- Guide
- Imaging
- Image analysis results
Image Analysis Results
When a run finishes, the middle of the workspace switches to what it found. Three tabs under the analysis's name move between the run's views, Viewer, Measurements, and Gallery, and Add to and Export at the end of that row act on the view that is open. On a phone they are the plus and download icons beside Comments. A run that detects objects also lists them, each with its cut-out, as Objects describes.
Viewer
Viewer is the single image you were tuning, full size, with the inspected step's overlay drawn on it and the row underneath to move between images. The tabs appear once a run has covered the set, and until then the Viewer fills the workspace.
On the Viewer, Add to puts the image on screen in a figure, a note in the Notebook or a slide deck. A new note started from a private image analysis is private too. Pick from This image, with outlines to send it with the inspected step's overlay drawn on it, or from This image to send the image alone. The outlines are the ones on screen when you send it, and they stay that way after later runs.
Measurements
Measurements is where a finished run opens, and it puts the images side by side. A pipeline that measures several things per image, or every object separately, shows a chart over the table it is drawn from.
Picking a Chart
The button at the top left of the chart names the chart on screen, with a picture of its kind. Press it to pick another of the charts this run can be drawn as. Each is listed with its kind, such as Bar or Box plot, and what it plots. Other measurements by image at the end of that list plots the numbers the named charts leave out, one at a time. Under Draw as, pick another kind of chart for the same numbers.
The settings button beside it opens the column picks, so you can take any of these somewhere else. A run that detects objects shows the objects table under the chart instead.
On Measurements, Add to puts the chart in a figure, a note in the Notebook, a report or a slide deck. A new note or report started from a private image analysis is private too. In a note, a report or a deck the chart redraws as the run's numbers change. A chart that can't be redrawn from the saved results, such as the bars comparing each image, goes in as a new figure instead.
Accept run 6 at the foot of the numbers settles them and puts them in a note, as Accepting a Run describes below.
Reading the Numbers
A fraction like area covered is shown as a percentage. A per-object pipeline writes a row per object, so a dozen images runs to thousands of rows. A z-stack writes a row per focal plane that ran, named for the image and the plane, such as Plate 3 · z4, so a stack trimmed to 13 slices writes 13 rows, one per slice.
A pipeline that writes one row per image, such as a count of objects, also records what each setting left on Auto came to on each image. Each of those settings gets a column after the measurements, named for the setting, such as Threshold used or Background offset used, and a count shows it beside each image's bar. Read down it to see whether the automatic cutoff moved across the set. The chart, the headline number, and Compare leave these columns out.
Gallery
Gallery gives one card per image with its headline number and how many checks fired. Click a card to open that image in the Viewer. An image the pipeline couldn't analyze is named in a band above the cards, with the reason.
Objects
A run that detects objects lists every one in a table. Its cut-out comes first, then each number the run measured, such as Area (µm²), ECD (µm), Convexity and Eccentricity. The table sits under the image in the Viewer, and under the chart in Measurements, where an Image column names the image each object came from.
A pipeline that only finds objects, with no measuring step after it, still lists what finding them measured.
| Column | What it is |
|---|---|
| Area (px²) | How many pixels the object covers. On an image with a scale it reads in µm² instead |
| Mean brightness | The average of the object's pixels |
| Brightness std dev | How much those pixels vary |
| Integrated density | The object's area times its mean brightness, so the total signal in it |
| Color | The color the object was sorted into for the color filter |
| Confidence | The detector's score for the object, when the detector gives one |
Sort by Mean brightness to see which objects sit nearest a brightness cutoff.
- Click a column heading to sort by it, largest first, and click again for smallest first.
- Small and Large set the size of the cut-outs.
- On an image with a scale, µm and px switch the sizes between microns and pixels. The exported CSV keeps both.
- Click an image name to show only that image's objects, and the × on its tag to go back to every image. The search box above the table finds an image by part of its name.
- Export beside the view tabs, then Objects, saves the table in one of three formats. When the table is narrowed to one image or a search, it can export only those objects.
| Format | What it holds |
|---|---|
| CSV | One row per object, with every size in both microns and pixels |
| Excel workbook | One row per object with its picture beside its numbers, up to 5,000 objects |
| ZIP of the CSV and pictures | The CSV and a PNG of every object, each named in the CSV beside its numbers, for any size of set |
A row with a blank square where its cut-out goes belongs to an image you haven't opened yet. Open that image in the Viewer and its cut-outs fill in.
When a filter or the detector threw objects away, the heading says how many were kept of how many were detected, such as 12 of 15 detected, and Show rejected (3) adds the rest as grayed rows with the reason beside each cut-out.
Filtered by size, 9 px names the filter that dropped it and the value it was dropped on, so you can see whether your cutoff is landing where you meant. Outside the area counted and Glare are the detector's own. A run made before Conspecta started keeping rejected objects has no Show rejected, so run the set again to get it.
The Results Table
Every run writes its numbers into a results table automatically. That table belongs to the image analysis itself.
It takes the analysis's sharing too, so the numbers from an image analysis you keep private stay yours, and sharing the image analysis hands them to the project. To read one run against another, Compare Image Analysis Runs puts several of them into a single table.
Exporting
Every file a run can be saved as is under Export, at the end of the row of Viewer, Measurements and Gallery tabs, or the download icon beside Comments on a phone. It appears once a run has results.
| Section | What it saves |
|---|---|
| Results table | The run's rows as Excel (.xlsx), with the headings and percentages the table shows, or as CSV (.csv), with the column names the results table stores, for a script that reads them |
| Objects | For a run that detects objects, the objects table as CSV, Excel or a ZIP with a picture of each object, as Objects describes |
| Chart | On Measurements, the chart as a PNG, SVG, TIFF or PDF picture |
Export in the Analysis tab's header writes the list of every image analysis in the project as CSV or Excel, with each one's name, type, status, how many images went in, how many runs came out, when it last ran, and its headline result. Check rows first and Download on the bulk bar opens the same dialog box on the rows you checked, where checking a folder includes what is inside it.
Export in the Images tab's header downloads image files. Scope is every collection in the project, or the ones you checked, which is what Download on the bulk bar opens on, and checking a collection includes the images inside it. One image comes down on its own and several as a ZIP, up to 256 MB at a time.
What Produced These Numbers
Once a run has covered the set, the sidebar's steps lock to the ones that run used. Analysis run at the top shows the run whose numbers are on screen, with when it finished and how much it covered. Under it, a run that checked control images or hand-counted marks lists how each came out, such as Controls Every control image measured inside its expected range, and Hand count Found 20 of 20 marked objects. They stay there when you open the image analysis later.
Right after you run the set, Accept run 6 sits under it too, the same button as the one under Measurements. Other runs under it lists this image analysis's remaining runs, and the link under them, such as Show 2 more runs, opens the rest. Pipeline under it holds the locked steps.
The settings listed there are the ones that produced the numbers beside them. Change a setting, or which slices of a z-stack run, without running again and a note appears over the results naming the run that produced them.
Edit, at the end of the Pipeline heading, unlocks the steps. The same button then reads Done and locks them again, and running the set locks them too.
Going Back to an Earlier Run
Each run in the list shows its headline number, so you can compare what each one found. Use these settings, the clock button at the end of a run's row, brings that run back. Its measurements return to the screen and its steps return to the sidebar.
Nothing runs again, so it takes no compute and adds no run to the list. The run moves to the top of Analysis run, and Accept run 3 at the foot of Measurements now names it, so you can accept the run you came back to.
The buttons show while the steps are locked, so press Done first when you are editing.
Bringing a run back replaces what is in the results table with that run's numbers. A row you typed into the table yourself belongs to no run, so it goes with the rest. To keep one, copy it somewhere else first. To get a later run's numbers back, press Use these settings on that run.
Runs recorded before September 2026 kept no copy of their measurements. On those, Use these settings brings the steps back and says so, and you run the set again to measure it. An image analysis keeps every run it has ever recorded.
Comparing a Run With the One on Screen
Compare with the run on screen, the two-arrow button on a row under Other runs, sets that run beside the run on top. Nothing runs again. While you compare.
- Measurements lists each image with its count on both runs, such as Run 3 and Run 6, the Difference between them, and a Total for each run.
- Gallery shows one image above the cards with one run's objects drawn on it. Switch between Run 3 and Run 6 to see which objects split, merged or went missing, and press a card to show that image.
An image a run kept no objects for reads Not stored, and the totals add up only the images both runs counted. A run that detected with an AI model has no compare button. Stop comparing, or the same button on the row, goes back to the run on screen.
Accepting a Run
When the numbers are right, Accept run 6 settles them. It sits at the foot of Measurements and names the run on screen.
Press Accept run 6 in the dialog box to accept the run on its own. To put it in a note as well, check Add to a note first and pick a note. Under Add to the note, keep checked what the note gets.
- The steps and their settings, for the method.
- A copy of the counts as a table, for the numbers. On an image analysis that measures, it's called A copy of the measurements as a table.
Accepting a run does three things.
- The image analysis moves to Ready.
- The run shows who accepted it and when.
- A note you picked gets the steps and a sheet named after the image analysis and the run, such as Nuclei count (run 6).
Open in the message that follows takes you to the note.
The run whose numbers are on screen is the one that can be accepted, once it has finished, and only once. That is the newest run until you bring an earlier one back with Use these settings, which makes that run the one you can accept. Anyone who can edit the project can accept a run.
What the Note Gets
The note lists the same steps as Pipeline, in the same order. The text names the method used for each setting left on Auto and the range it came to across the set, such as Threshold: Auto (Otsu), resolved to 96 to 131 across 12 images.
When the image analysis was built from a template, the line under the run's heading names that template and the version it was built from, such as "Built from the template Nuclei count, version 2". If the run changed any of that version's settings, the line says how many.
That sheet is a copy, listed in Sheets with the rest. Its columns have the same headings as the results table, and its rows follow the order of the images in the image analysis. Later runs rewrite the image analysis's own results table and leave the copy's numbers as they were accepted.
The copy is shared with the project when the image analysis is. A private image analysis gets a private copy, so teammates reading a shared note see a placeholder until you share that sheet from Sheets. Changing the image analysis's sharing later leaves the copy as it is.
Tip: Reopening a finished image analysis restores its numbers without re-running anything, which uses no compute usage. Measurements and Gallery come back in full, and Objects fills in for the images you run again.