is an open source application designed for high throughput analyzing of microscopy images in the biological sciences.
With imageC image processing pipelines can be defined which are applied on a set of images to analyze. The resulting data is stored in a file-based database (duckdb) and can either be viewed directly in imageC using the built-in reporting tool, or the data can be exported to other file formats.
imageC (EVAnalayzer 2), the direct successor of EVAnalayzer an imageJ plugin with more than 3000 downloads (June 2024), is a standalone application written in C++. The main goals were to improve performance, allow the processing of big tiffs and improve usability.
| imageC | EVAnalyzer | |
|---|---|---|
| BioFormats support | x | x |
| OME-XML support | x | x |
| XLSX report generation | x | x |
| Max. channels | 10 | 5 |
| Max image size | no limit | 2GB |
| Support for big tiff | x | - |
| AI based object detection | x | - |
| Database based result processing | x | - |
| Heatmap generation | x | - |
| Image density map generation | x | - |
| In image object marker | x | - |
| Built-in reporting tool | x | - |
| Side by side image comparison | x | - |
Covert to
bfconvert -no-upgrade -noflat -pyramid-resolutions 4 -pyramid-scale 2 -tilex 512 -tiley 512 /workspaces/imagec/test_local/bigtif/GMEV5minM1OT3_0001.btf /workspaces/imagec/test_local/bigtif/GMEV5minM1OT3_0001.ome.btf
This section contains information needed for anyone who is interested and especially for those who would like to contribute.
All data generated during an analysis run are stored using the file based database duckdb. Following database schema is used:
The data generated for each detected ROI are stored in the objects table.
The row objects.values is a map containing the measured values.
The key of the mep describes which data is stored, the value is the double formatted measurement value.
The measure channel key is a 32 bit unsigned integer coding the measurement channel and the image channel to which the value applies.
The upper 16 bits are coding the measure channel, the lower 16 bits the image channel whereby a image channel of 0xFFFF is the this channel meaning use the channel stored in objects.channel_id.
+-----------------+-----------------+
| Measure channel | Image channel |
+-----------------+-----------------+
| MSB | LSB |
+-----------------+-----------------+
Example: The measure channel AREA_SIZE = 2 for this channel is coded as 2 << 16 | 0xFFFF = 0x2FFFF = 196607
Example: The measure channel CROSS_CHANNEL_COUNT = 17 for channel 0 is coded as 17 << 16 | 0x0 = 0x110000 = 1114112
To calculate the average value of the area size of image 4261282133957314495 following SQL query can be used:
SELECT
AVG(element_at(values, 393215)[1]) as val_avg
FROM
objects
WHERE
objects.image_id=4261282133957314495The object.image_id identifies an image unique per run.
This id is calculated by the fnv1a hash: object.image_id = fnv1a(analyzes.run_id + <ORIGINAL-IMAGE-PATH>)
valgrind --tool=massif --log-file="filename" ./build/build/tests "[pipeline_test_spots]" valgrind --gen-suppressions=all --tool=massif --log-file="filename" ./build/build/tests "[pipeline_test_nucleus]"
docker build --target live -t joda001/imagec:live . docker build --target build -t joda001/imagec:v1.7.x . docker push joda001/imagec:v1.7.x
Install MSYS2 and following packages:
pacman -S --needed base-devel mingw-w64-x86_64-toolchain
pacman -S mingw-w64-x86_64-catch
pacman -S mingw-w64-x86_64-pugixml
pacman -S mingw-w64-protobuf
pacman -S mingw-w64-x86_64-opencv
pacman -S mingw-w64-x86_64-nlohmann-json
pacman -S mingw-w64-x86_64-libtiff
pacman -S mingw-w64-x86_64-python-mingw-ldd
pacman -S mingw-w64-x86_64-qt6-base
pacman -S mingw-w64-x86_64-libxlsxwriter
Add following ENV variables:
C:\msys64\usr\bin
C:\msys64\mingw64\bin
Add powershell permissions:
set-executionpolicy remotesigned
Execute make.ps1 and build_local.ps1.
The EXE file will be placed in build/build/imagec.exe
mingw-ldd.exe imagec.exe --dll-lookup-dirs C:\msys64\mingw64\bin strip.exe imagec.exe
Many thank's to the authors of following open source libraries I used:
| Title | Link | License |
|---|---|---|
| nlohmann/json | https://github.com/nlohmann/json.git | MIT |
| zeux/pugixml | https://github.com/zeux/pugixml | MIT |
| protocolbuffers/protobuf | https://github.com/protocolbuffers/protobuf | Google Inc. |
| opencv/opencv | https://github.com/opencv/opencv.git | Apache-2.0 |
| qt6 | https://code.qt.io/cgit/ | LGPL-3.0 |
| libtiff/libtiff | https://gitlab.com/libtiff/libtiff.git | Silicon Graphics, Inc. |
| jmcnamara/libxlsxwriter | https://github.com/jmcnamara/libxlsxwriter.git | FreeBSD |
| tukaani-project/xz | https://github.com/tukaani-project/xz | GPL-3.0 |
| madler/zlib | https://github.com/madler/zlib | Own |
| nih-at/libzip | https://github.com/nih-at/libzip | Own |
| duckdb/duckdb | https://github.com/duckdb/duckdb | MIT |
| ome/bioformats | https://github.com/ome/bioformats | GPL-2.0 |
Thank's to the authors of imagej I ported some image processing algorithms from to C++.
imageC is the follower of evanalyzer.