Presentation on theme: "Neuronal Reconstruction Workshop"— Presentation transcript:
1Neuronal Reconstruction Workshop Darren R. Myatt*,Slawomir J. Nasuto,Giorgio A. Ascoli.
2More Acknowledgements Thanks also go toTye HadlingtonNathan SkeneKerry Brown (GMU)Thanks specifically do not go to the Heathrow Airport Security team
3Requirements for this workshop Laptop running/emulating WindowsWINE should be ok, except for possibly the 3D displayA reasonable amount of RAM1 Gig recommended, although 512M will be OK – less is possible, but not greatA standard 3 button mouse/trackball with mouse wheelNot strictly necessary but strongly preferable – I have a few spares to hand outEither a working CD-ROM drive or USB port that will recognise a flash driveIf you have neither of these, then I will begin to suspect that you are in league with the Heathrow airport security team in making my life more difficult than it needs to be
4Workshop AimsProvide participants with direct experience of reconstructing neurons and the challenges involved in resolving ambiguitiesGive a tutorial with the freeware Neuromantic applicationSemi-manual reconstructionSemi-automatic reconstructionTo generate discussion about best practice for reconstructing dendritic treesConsistency remains a problemGather feedback and recommendations on improvement for the Neuromantic toolThe workshop length is not set in stone but will probably last for around two hoursThese differences also cause physiologically significant differences in behaviour when simulating.
5Why Reconstruct Neurons? Allows the validation and refinement of simulations of neuronal behaviourCompare between simulation (via NEURON or GENESIS) and electrophysiological testingGaining large enough populations of reconstructed neurons allows insight into the morphological variation observed in each class.Facilitates the identification of dendritic abnormalities associated with brain diseaseEpilepsy, Alzheimer’s disease, some forms of retardation etc.Compare statistical properties of trees between control and experimental conditions (via L-Measure, for example)Slap some neurobiotin in and, voila!Remember to mention correlation/causation.When using to test difference between groups, if the radius errors (or any errors) are systematic then true differences will still be found.
6Is it Live or is it Memorex? Two main options for reconstruction…Live imaging (NeuroLucida)Advantages: no real memory requirement, no discretisation in Z.Disadvantages: specimen degradation over time and Z drift on stageReconstruction from an image stackAdvantages: minimal specimen degradation and Z driftDisadvantages: can require large amounts of storage and Z values are usually discretised.A motorised stage is strongly preferred.
7Flavours of Reconstruction Reconstruction methods may be split into 4 (or possibly 5) broad classesManualSemi-manualSemi-automaticAutomaticSo automatic that you don’t even need to turn up to work any more
8Manual Reconstruction User has to do define every neurite compartment with very little or no assistanceIncredibly laborious and time consumingCamera LucidaPencil and paper tracing via a system of prisms (it still exists!)Neuron_MorphoFreeware plug-in for ImageJOriginal inspiration for NeuromanticMaybe NeuroLucida here too?
9Semi-manual Reconstruction Each segment is still added manually by the userApplication gives some assistance in some elements of the task to reduce effort e.g. auto focussing, useful visualisationNeuroLucida (without AutoNeuron), Neuromantic on manual modeGenerally considered to be the most accurate method of reconstruction, but still highly time consuming
10Semi-automatic Reconstruction Application requires constant user-interaction, but the application requires mainly topological information.Define beginning and end points of a dendrite, and the neurite is traced out automaticallyNeuronJFreeware plug-in for ImageJ (single image only)Derived from the robust LiveWire algorithmNeuromanticSemi-auto tracing is a 3D extension of the NeuronJ algorithm with post-processingAlso includes radius estimationCommercially, Imaris from Microbrightfield has some similar functionality.
11Automatic Reconstruction What everybody really wants…Current automatic techniques are generally limited to high quality microscopy data (e.g. confocal fluorescence)AutoNeuron for NeuroLucida, NeuronStudioNumerous skeletonisation techniques, and also the Rayburst algorithm.The outputs frequently require cleaning up to bring reconstruction accuracy up to the required standardNeuronStudio is freeware, but not currently available. Mount Sinai School of Medicine.
12Which flavour to choose? t(Automatic)+t(Clean Up)<t(Manual)?Realistically, the clean up time will always be non-zero, except in trivial casesWith noisy data, fully automatic reconstruction is unlikely to be possibleA good reconstruction application shouldmake it as easy as possible to spot errorshave good manual editing capabilities to facilitate clean upClean up also applies to semi-automatic technique.
13Issues with reconstruction Interuser/Intrauser variation…Different users on the same systemThe same user on different systemsEven the same user reconstructing the same neuron on the same system!Thin dendrites (relative to image resolution) are a particular problem, as errors in radius estimation can have a large impact on surface area and cross-sectional area.Increased automation should increase consistency, but accuracy may still be a problem.However, systematic errors are definitely preferable to large scale random errors
14Example from Jaeger, 2001These reconstructions were performed in NeuroLucida by experienced usersSurface area range shows over 20% variation, which has a lot of implications for behavioural simulationsand this is just variation over individual dendrites, not a whole dendritic tree!For this example, the dendrites were around 1 micron in diameter. PiLarge variation in branch points, also – Talk about branch order
15Pyramidal Neuron Example All 10 participants were complete novices at neuronal reconstructionInterquartile range of surface area shows around 15% variationInterquartile range of volume is around 30% variationIncludes thicker neurites as well as thinReconstructions were performed by novices – MSc, PhD students in our lab (Cybernetics)Interquartile range – after all, at least 2 of the participants never really got the hang of it.Consistency information was given.
16NeuromanticFreeware application for making 3D reconstructions of neurons from serial image stacksProgrammed in C++ BuilderCan function on any form of microscopy data from non-deconvolved widefield stacks upwards.Semi-manual tracingManually position new compartments, which may then be edited afterwards as necessarySemi-automatic tracingLonger neurite sections can be traced out automatically, and the radius is calculated at each pointThe neuron can also be visualised in 3D to help identify and correct errorsFree for research – hope to not only provide freeware, but also offer top quality reconstructionCan function with any type of microscopy, although I’ve mainly tested it on non-deconvolved TLB stacks.* Semi-manual tracing is what the basic Neurolucida package allows you to do.
17Overlaid Reconstruction Basic InterfaceMode ButtonsMode optionsOverlaid ReconstructionImage StackStack BarThis is a CA1 pyramidal neuron from a Sprague Dawley rat.Image Processing
18Installation Time! CD/Flash drive contains Neuromantic directoryStack containing basal tree of a pyramidal neuronSimply copy the Neuromantic directory onto your computer somewhere, and it should be fine (hopefully!)Copy the stack to a directory nearbyRun the Neuromantic executable V1.4.1 to make sure everything is workingI included the V1.4.0 executable as a back-up.I’ve also included a variety of neurons from NeuroMorpho as I had quite a bit of space free
19Getting StartedAn updated manual may be found in Manual.pdf in the Neuromantic directoryLoad in the stack by pressing F2 or File->Load Stack and selecting the first imageWait for a while under the stack loads (it’s 387 Megabytes in total with 86 images) – the status bar shows the current progressHalve stack size if you are forced to use virtual RAM otherwise (Options->Stack->Halve Stack Size)
20Stack NavigationMost functionality is always present on the mouse for speedDrag the stack around with the right buttonZoom in/out by rolling the mouse wheel (or -/+ keys for those without)Use the stack bar or hold down the middle mouse button and move vertically to scroll through the different images (z axis)Middle clicking the mouse button auto-focuses at that position (+/- 5 slices)Hold SHIFT while middle clicking to auto-focus over all imagesMention capability of reversing mouse wheel zoom.
21Semi-manual Reconstruction Each compartment is added by dragging a line from one edge of the dendrite to the other, thus providing an estimate of the radiusThe compartment added is of the type defined by the radio buttons in the Manual panel to the rightEvery time a new compartment is added its parent is set to the currently selected compartmentSo add a compartment, then auto-focus on the next position down the dendrite, then add the next etc.In order to create a branch point, select the desired compartment with a left mouse click, then carry on as before
22Selecting Compartments As you move the cursor towards the centre of a compartment it will change, indicating that you can manipulate that segmentLeft click a compartment to select itSHIFT whilst selecting to add to the current selectionCTRL whilst selecting to select an entire branchALT to select all the compartments of the same typeCTRL+I inverts the current selectionCTRL+D deselects all compartmentsUsing these controls it is possible to efficiently select any set of compartments, such as a subtree.There are also some specialised selection commands in the Edit->AutoSelect menu
23Editing CompartmentsSelected compartments can be dragged around in the x/y plane using the left mouse buttonThe Z value is altered by selecting a compartment, navigating to the new desired image slice, and then pressing CTRL+C (or Edit->Set Z To Current Slice)The radius of a compartment is altered by holding down CTRL, and dragging with the middle buttonPress DELETE to delete all selected compartment
24Semi-automatic Reconstruction Newly added to the applicationStill a bit of a Work In Progress, as it is not as intuitive as I would like yetEmploys an extension to 3D of the semi-automatic algorithm used in NeuronJIncludes estimate of dendritic radiusAdditional post-processing to improve accuracyThis is not the final interface for the automatic reconstruction.
25Semi-automatic Reconstruction Employs Steerable Gaussian Filters to perform the image processingEfficiently yields information on the position of neurites and flow direction from eigen analysis of the Hessian matrixThe standard deviation of the Gaussian determines the radius of the neurites detectedA graph search (via Djikstra’s algorithm) is then performed to calculate the optimal route via the defined cost functionThere are problems if the standard deviation is significantly too large/small
26Patchwork MethodPre-processing on the entire image stack is expensive in both time and space.For the basal stack used in this workshop, around 10Gigabytes of RAM would be requiredTherefore, to avoid this issue, only the necessary patches of the image are image processed and routed.
27ConclusionsDiscussed reconstruction in general and some of the challenges associated with itGiven participants experience of the Neuromantic application, in terms of both its semi-manual and semi-automatic capabilitiesI hope you have enjoyed yourselves!