Welcome to the Imaris 9.7 release notes. Please have a look to the overview of Imaris Release notes for information about prior released features and fixed bugs.
Version Date: February 9, 2021
Imaris 9.7 addresses two themes: Dynamics and Structure. New measurements and new plots provide useful information and analysis.
Imaris 9.7 facilitates the investigation of dynamics of individual cells by providing an easy mechanism to synchronize measurements from an entire population based on “events” independently defined for each cell. For example in a movie of dividing cells where divisions happen at different time points one can define an event for a cell when it undergoes cell division and then plot measurements synchronized by their time relative to the event.
Imaris 9.7 lets users address an important question: “Is there an interaction (spatial relationship) between spots and surfaces?” To answer this question Imaris computes the number of spots within a distance d of the surface as a function of the distance and compares the observed function with simulations of spots positioned under complete spatial randomness. Using this approach from spatial statistics users can study interactions between spots and surfaces.
Vantage Time Plots
For the study of dynamical phenomena we have added a Time plot to the Vantage module. It plots statistics values measured by surpass objects, i.e. spots, surfaces, cells, or filaments. The plot shows median values as well as 25 and 75 percent quantiles over time and it lets you easily set up a comparison between different surpass objects or between different classes in a single surpass object.
When objects are tracked Imaris automatically records “Time Since Track Start” statistics values. In the Vantage Time Plot you can choose this time as the x-axis to for example produce an MSD plot (mean squared displacement plot).
Other choices for the time axis come from the definition of events. For example when cells are tracked with the lineage tracking algorithm every cell division creates an event and this leads to time statistics called “Event Time Event = TrackSplit” which Imaris computes to enable plotting of statistics values around the time of cell division.
This plot shows the volume occupied by chromosomes before and after mitosis. To get a graph like this from many cells that divide at different time points it is essential to synchronize all measurements by their time relative to the event which is captured in the “Event Time Event=TrackSplit” statistics. Imaris 9.7 produces this measurement automatically.
Imaris 9.7 introduces the concept of “events” to allow alignment of time series data by “events” on the time axis. Consider as an example three cells in the same image undergoing mitosis at different times, as shown in figure 1c. The graph in Figure 1a shows the intensity of the DAPI channel of the three cells plotted against acquisition time. Figure 1b shows the same graphs after aligning them by the events indicated on the time axis of Figure 1a.
Figure 1. Intensity of DAPI in three cells undergoing mitosis. a) Intensity plot versus acquisition time. b) Intensity plot against “EventTime” after defining events at the peak of chromosome condensation for each cell individually. c) Time series with the cell events indicated in pink.
Event Tab in the Creation Wizard
Events can be defined in the creation wizard of Spots or Surfaces. To do so one has to first classify objects into different classes and then one can set up “ClassChange” events for objects within a track when they “switch” from one class to another. This approach may seem somewhat indirect but it comes with the benefit of unlocking all the possibilities of the classification tools for the definition of events.
The event setup dialog lets you add any number of events and for each event you can define the class-change you are interested in. For example you may have set up an “inside” and an “outside” class based on the shortest distance between spots and surfaces to label spots as “inside” when their distance to the surface is negative and “outside” otherwise. You can then set up an event to capture when a spot changes from “outside” to “inside” and another event for when it changes from “inside” to “outside”. Each event you define produces a corresponding event time statistics for all those objects that are in a track which has such an event. This statistics value then becomes available as an x-axis for the Vantage Time plots.
TrackSplit and TrackMerge Events
When objects are tracked in such a way that tracks divide or merge --- through lineage tracking or connected components tracking or maximum overlapping cells or manual tracking --- then the objects immediately before a division or immediately after a merge will get a TrackSplit or TrackMerge Event and for all objects within a track that has a division or a merge an event time “Event Time Event= TrackSplit” will be computed.
Manual Event Editing
Events can be manually edited after object creation in the “Edit Events” tab. To assign an event to an object, select the object and press “assign” on the event you want to assign to that object.
It is also possible in Imaris 9.7 to add “Image Events” to some time points of a time series image. These will also lead to event time statistics for Spots and Surfaces and Cells and Filaments so that again those event time statistics can be used as the time axis in Vantage Time.
To set up image events the time bar of Imaris has received an event toggle button that adds an event at the current time point if the current time point doesn’t already have an event. In case the current time point has an event the event toggle button removes the event from the current time point.
Existing image events are rendered on the time axis as yellow rectangles.
Event Time Statistics
When Events are defined for objects in a track Imaris 9.7 will automatically compute some new statistics for each object in the track.
Time since nearest previous Event. Before the first Event it has negative values.
Time to nearest next Event. After the last Event it has negative values.
EventTime is computed for the purpose of plotting statistics “around the event”. It’s definition is somewhat arbitrary but it often fulfills the purpose of plotting statistics “around the event” quite well.
Surfaces Overlap Statistics
To help study the relation between two sets of surfaces Imaris 9.7 calculates statistics values measuring surface-surface overlap.
Overlapped Volume to Surfaces Surfaces - The volume of the overlap region between this specific surface and all surfaces from the other Surface.
Overlapped Volume Ratio to Surfaces - The ratio of the Overlapped Volume to Surfaces and the Volume of the specific surface.
Vantage Spatial Statistics
Imaris 9.7 provides some useful tools to study the question: “Is there an interaction between spots and surfaces?” and the follow up question: “At what distances do we observe attraction/repulsion?”
The starting point of this analysis is the measurement of the cumulative number of spots within a distance d from the surface and its less reliable1 counterpart the number of spots at a distance d from the surface.
Cumulative Plot (within distance d)
Histogram Plot (at distance d)
Imaris 9.7 then simulates randomly positioned spots (the same number as observed) and generates a randomization envelope around the expected value of the simulations. This facilitates a useful comparison. Since the simulated spots are by construction neither attracted nor repelled from the surface they provide a useful null hypothesis and deviations from this null hypothesis indicate attraction or repulsion.
Cumulative Plot (within distance d)
Histogram Plot (at distance d)
Inference from the cumulative graphs
Where the observed cumulative function exceeds the simulations it indicates “attraction” or more plainly that the observed number of spots within the respective distance to the surface is greater than what is expected from random simulations. Where the observed cumulative function is lower than the simulations it indicates “repulsion” or more plainly that the observed number of spots within the respective distance to the surface is lower than what is expected from random simulations.
Inference from the histogram graph
In the histogram graph we can observe at which distances from the surface spots occur in greater or lesser numbers than expected from random positioning. In the example it is at distances up to 3.5 micrometers from the surface that we observe a greater number of spots than expected from random positioning.
The probability density graph
The probability density graph is similar to the histogram graph in that it represents that density of spots at a distance from the surface. This graph is more smooth than the histogram because it is constructed using a kernel density estimate. The kernel smoothing is also applied to both the observed data and the simulated data (following [Duranton 2005]). The kernel width is displayed in the legend of the graph.
The randomization envelope of the random spots is computed from 1000 simulations such that it encloses 98% of the simulations, 1% of the simulations were above the randomization envelope, and 1% below the randomization envelope. It is plausible to reject the hypothesis that observations are completely random when the observed curve lies outside of the randomization envelope.
Please take care to understand that the region for the simulation of random spots forms the basis for inference. It is deviations from complete spatial randomness in the simulation region that you can detect with the plots of Imaris9.7. Imaris9.7 provides 3 choices for the region within which random spots are simulated: Inside the surface, outside the surface or inside- and outside of the surface. Furthermore Imaris9.7 will automatically restrict the distance from the surface (outside and inside) within which spots are simulated to the distance that is displayed on the x-axis of each plot.
To visualize the region within which random spots are simulated you can use the Surface distance transform function (see below) to compute a distance transform from the surface and consider only distances in the range that is displayed on the x-axis.
The methods employed by Imaris 9.7 are based on well-established methods from the field of spatial statistics. In statistical terminology Imaris9.7 plots the “intensity” of the observed spots as a function of the shortest distance to the surface [Baddeley 2015 chapter 6.6]. In [Baddeley 2015 chapter 10] you can find a description of randomization envelopes and hypothesis tests. When building the methods for Imaris 9.7 the analysis of [Gomariz 2018] served as a model.
Image Intensity as a Function of Distance to Surface
Imaris 9.7 provides a plot showing the average image intensity of each image channel versus distance d from the surface. This plot lets you inspect whether the intensities in some channel change as a function of distance from the surface. Clearly we expect this for the channel from which the surface was computed. You may find that other channels also exhibit a change of intensity with distance.
Surfaces Distance Transform Native in Imaris
Via the “Mask Selection” or “Mask All” buttons on the Surfaces Edit dialog it is now possible to compute the Distance Transform. The result is a new floating point image with a channel that holds at each voxel the value for the shortest distance to the surface from the center of that voxel. The distance is computed inside and outside of the surfaces. The values of the Distance Transform inside the surfaces will be negative.
Unique New HostID Simplifies Workflow
The computation of an identifier for the computer on which a license is installed has been improved for Imaris9.7. Where previously users had to choose a HostId, Imaris now automatically computes a unique ID and no user choice is required.
Click here for instructions on how to update your license.
We want to thank Alvaro Gomariz and César Nombela-Arrieta for their support and for the discussions about spatial statistics and its use in Imaris.
[Baddeley 2012] Baddeley, Adrian, Ya-Mei Chang, Yong Song, and Rolf Turner. "Nonparametric estimation of the dependence of a spatial point process on spatial covariates." Statistics and its interface 5, no. 2 (2012): 221-236.
[Baddeley 2015] Baddeley, Adrian, Ege Rubak, and Rolf Turner. Spatial point patterns: methodology and applications with R. CRC press, 2015.
[Gomariz 2018] Gomariz, Alvaro, Patrick M. Helbling, Stephan Isringhausen, Ute Suessbier, Anton Becker, Andreas Boss, Takashi Nagasawa et al. "Quantitative spatial analysis of haematopoiesis-regulating stromal cells in the bone marrow microenvironment by 3D microscopy." Nature communications 9, no. 1 (2018): 1-15.
[Duranton 2005] Duranton, Gilles, and Henry G. Overman. "Testing for localization using micro-geographic data." The Review of Economic Studies 72, no. 4 (2005): 1077-1106.
[Ripley1981] Ripley, Brian D. Spatial statistics. Vol. 575. John Wiley & Sons, 2005.
84 Bugs Fixed in Imaris 9.7
Hard to determine which license is selected in Administrator - Replace QTableView with QTreeView in Imaris Administrator
AddFilaments() in Imaris 9.1.x works like SetFilament() previously, i.e. deletes the current Filament
OME Tiff files from Micromanager (1.4.23) does not get converted
Edit Surfaces step of creation wizard is non-functional
some .lsm files open with incorrect intensity values
2D tiles of ND files are not correctly read in but superimposed
Help for XTension Surfaces Split misleading
Formula % of data set colocalized wrong
Observed Folder Lists and Observed Folders can be lost when Imaris and ImarisViewer are opened at the same time
XT Classify Spine is broken if there are more than two separate dendrite
Statistics calculation happens too often while moving the frame