Skip to content

Results

EVAnalyzer stores the results of each analysis in a DuckDB database file, in a folder of its own named by the start time and the job name:

<project folder>/results/<date>__<job_name>/<job_name>.evadb

The folder also holds <job_name>.evaproj, a copy of the exact settings that analysis used.

Results written by the current version use results format 2, which can hold profiles. Versions up to 1.1.0-alpha.3 can’t open them (“these results were saved by a newer version of EVAnalyzer … update EVAnalyzer to open them”). Older results open as before; when an analysis adds to an older results file, the file is upgraded to format 2.

The results hold the objects of the project classes whose Results setting is Objects (the default), with their measurements. Not stored are:

  • objects without a project class (Unset, or with their last class removed),
  • intermediate classes - an object with only intermediate classes is not stored, one that also has a stored class is stored without them; an intermediate class is not listed among the results’ classes,
  • parent and colocalization partner entries pointing to an object that is not stored.

Results are a mode of the main window: switch to Results in the bottom bar (Ctrl+3).

In Results, the top bar holds the results’ controls:

  • Analysis ▾ - the analysis shown, picked from every analysis of the project’s images (ANALYSES · N). With none open the chip says Pick an analysis; with none at all the menu says “None yet - Run writes one”. The same menu has Refresh, Show the output folder and, for the open analysis, Delete “name” - see Deleting an analysis.
  • The project’s views as tabs - each a List, Plate, Charts, Tracks, Paths or Trees - and + (New view) to add one.
  • On the right a Z/T stepper when the data has Z-stacks or time-lapse frames - it applies to whichever view is open -, Full | Split with image - the results alone, or beside the image (of the charts only for Profile: “Split view works with List, Plate, Tracks and the Profile chart” - another chart shows full width, and the split comes back with Profile) - and Export….

Each view has its settings in a panel on the left and its content beside it. The bottom bar shows the number of images (“81 images · 1 disabled”).

Results mode, List: the results bar on top, the Images, Classes and Columns lists and A row per on the left, the table of objects beside them

With no analysis open, Results shows the Analyses card: every analysis of the project with when it was written and its size. Click one to open it. Its ⋯ (or a right-click) has Open, Show in the folder (its own folder) and Delete from disk….

The Analyses card with the analysis dilution_series, when it was written and its size; the menu of one open: Open, Show in the folder, Delete from disk…

Delete from disk… in the Analyses card - or Delete “name” in the Analysis ▾ menu for the open one - deletes an analysis you no longer need, to free its disk space. The dialog Delete the analysis “name”? shows what goes:

  • the analysis’ folder, relative to the project (e.g. results/20261023T091000Z__serum/),
  • Results (the results database), Control images (N files) from Save Image steps, Settings of the run and Other files, each with its size,
  • Freed - the total.

The dialog Delete the analysis “dilution_series”?: its folder, Results 19.5 MB, Settings of the run 102 KB, Freed 19.6 MB, and the buttons Cancel and Delete 19.6 MB

“Its folder is deleted from the disk, with everything in it. This can’t be undone - export what you still need first.” The project, its images and its views stay. If the analysis is open, EVA closes it first.

Delete size (e.g. Delete 860 MB) deletes it; the list refreshes, and if the deleted analysis was open, Results goes back to the Analyses card. A file another program still has open can’t be deleted: the dialog says so (“Couldn’t delete 1 file - close the program using it and try again”) and its button becomes Try again. An analysis that is still running can’t be deleted - stop it first.

Only analysis folders in the project’s results folder can be deleted this way. Connected to a server, it deletes the analysis on the server.

A view is one way of looking at the results, with all of its settings: a List with its classes, columns and aggregations, a Plate with its value and colors, a chart with its axes, and whether the image is shown beside it. The views belong to the project, not to one analysis: every analysis of the project opens with the same views, so a view set up once - say “Spots per cell” or “QC” - shows the next analysis the same way.

A new project has four views: List, Plate, Charts and Tracks. The tab of a view says its kind (LIST, PLATE, CHART, TRACKS, PATHS, TREES) unless its name starts with it already. A Tracks view is dimmed while the analysis has no tracks.

  • Click a tab to show the view. The view shown saves itself: every change of its settings is written into the project (and the project has unsaved changes).
  • + opens NEW VIEW: a List, Plate, Charts or Tracks view with the default settings, a Paths view (only when the analysis has tracks) or a Trees view (only when it has traced neurons), or A copy of the view shown. A new view is named after its kind (“List 2”), a copy ”… copy”.
  • Double-click a tab to rename it in place - Enter takes the name, Esc keeps the old one.
  • Right-click a tab for its menu: SHOW AS List, Plate, Charts, Tracks, Paths or Trees (the view keeps the settings of each kind, for switching back), Rename…, Move left, Move right, Duplicate, Back to how it was saved and Delete (not the last view).
  • With more views than fit, the tabs scroll and All views lists them all (VIEWS · N).

Creating, renaming, moving and deleting views can be undone with Undo (Ctrl+Z).

Views name classes by their id and columns by their key, so they fit every analysis of the project. When a view names a column the analysis shown doesn’t have - for example a measurement of a step it didn’t run - a banner says so: “Not in these results, left out of the view: … - the view keeps them for other results.” The view keeps the column for the next analysis that has it.

The CLI can export a results database the way a List view shows it - see results export --view.

The results bar with the default views List, Plate, Charts and Tracks, a copied view renamed "Wells by count" with its PLATE badge, and the + menu open: NEW VIEW List, Plate, Charts, Tracks, Paths, Trees or a copy of the view shown

The List view shows one row per detected object. The default columns are Image, Object ID, Class (with the class’s color), Area [px], Area [nm²] and Circularity; more are under Columns:

  • General - Object ID, Track ID (the track an object belongs to, set by a tracking step), Image, Class, Count.
  • Geometry and Shape - area, perimeter, circularity, solidity, eccentricity, Max Feret [px], Min Feret [px], Major Axis [px], Minor Axis [px], Equivalent Diameter [px] and Extent.
  • Coloc - which partner classes from Colocalize Objects an object has - per class and per combination of classes, see Coloc columns.
  • Intensity - per channel Avg, Sum, Min, Max, Std Dev, Median, Lower Quartile, Upper Quartile and MAD Intensity (Ch n).
  • Measurements NEW - the values of the pipeline’s measurement steps, named step ID · measurement (e.g. coloc_01 · Pearson). A measurement of the whole image shows the image’s value in every row of that image. A measurement that points to another object - the nearest neighbour, the parent - has a second column … · object; click its cell to see that object.

The settings panel has, under SHOW:

  • Images - All images, Only enabled or Only disabled (the images left out).
  • Images and Classes - check lists with All / None and a search; click the title to fold or unfold them.
  • Columns - Find a column searches all of them; below it the columns in foldable sections - GENERAL, GEOMETRY, SHAPE, the intensities, the measurements, … -, each titled with how many are ticked (“SHAPE · 1 of 12”). A section with a ticked column starts unfolded, the others folded; while you search, every section is open. The coloc columns have a box of their own (Coloc columns), and the CHILDREN and LINKED OBJECTS lists of Parents and children come after it.

Above the table, N rows and the page; use the arrows to page through the results rather than scrolling through everything at once.

Click a row to select it; it stays highlighted. Show on image ↗ - or a double-click - opens the object on its image: the image opens in the editor with the object marked, and a bar over it, RESULTS · Row 12 of 200 ‹ › Back to results, steps to the previous or next row’s object or takes you back.

After Show on image ↗: the image with the object marked and the bar "RESULTS · Row 12 of 200 ‹ › Back to results ✕" at the top

A row per (under ROWS) switches the table between Objects (flat, one row per object), Images (one row per image/class combination) and - when the analysis has colocalization partners - Coloc pairs. With Images, Aggregate picks one or more of Average, Minimum, Maximum, Std Dev, Sum, Median and Skewness. Every numeric column is duplicated per selected aggregation (for example Area (px²) [avg] and Area (px²) [sum]), so you can compare, say, average object size against total covered area per image.

A row per: Images, Aggregate: Average and Sum - columns doubled per aggregation

Transpose places the classes next to each other instead of below each other - handy when you want to compare classes per image, or paste the table into a spreadsheet that expects one column block per class.

  • With A row per: Images, the table has one row per image, and for every class one block of columns (one column per measurement and aggregation).
  • With A row per: Objects, the n-th object of each class in an image shares row n, with one block of columns per class. A class with fewer objects in that image leaves its block empty in the extra rows. The image name is the first column, the class is given by the block.

The Classes list decides which classes become column blocks.

Transpose on, a row per image: one column block per class

When the analysis has Colocalize Objects partners, COLUMNS has a COLOC box. Its columns tell for each object which of the partner classes it colocalizes with:

ColumnAn object counts when it has
Coloc with Ba partner of class B (others may be there too).
Coloc with at least B, Ca partner of class B and one of class C (others may be there too).
Coloc with exactly B, Cpartners of class B and C and of no other class picked under With.
Coloc with no classno partner of a class picked under With.

The box holds:

  • With - the partner classes the columns are made of. Default: all of them. A partner of a class not picked doesn’t count for exactly and no class.
  • Coloc counts - what a cell holds: Objects (1 for an object that colocalizes, else 0 - with A row per: Images the Sum is how many objects, the Average their share) or Partners (the number of partner objects of those classes, 0 without a match). Coloc with no class is only offered with Objects. The header is the same either way.
  • the columns in three foldable lists, WITH EACH CLASS, WITH AT LEAST and WITH EXACTLY (with Coloc with no class), titled like the sections (“WITH AT LEAST · 1 of 3”).
  • Only combinations that occur (on by default) - offers only the combinations some object has. Off, every combination of the classes under With is offered; with more than 8 classes only those that occur.

Example: with the partner classes cy5@spot, cy7@spot and gfp@spot and a Cell row per image, Coloc with exactly cy5@spot, cy7@spot with Objects and Aggregate: Sum counts the cells that have a cy5 and a cy7 spot but no gfp spot.

Coloc counts, With and Only combinations that occur are one setting for the List, the Plate view and the Charts: the Plate and Charts settings show Coloc counts too, and their property pickers - and the export’s Measurements - offer the same coloc columns.

A row per: Coloc pairs gives every object one row per partner it colocalizes with: the object’s own columns, then one set of measurement columns per partner class. A row built from a partner of one class shows a dash in the columns of every other partner class - those partners get rows of their own. Rows of the same object share a shading.

Coloc pairs A row per: Coloc pairs - each object once per partner, partner columns per class, dashes in the other classes' columns. src/assets/screenshots/screenshot-results-coloc-details-view.png

When a pipeline links objects with Relate Objects - spots to the cell they lie in, say - each child stores one parent per parent class. COLUMNS then has two more lists, after the coloc box:

  • CHILDREN - one entry per link, “Spot in Cell”. Ticked, every parent row gets the children’s Count (Spot · Count) and, per aggregation picked under Of the columns (the same choice as Aggregate: Average, Minimum, Maximum, Std Dev, Sum, Median, Skewness), that aggregate of every ticked numeric column over its children (Spot · Mean · Area [px]). The default aggregation is the average.
  • LINKED OBJECTS - one entry per parent class, “Parent · Cell”. Ticked, every child row gets its parent’s ID (Parent Cell) and its parent’s value of every ticked numeric column (Parent Cell · Area [px]).

The counts and IDs are links: click a children’s count to list just those children, a Coloc with … count of one class (with Coloc counts: Partners) to list the partners, a parent’s ID to show the parent. A bar above the table names what is shown (“Spot of Cell 3f9a0c21d4e7”); ← Back goes back to the full list.

Parents and children List view with CHILDREN "Spot in Cell" ticked and Of the columns: Average - the cell rows with Spot · Count and Spot · Mean columns, one count cell hovered as a link. src/assets/screenshots/screenshot-results-children.png

Hover a row to see its eye: Leave this image out of the results (or Take this image back into the results). It applies to the row’s whole image, see Leaving images out. The row of an image left out carries a badge - Left out when you left it out, Excluded when an Exclude Image step of the pipeline did; hovering Excluded shows the step’s reason.

To leave several images out at once, tick their rows (the box at the start of each row, Shift-click for a range, the box in the head for the whole page). The bar above the table counts them (“3 images ticked · 12 rows”) and offers Take back in, Leave out and Clear.

The Charts view plots the results instead of listing them. Pick the chart as a tile: Histogram, Scatter, Boxplot, Over time or - when the analysis has profile measurements - Profile.

  • Histogram bins a chosen numeric property, optionally restricted to one class, and shows the object count and the distribution across bins. Under Display: the number of Bins and Counts on a log scale.
  • Scatter plots two numeric columns (X and Y), optionally restricted to one class. Large datasets are downsampled for drawing - Points drawn sets how many (500, 2 000, 10 000 or All; “2000 of 41227 objects plotted”). Click a point to see its object on its image.
  • Boxplot draws one box (quartile box, median line, whiskers, and outlier dots) for a chosen property. A box per groups them by class, image, well, plate row or plate column. Click a box to see its objects in the List.
  • Over time NEW follows a property through the frames of a time-lapse. A line per Class, Image, Well or Track (tracked objects); each line’s spread is drawn as a band - under DISPLAY Mean ± SD or Median + IQR. Relative to the first time point shows the change instead of the value, Individual tracks, faint draws every track behind the lines. For results with divisions or fusions, History picks Own track (default: each track its own line) or With parents’ (a line per cell from its first ancestor: a daughter’s line starts with its mother’s track) - with A line per Track or individual tracks. Export data… above the chart writes the lines’ values per time point as a table.

The Over time chart: the mean area of gfp@vesicle per time point with its standard deviation as a band, every track faint behind it

Charts: results histogram

Charts: results scatter

Charts: results boxplot

NEW A profile is a measurement with several values per object, each at a position - for example the number of intersections per radius of Measure Object Sholl. The value pickers of the List, Plate and the other charts leave profiles out; the Profile chart draws them.

  • Profile picks the profile measurement, when there is more than one.
  • A line per Class, Image, Well or Plate; each line’s spread is drawn as a band - Summary Mean ± SD or Median + IQR. An object without a value at a position its line has counts as 0 there (it doesn’t reach that radius) - for the median and quartiles too.
  • The selected object’s curve - select an object in the List or in the split view to draw its own profile on top. Mark its highest value marks where it peaks, and the cards below the chart show SELECTED · MAX (the highest value and its position), REACHES (its last position) and MEAN UP TO THERE.
  • Export data… above the chart opens the export with the profile picked.

With Split with image, the Profile chart has the image beside it. Click an object on the image - or use ‹ › (or ↑ / ↓) - to see its curve and its circles: ‹ › step through the objects with the profile on all images, the image’s head showing which one (n / total). For Measure Object Sholl the circles are drawn on the image around the object it was measured around (its Centre From parent, e.g. the soma), else around the skeleton itself; the circle of the highest value is drawn stronger.

Charts › Profile: the Sholl intersections per radius of the skeletons, mean ± SD as a line with its band; the selected skeleton's curve on top, its highest value marked, and the cards SELECTED · MAX, REACHES and MEAN UP TO THERE below

Plate lays results out as a physical plate/well grid instead of a table or chart - useful for spotting spatial patterns across a multi-well high-content screening plate. It drills down through three levels: Plate → Well → Image.

The settings panel holds the same settings on all three levels: Class, Property (the metric to color by), Aggregate, Colors (Excel, Viridis, Plasma, Inferno, Cividis, Coolwarm, Red-Blue, YlGnBu, Haline, Algae, or Thermal) and Group by. The legend shows the active color range - click it to switch between Auto and a Manual min/max.

Click a tile to select it: the card on the right shows its value - and for a well, its images with their own values. Open well › / Open image › drill in, or simply double-click the tile. Inside a well or an image, the plate’s head shows where you are - ‹ Plate › Well C3 › image - click a part to go back there, or ‹ to go one level up.

Group by decides how images are assigned to wells:

ModeHow the well is found
AutoThe well is detected from the image name, e.g. A01_01.tif or Exp_B3_s2.tif.
Custom regexYour own regular expression on the image name. Group 1 is the well (B03), group 2 the plate row (B), group 3 the plate column (03), group 4 the image number inside the well (01) - e.g. ^(([A-H])([0-9]+))_([0-9]+).
FolderOne well per image folder, placed on the plate when the folder is named like a well (B03/image_01.tif).

Regex in use shows the regex - for Auto and Folder read-only, so you can see what was detected.

Grouping, plate size and well layout are project settings: they are stored in the project file, are the same values as in EVA › Project settings… › Plate & grouping and the Images panel’s Plate, and are used by the plate and well export. Changing them in one place changes them everywhere.

Each tile is one well, colored by the aggregated metric across everything grouped into it. Wells are placed by decoding their group label into a row/column coordinate (e.g. A14). Pick a Plate size - Auto (the smallest standard layout that fits your data) or a fixed layout from 1 to 3456 wells (e.g. 96 Well (8 x 12), 384 Well (16 x 24)).

A 96-well plate heatmap, one well selected - its card on the right with the value and the well's images

Drill into a well to see its individual images (fields) laid out as their own grid. Well layout sets that grid:

  • Auto - the smallest square-ish grid that holds the highest image number in the well, images placed in number order.
  • Rows × Columns - a fixed grid, with a free choice of which image goes where.

Rearranging images by drag and drop. If the images were not acquired in the order the grid shows them (e.g. a meander scan), drag an image tile onto another position - the two images swap places. The new order is stored in the project’s plate settings and applies to every well of the plate. If the layout was Auto, the grid shown becomes a fixed grid of the same size first. The same order can also be edited number by number under Images in well in Settings.

Inside a well: the images as a grid, the breadcrumb "Plate › Well C3" in the top bar, one image left out (crossed out)

An image can be excluded from the statistics, for example because it is out of focus or contains an artifact. Click its eye in the well’s card, select its tile inside the well and click Leave this image out of the results, or use the eye of its row in the List; Use this image in the results brings it back.

An Exclude Image step in the pipeline does the same during the analysis: the images it rejects are stored as left out, with its reason, and can be taken back in by hand like any other.

An image left out:

  • is left out of every aggregated value - the well’s value in the plate view, and the plate/well exports;
  • is shown crossed out in the well view, and the well containing it gets a small corner badge in the plate view;
  • is shown struck through and dimmed in the List view, with a Left out or Excluded badge - and listed alone with Images: Only disabled;
  • is skipped by exports that cover all images. If you select it explicitly in the export’s images, it is exported anyway.

The setting is stored in the results database, so it is kept when the results are opened again.

An image whose analysis stopped with an error only holds the objects found before the error. Such an image is marked with a warning badge in the plate and well views (and in its well), drawn in the warning color in the List view, and - like an image left out - left out of aggregated well values.

If the whole analysis did not finish (cancelled, failed or interrupted, e.g. by a crash), Results shows a warning banner at the top: the results are incomplete.

The innermost level bins a single image’s objects into square tiles - the Square size (36, 48, 64, 128, 256, or 1024 px) - colored by the same metric and aggregate as the levels above. Click a tile to select it; Show on image ↗ shows that region of the image in the editor, marked with a rectangle.

One image binned into tiles, one tile selected with Show on image ↗ in its card

When an analysis has tracked objects - from a tracking step such as Track LAP, Track Kalman, Track Overlap or Track Nearest Neighbor - the results bar offers a fourth view, Tracks: one row per track instead of one per object.

One row per track. The table’s columns:

ColumnWhat it is
TrackThe track’s number (#12), with a dot in its track color - the color of its trail on the image when trails are colored by track.
ClassThe class of the tracked objects.
ImageThe image the track is on.
Time pointsThe first and last time point and, in brackets, at how many time points the track has an object: 3–41 (37) - fewer than the span when it has gaps.
the motion columnsThe values of Measure Track Motion, see below - by default Moved, VCL, LIN, Turning, Max distance and MSD alpha.
Area pxThe track’s mean object area over its time points, in pixels.
Area over timeA small curve of the object’s area from the first to the last time point, scaled to the track’s own smallest and largest area - a cell that grows, divides or shrinks shows at a glance.

Click Track, Time points or a motion column’s head to sort by it - the time points and values largest first, tracks without a value last; Track sorts by image and track number.

NEW Columns picks the motion columns (6 of 16), any of the 16 values of the Measure Track Motion step after the tracking. Without that step they show –. The columns picked are kept with the view; with many of them the table scrolls sideways and the track’s details stay in view. Each head names its unit, e.g. VCL µm/min:

ColumnUnitWhat it is
Path lengthlengthThe length of the whole path: every step from time point to time point added up.
MovedlengthThe displacement: the straight distance from the first to the last position.
Max distancelengthThe farthest the track ever got from its first position - high for a track that went out and came back, unlike Moved.
VCLspeedCurvilinear speed: the path length over the track’s duration. This is the track’s Speed.
VSLspeedStraight-line speed: Moved over the duration.
VAPspeedAverage-path speed: the length of the smoothed path (each position averaged with its neighbours) over the duration.
LIN-Linearity, VSL / VCL: 1 a straight line, near 0 an object that wandered around. This is the track’s Straightness.
STR-Straightness of the smoothed path, VSL / VAP.
WOB-Wobble, VAP / VCL: how much the path swings around its smoothed course (1: not at all).
ALHlengthLateral head amplitude: twice the mean distance of the positions from the smoothed path - how wide the track zig-zags.
BCF/timeBeat-cross frequency: how often the track crosses its smoothed path, per time.
Turning°The mean turning angle from one step to the next, + counter-clockwise: near 0 for random turns, one sign for a track circling one way.
Turning abs°The mean turning angle without its sign (0-180°): how much the track turns, whichever way.
Angular speed°/timeThe turning angles added up, per time.
MSD alpha-The exponent of the mean squared displacement over the lag: about 1 a random walk, 2 a straight line, below 1 a track that stays in a bounded area.
MSD Dlength²/timeThe coefficient of the mean squared displacement - how fast the track spreads out.

Lengths (Moved, Path length, Max distance, ALH) are in µm when every image listed has a pixel size and a frame interval, otherwise in px. Speed unit picks px/frame, µm/min or µm/s for the speeds, and the time of BCF, Angular speed and MSD D (per frame, minute or second; MSD D in px² with px/frame, else µm²). Without a pixel size and frame interval, the table stays in px and px/frame. Measure Track Motion leaves out a value a track can’t have - with too few positions for the MSD, say - and the cell shows –. How each value is computed is on the Measure Track Motion page.

The bar above the table also sums up what is listed: “412 tracks · 9864 objects · median 31 time points”.

Above the table, Class and Image pick which tracks to list. Filters opens the other filters (SHOW ONLY TRACKS); its badge counts the filters changed:

  • At least N time points - leave out short tracks.
  • From the first time point - only tracks that are there from the start.
  • Not touching the edge - leave out tracks that touch the image’s edge, which may have left or entered the field of view.
  • Reset filters - back to all tracks.

With Track LAP, Track Kalman or Track Overlap and merges or splits switched on, tracks have a lineage: a division ends a track and starts two, a fusion ends two and starts one. The summary then also counts them (”· 3 divisions · 1 fusion”), and a selected track’s card names where it came From and what it goes Into (“#3 (division)”, “#7 (fusion)”). On the image, a line joins a track to its parent’s last point, with a ring where the division or fusion happened; a selected track is drawn together with its parents and children.

Click a track to see it on the right: its numbers - Time points, Moved, Speed, Straightness and, with Measure Track Motion, Max distance, MSD α and Turning -, its Path from its start (drawn from the origin, with the drift the step corrected taken away), with Measure Track Motion its MSD over lag (log–log), its Area over time, and The track’s time points with the position (X px, Y px) and area at each. Show in image opens the split layout at that track.

Results, Tracks: Class, Image, Filters and Columns on top, a row per track sorted by VCL with its time points and motion columns; the selected track's numbers, path from its start, MSD over lag, area over time and time points on the right

With Split with image, the tracks of one image are listed beside it, and the image draws each track as a trail through its time points. The table there is shorter: Track, Time points, Speed (VCL, in the speed unit picked - – without Measure Track Motion) and Area over time, sortable by the first three. Click a row to select its track on the image - or an object to select its track (Click an object to select its track). ‹ › step to the previous or next image with tracks; T under the image steps through the time points or plays them.

The panel at the top right of the image sets the trails: Trail Off, the last 10 time points or the Whole track; All tracks or Selected only; Color by Track or Class. With a trail of N time points, − / + make it shorter or longer (by 1, or by 5 above 20); the length is remembered.

Results, Tracks split with the image: one image's tracks as a table, their trails on the image, the selected track highlighted

To follow a property of the tracks over time, use the Over time chart with A line per: Track.

NEW A Paths view (+ › Paths, offered when the analysis has tracks) compares how the tracks move, side by side:

  • Paths from their start - every track drawn from the same origin, so straight movers point outwards and wandering ones stay in a tangle near the centre. At most 2000 paths are drawn (an even sample); the header says how many there are.
  • Motion map - a dot per track: its straightness (LIN, 0 to 1) against its Turning angle or its MSD α, picked under Motion map, y axis. Directed movers sit at the right, confined or randomly moving ones to the left; an MSD α above 1 means directed, below 1 confined motion. At most 5000 dots are shown. The map needs a Measure Track Motion step after the tracking - without one, it says so.

Click a dot to select its track: its path is highlighted, and the caption names it with its values (“#12 · A01_01 · LIN 0.91 · turning +2.3° · MSD α 1.84”).

The settings on the right: under DATA the Class and Image, At least N time points (10 in a new view), From the first time point and Not touching the edge (on in a new view) - as in Tracks; under DISPLAY Color by Track, Well or Class, and Motion map, y axis Turning or MSD α. The view doesn’t sort tracks into motion types or apply thresholds - it shows the values to judge by.

Results › Paths: every track drawn from its start on the left, the motion map (LIN against turning) beside it - the straight movers bunched at LIN 1 on the right -, the DATA and DISPLAY settings on the right

NEW A Trees view (+ › Trees, offered when the analysis has traced neurons from Trace Filaments) shows each neuron’s branching:

  • NEURONS - the head counts them (“12 in 3 images”). Each image is a group row, its neurons below it by number (#1, #2 …) with their Length px, Primary branches, Branch points and Tips, the values of Measure Filament Tree. Click a column’s head to sort by it (▾ marks the column; Length px at first) - the neurons are sorted within their image. With only Trace Filaments, the list says “Add Measure Filament Tree for branch levels and lengths.”
  • The dendrogram of the selected neuron, its totals in the header (“Length 1374.85 · Primary 6 · Branch points 19 · Tips 25 · Highest level 4 · Longest path 156.58”): left to right the path distance from the soma, a row per tip, each branch point between its children. Color by Branch level or Strahler order. Zoom with the mouse wheel, pan by dragging.

Click a branch to select its segment: the cards show its BRANCH LEVEL, STRAHLER ORDER, PATH DISTANCE, LENGTH and STRAIGHTNESS, and with Split with image the segment is selected on the image. A click on the image selects the neuron and segment under it.

Results › Trees: the traced neurons with length, primary branches, branch points and tips; the selected neuron's dendrogram colored by branch level, its totals in the header

Split with image in the results bar puts the results on the left and the image on the right, with a divider between them that you can drag. It works with List, Plate, Tracks and the Profile chart.

The analysis’ objects are drawn on the image, in their class colors, with their real outlines - the results keep each object’s mask. (With Keep object outlines in the results switched off in Project settings › Objects, the results file is smaller and the objects are drawn as ellipses.) Select a row in the List and the image flies to its object - a yellow crosshair and a card (class, ID, area, the mean of a channel, position, and for related objects its Parent and Children) mark it. Click an object on the image to select its row.

  • ↑ / ↓ (or ‹ ›) step to the previous or next object; Esc lets go of it.
  • Follow selection switches the flying off - the image then stays where it is.
  • Color by Class or Value - with Value, the objects take the heatmap’s colors for the plate’s Property. Value needs a property each object has a value of - not Count.
  • In Plate, a well shows its images in a filmstrip under the image.
  • Image opens a drawer over the image with the Classes of the analysis and the image panel. In the Classes list, the eye shows or hides a class’s objects on the image, and a click on a class draws its objects on top of the others; the classes can’t be edited here.
  • Open in Images ↗ opens the image in the editor.

The width of the results, whether the split is on, and Follow selection are remembered. With Reduce motion on (EVA › Project settings… › Viewport), the image jumps to the object instead of flying there.

The list on the left, one row selected; the image on the right with the object ringed and its card

The plate heatmap on the left, a well selected; its images as a filmstrip under the image, Color by: Value

Export… in the results bar opens the export card below it: what to export, the Format, the Classes and the folder (To). More options › shows the rest. Every checked option is written out together in one run. The card starts from the view shown - its columns, classes and aggregations -, each time it opens: a change to the table since the last export (A row per, Aggregate, Transpose) is taken over. Settings only the export has - Format, the folder, Z planes and time points, the plate options - keep their last values.

The export card below Export…: what to export, Format, Classes and the folder

What:

  • Objects - one row per object - with Images to pick (all by default; images left out are skipped unless you select them explicitly), and optionally A row per coloc pair (as Coloc pairs in the List) and Each image in its own file.
  • Per image - the same aggregated-by-image data as a row per image in the List; optionally the Image heatmap (the per-image spatial heatmap grid, with its own square size).
  • Time course - per time point NEW - for time-lapse data: the values per frame, as the Over time chart shows them. A line per Class, Image, Well or Track; Statistics - any of mean, SD, median, Q1, Q3, min, max, sum (default mean and SD); Tables Long, Wide or Both (default; not offered for Track); optionally Relative to the first time point (%), and with A line per Track Each track with its parents’ history (as History: With parents’ in the chart).
  • Measurements - one row per value NEW - when the analysis has measurement steps: every value on a row of its own, with what it is about - image_name, image_rel_path, z_stack, t_stack, region, object_id, related_object_id, object_class_name, step, measurement, unit, key, value - and, in results of a current version, position: the position of a profile value (e.g. the radius of a Sholl value), empty for every other value. This long layout reads straight into R or Python, and a new measurement needs no new column. The measurements picked under More options › Measurements are written (none picked: all), of the chosen images, Z planes and time points. Written as measurements.xlsx / .csv.
  • Tracks - one row per track NEW - the tracks table: who (image, well, class, track), when (time points, duration), how it moved (path, distance, speed, straightness - and with Measure Track Motion its 16 motion values), whether it touches the edge or is there from the start, then its measurements. Lengths in µm when the pixel size is known, durations in minutes when the frame interval is. For results with lineage, the columns Parent tracks, Child tracks, Lineage (the family’s first track), Generation, Starts with and Ends with (e.g. Division) follow Track; several IDs are written as 3;7. Options: Statistics of the measurements (default mean), At least N time points, Only tracks from the first time point and Leave out tracks touching the image edge.
  • Profile - per radius (e.g. Sholl) NEW - when the analysis has profile measurements: Profile picks the measurement; What is The lines (per line and radius: n, mean, SD, median, Q1, Q3 - with A line per Class, Image, Well or Plate, as in the chart) or Each object (per object and radius: its value); Tables Long, Wide or Both. In the wide tables a radius an object doesn’t reach is 0, as in the chart.
  • Per well - plate layout (XLSX) - the plate as a heatmap; optionally Plate as list, Wells as heatmap and Wells as list, with Colors for the heatmaps. Grouping, plate size and well layout are taken from the project’s plate settings - the same ones the Plate view uses.
  • Images with objects drawn (PNG) NEW - a PNG per image and analysed plane, its channels in their colors and the objects of the export’s Classes drawn in their class color. Draw the objects picks Outline (the object’s contour), Filled 50 % (half transparent, contour opaque) or Box (its bounding box). Objects whose outline the results don’t keep (Keep object outlines in the results off) are drawn as their box. An image analysed as a z-projection is written projected. Brightness and size are under More options › Images.

Format - XLSX (default), CSV, or Parquet.

More options ›:

  • Name prefix - put in front of every output file name as {prefix}_{name} (e.g. exp1_list.xlsx), so exports of several experiments can share one folder. Characters that aren’t allowed in file names are replaced by _.
  • Transpose the tables - writes the object and image tables transposed, with the classes next to each other.
  • Measurements and Aggregations - which columns and aggregations to write.
  • Z planes and Time points - All (default) or a Range.
  • Images - for Images with objects drawn: Brightness As in the viewer (each channel’s display range from the viewer) or Auto per image (each image scaled to its own 0.5 % - 99.5 % values) - channels hidden in the viewer are left out either way, and Size - the longer side at most 2048, 4096 (default) or 8192 px; smaller images keep their size, large ones are read at the pyramid level that fits.

The export card with More options open: name prefix, transpose, measurements, aggregations, Z planes and time points

The image heatmap and the plate and well exports are XLSX-only. Parquet goes further: picking it ignores every other setting (filters, columns, grouping, checkboxes) and writes a single objects.parquet file - a raw, unfiltered dump of every column in the results database (and, when the analysis has measurements, measurements.parquet beside it), meant for downstream tools that read Parquet natively rather than for a human to open.

In CSV exports, a text - a class, image or measurement name - that starts with =, +, -, @, a tab or a carriage return gets a ' in front, so Excel or LibreOffice don’t run it as a formula (a project shared with you could carry such names). Numbers, negative ones too, are written as they are.

In XLSX exports, values that come from (or, for a well/plate, are affected by) an image left out are grayed out and struck through.

Click Start export; a progress bar tracks the run and a message confirms completion or reports an error. For XLSX/CSV, results are named for what they contain (list.xlsx/.csv, grouped_by_image.xlsx/.csv, measurements.xlsx/.csv, plate.xlsx, well.xlsx, plate_list.xlsx, well_list.xlsx, heatmap_{image}.xlsx, time_course.xlsx - a Long sheet plus the wide sheets - or time_course.csv plus time_course_{name}.csv per wide table, tracks.xlsx/.csv, profile_lines.csv (long) and profile_lines_{statistic}.csv (wide) or profile_objects.csv and profile_objects_wide.csv - in XLSX one workbook with a sheet each; the PNGs go to images/ with the image’s folders kept, e.g. images/B1/img.png - with _z0002 / _t0013 added when the results have several Z planes or time points), each preceded by the name prefix if one is set.