013

Just back from UCL Neuroscience 2013. Its not really  radial arm maze at all, but it is neuroscience and mainly behavioral so it is as good a place as any to note and discuss! These notes are not meant to be science journalism but just an experiment putting some part of my meeting/lab notes live to see if I get any interesting conversations going. If you are a grad student, presenting author or scientist feel free to mail me on this. You'll probably need the meeting notes to follow this.

  1. Initial notes related to application of our new analytical techniques to our scientists legacy data (Something like Have you run a big trial and it doesn't "quite" show what you think it should? Do you have any legacy/existing data that you feel *should* show effects but just doesn't? We've got a number of new analyses that I think should show much cleaner, more significant effects from than for example typical latencies, corridor tests, mean distance to platform etc  in MWM). I will contact my Linkedin MWM and our Darwin club about these so need not be detailed here (but you an email me). One interesting aside is that the new methods should be more suitable for superposition which is nice and elegant.
  2. Maria Fitzgerald's lecture on development of nociceptive circuits was of of important clinical and ethical significance. What I thought of though was a series of in vivo tests that you could do by direct stimulation of dorsal horn (with a phased transdermal array?). If we used a 16x16 dot array of gel electrodes on a flexible backplate and controlled the sinusoidal (or maybe trochoidal) inputs this might do the job. I guess we are supposed to avoid T10 (paraventricular nucleus of the hypothalamus and secondarily the supraoptic nucleus of the hypothalamus) or maybe not? Tell me!)  This fits it with some work here which I've nominally titled Complex Structure Elucidation in formerly stochastic processes but I'll need to expand on this elsewhere.
  3. Parashev Nachev from the Institute of Cognitive Neuroscience gave an excellent talk - of which I wrote down (or perhaps re-interpreted as Complexity looks like noise but it hides non-stochastic deep structure. Well he might not have actually said that - perhaps I just wrote down what I've been saying (and acting on for years) and was just re-iterating the deep structure that I thought was there. Too many Cambridge Balls this week I think in sleep deprivation. Good amusing comments on the H-Index (or what we call the Hirsch number) by John Hardy too. 
  4. Good informed discussion with  - discussed how I'd been creating data mining tool for my own research in the early days of PubMed (and as I remember ISI) - but now I used Google Scholar. I've looked at the PMC API now and I can see that in the first instance it looks far superior to Google as the tags are likely to be stable and not changed arbitrarily like Google.
  5. Had a good talk with Linda Partridge on behavioral assays of Drosophila and possible confounding factors and some observations on histology. Good talk at Marinella Cappelletti's poster on training and brain stimulus (I'm still interested to discover as to whether a dc offset would improve data). Very nice work by Debra Fine in her study of Increased Capacity of Time Perception in Autism Spectrum Disorder. Still interested in on population and effect distributions - I bet we could pull that out of the existing data sets ad tell an even better cleaner story without further experiments. I've got some related notes on configuration classes here too but I'm going to need to look at the literature to articulate them properly. Excellent and detailed discussion of virtual maze in the poster session with Michael Krumin and showed him some of our notes and plots on improving anaylsis of spatial learning. Interesting to see however that Probit analysis is currently not used where it could be most effective. Generally there is a lot of used of parametric statistics were non-parametric measure might be much strong doe to population distributions.
  6. Interesting discussions of O'Keefe's original quad electrode and microlithograpic electrodes. This is where my extra post-doc skills in micro-photolithography  and nano-lithography (include some prep work for the AFM) and EEG electrode balancing - and frankly basic undergrad physics texts that I read as a kid on  get me thinking of the type of noise balancing array you could build with nanolithograpy on carbon filaments. I'm not sure however that the current practice on electrode impedance makes sense to me if you really want to get clan, easy to interpret data - it looks as though there is a better way fairly clear.
  7. Made some really interesting observation of Mitochondrial VCP structure (and the translocase complex of the outer membrane) in Fernando Bartolome's talk which were fascinating to me but probably blatantly obvious to everyone else so I won't bore people by recounting my banal observations here.
  8. Kenneth Harris gave a great talk on the effects of locomotion  and the V1 response to position and speed look like keys to developing new spatial navigation tasks (both are needed to get a strong V1 response). I'm left wondering if the virtual air-ball maze V1 results replicate in Morris Water Maze? Hopefully they do in dry mazes and open field.  I thought of the V1 data as a Laplacian and a weighted area tracking algorithm came into my head at this stage and there are lots of notes of these.
  9. More great posters from Abigdail Andrews, Siri Ranlund, Padraig Gleeson, Cyrille Rossant. Must come back to jot down my notes on these later. 
  10. Really enjoyed Julija Krupic on spatially periodic cells in the parahippocampal region.  First thoughts were on possible validation of some new protocols (particularly relating to cross modal scaling of grid cells). I thought that having taken the  data into Fourier space (an excellent strategy) there was something very obvious and important that she missed on the interpretation of place filed FT vs grid filed FT - - something that brngs out a whole lot of extra understanding if you think it through - well if you are used to thinking in Fourier space - if I get time I'm definitely have to point it out in the meanwhile check here and see if you can see it. I'm sure John O'Keefe can!
  11. Schimdt-Hieber's Cellular mechanisms of spatial navigation (in stellate cells) presented a fascinating model in terms of contrasting Oscillatory Interference (re:theta waves) vs attractors. Unfortunately just as we got the really exciting results slide for  Oscillatory Interference (and I was literally leaning forward in my seat) - Christoph seemed to run out of time and skipped forward. Drat.

http://www.ucl.ac.uk/neuroscience/features/2013_ucl_neuroscience_symposium