------------------------------------- NEURODEBLUR RELEASE NOTES ------------------------------------- NeuroDeblur 1.5.7 (18/12/2025) – Update Recommended - FIX: Resolved an error occurring when using custom block sizes for deconvolution splitting. Previously, an issue arose if the remainder of the split (e.g., a remainder of 10 when processing 300 files with a block size of 30) was smaller than twice the PSF size in that dimension. The algorithm now automatically adjusts the number of blocks to ensure no block is too small. - FIX: Addressed a "Windows access error" that could occur when overwriting an existing output folder. The automated delete-and-recreate process has been redesigned to be more robust. - FIX: Resolved an issue that occasionally caused visible striping artifacts at block boundaries along the z-axis. - FIX: Corrected a scaling error in the calculation of confocal PSFs. - FIX: Any Possibly occuring NaN value in the deconvolved images are substituted by zero values now. - FIX: Background subtraction settings were not correctly restored when applying "restore settings" from the GUI. - IMPROVEMENT: Enhanced error handling across all deconvolution tools to better detect and manage corrupted TIFF files. ------------------------------------- - NeuroDeblur 1.5.3 (10/23/2025) – Update Recommended - Fixed an issue where some OME-TIFF files were not imported correctly. - Further improvements and corrections to the (still experimental) PSFExtractor module. - Significant performance boosts in GPU-based adaptive background removal and FFT calculations across multiple GPUs. - IMPORTANT FIX: Fixed a bug that caused deconvolution in CPU mode to run on a single core, severely slowing down performance. - Numerous internal code improvements and restructurings to enhance overall program stability. ------------------------------------- NeuroDeblur 1.5.2 (09/09/2025) – Update recommended - IMPORTANT FIXES AND IMPROVEMENTS IN PSFextractor: The, still experimental, PSFextractor has been massively improved. Several major bugs were identified and corrected (e.g., extracted PSFs were sometimes not centered properly, the option for rendering the extracted psf symmetrical did not allways work as expected). To improve bead recognition, the bead extractor now utilizes a theoretical PSF model to estimate the expected size range and shape of valid beads. Consequently, several optical parameters must now be specified in the bead extractor (e.g., microscope type, NA, refractive index) allthough they not directly influence the shape of the exctracted PSF. If you are using the PSFextractor, updating is strongly recommended! - IMPORTANT FIX: Resolved a bug that affected the correct calculation of the PSF for confocal data. - IMPROVEMENT: Further optimized the loading speed of 2D image stacks. - MINOR FIX: Performed various internal optimizations to improve overall program stability. - MINOR FIX: Applied several GUI enhancements for a cleaner appearance. ------------------------------------- NeuroDeblur 1.5.1 (08/24/2025) – Update recommended - IMPORTANT FIX: Resolved a major glitch introduced in the previous version that rendered the PSF extractor non-functionalIMPORTANT FIX: Resolved a critical issue introduced in the previous version that caused the PSF extractor to stop functioning. - IMPROVEMENT: More robust file reading – if a read error occurs when loading image data, the operation is retried up to three additional times before an error is raised. - IMPROVEMENT: Enhanced compatibility when reading non-standard OME-TIFF files. ------------------------------------- NeuroDeblur 1.5.0 (08/14/2025) – Update recommended - MAJOR IMPROVEMENT: Histogram calculation is now significantly faster thanks to the use of partial histograms. These are computed for each processed data block and later combined, eliminating the need to re-read all blocks just to compute the global histogram for the entire image stack. For large datasets, this can save several minutes—sometimes 10 minutes or more—of total processing time. - Added a Help menu entry to display a list of recent fixes and improvements (release notes.txt). - Various internal enhancements to improve stability and remove minor bugs. ------------------------------------- NeuroDeblur 1.4.6 (07/07/2025) – Update recommended - MAJOR IMPROVEMENT: NeuroDeblur’s sophisticated adaptive algorithm for adaptive background removal has finally been ported to the GPU, resulting in a vast increase of speed on systems equipped with modern NVIDIA GPUs (10× faster or more, depending on the graphics card). On systems without a suitable GPU, the program will continue using CPU processing as before. - FIX: In some cases, empty or incomplete data folders generated by NeuroDeblur were not properly removed after the program was terminated. This issue has been resolved in the current version. ----------------------------- NeuroDeblur 1.4.5 (06/27/2025) – Update recommended - MAJOR IMPROVEMENT: The adaptive background subtraction has been completely reworked. As a key enhancement, the optimal radii for rolling ball filtering in the three spatial directions can now be determined automatically based on the selected mode (auto, auto-conservative, or auto-progressive). Auto-progressive provides the strongest effect. In adaptive – user-defined mode, the filter radius can alternatively be specified manually (e.g., 70) or individually for each direction (e.g., 70:50:40). Several bugs have been fixed, especially those affecting the processing of image stacks with non-isotropic voxel sizes. These improvements result in significantly enhanced performance of the reworked adaptive background removal algorithm. ----------------------------- NeuroDeblur 1.4.4 (06/04/2025) - IMPROVEMENT: Deconvolutions initiated via the console (i.e., without using the GUI) can now be reloaded into the GUI for review or re-execution with different parameters. Previously, this functionality was not supported. - IMPORTANT BUG FIX: Stopping a running deconvolution from the GUI during certain stages (e.g., loading or saving data, background filtering) could leave orphaned tasks active. These tasks could significantly slow down the system and consume excessive memory. This issue has now been resolved. ----------------------------- NeuroDeblur 1.4.3 (05/28/2025) -IMPORTANT IMPROVEMENT: NeuroDeblur splits large deconvolutions into multiple blocks that are processed sequentially. In this version, the overlap between these blocks has been increased by a factor of four, leading to improved results with fewer stitching artifacts. The stripe-shaped intensity variations that were visible in some (but not all) samples after deconvolution should now be eliminated. -Fixed a glitch in the Color Selector tool window of the "Color Combiner," which previously caused only half of the window to be visible. ----------------------------- NeuroDeblur 1.4.2 (05/16/2025) - fixed a regression in the PSFextractor plugin causing the bead extraction process to crash if using OME-tiffs comprising channel names as input. ----------------------------- NeuroDeblur 1.4.1 (05/15/2025) - In all plugins the channel names are now displayed when working with multi-channel OME-tiff data. - a few minor bug fixes. ----------------------------- NeuroDeblur 1.4.0 (05/05/2025) - Deconvolution now runs about 15% faster with GPU acceleration. - Inspecting OME-TIFF data in the NeuroDeblur GUI (ImageViewer, ResultWindow) is significantly faster, as data no longer need to be copied to temporary folders. - Reading data blocks from large OME-TIFF libraries is now 10–20% faster and more memory-efficient, thanks to the use of Dask arrays. - The plugins ColorCombiner, Destriper, CLAHE, Sharpener, and PSF Extractor can now process OME-TIFF files directly, without requiring prior conversion into 2D image folders. You can select the desired image series (S), color channel (C), or time point (T) via the GUI. - Estimating global background intensity in OME-TIFF files no longer requires loading the entire image stack into RAM. The process now splits data into smaller blocks, ensuring no single block uses more than 50% of available memory—ideal for systems with limited RAM. - In the rare event of a read/write error during block processing, all deconvolution tools now attempt two additional retries before reporting an error. - The original image display window now includes a context menu option to save the currently displayed image as a JPG file. - Writing deconvolution results to 3D OME-TIFF libraries is now faster and more memory-efficient. - For each performed deconvolution, the PSF used is saved in a folder named "psf". - The option to read NGFF folders has been removed from the GUI but remains accessible via the console interface. - The licensing system has been updated: activating NeuroDeblur now requires a personal serial number. Hardware locking has been relaxed—system changes such as BIOS updates should no longer invalidate an existing license. - FIXED: A crash that occurred when pressing the ESC key during image extraction from an OME-TIFF library. - FIXED: A bug that could occasionally crash the GUI after completing deconvolution of an OME-TIFF image stack. - FIXED: A crash caused by OME-TIFF files with missing or invalid metadata. ----------------------------- NeuroDeblur 1.3.9 (03/02/2025) - The NeuroDeblur installer has been updated to optionally invoke the NVIDIA CUDA Toolkit installer, providing the necessary libraries for CUDA-accelerated FFT calculations. Previously, these libraries were integrated directly into the NeuroDeblur installer, a method prone to errors due to potential changes with each CUDA version. With this update, an internet connection is required during installation. However, if a compatible GPU toolkit is already present on your system, you may skip this step. Alternatively, the GPU toolkit can be downloaded from NVIDIA's official website and installed manually before or after the installation of NeuroDeblur. For more information and to download the CUDA Toolkit, please visit NVIDIA's CUDA Toolkit page. - By outsourcing the CUDA components, the installer size has been reduced by approximately 2.5 times, resulting in faster download and execution times. - Resolved an issue where temporary files generated during deconvolution were not always deleted after use. ----------------------------- NeuroDeblur 1.3.8 (02/26/2025) - Lowered the limitation for the minimal allowed voxel size from 0.1 to 0.01 µm. ----------------------------- NeuroDeblur 1.3.7 (02/18/2025) - IMPORTANT BUGFIX: Fixed an issue where the 'OriginalData Window' did not display the 'bit depth selector' for OME-TIFF data. The bit depth selector now allows a minimum of 8-bit instead of 10-bit. - Calculated FWHM values are now included in the PSF.info file. - FWHMx and FWHMy are now calculated separately. Previously, the larger of the two was reported as FWHMxy. - When loading a PSF for deconvolution, values for dxy, dz, FWHMx, FWHMy, and FWHMz are now imported if a PSF.info file (generated by PSFextractor) is found in the PSF folder. - Bubble help fields now display the allowed parameter limits when clicking on a field name. Almost all "bubble help" texts have been revised for clarity. - Allowed parameter ranges have been adjusted to more practical limits. - Fixed inconsistencies between allowed parameter ranges for GUI input fields and console tool. ----------------------------- NeuroDeblur 1.3.5 (01/29/2025) - MAJOR BUGFIX: Fixed a severe bug in psfdeconv.exe that occasionally caused incorrect PSF rescaling. Please update if you perform deconvolutions using a measured PSF. - FWHMx and FWHMy are not calculated seperately (before just a single parameter FWHMxy was reported that was the bigger on of the two) - FIX: Resolved a bug where the 'Make Windows Same Size' tool in the result window did not work correctly when the displayed image was part of a dataset in OME-TIFF format. - Ongoing development and major improvements in PSFextractor. ----------------------------- NeuroDeblur 1.3.3 (01/22/2025) - fixed a bug in the NeuroDeblur Installer introduced in V1.3.2 - minor improvements in the PSFextractor. ----------------------------- NeuroDeblur 1.3.2 (01/21/2025) - several minor bug fixes - improvements in the NeuroDeblur installer ----------------------------- NeuroDeblur 1.3.1 (01/15/2025) - Added new plugin for PSF-extraction from recordings of fluorescent beads (still experimental and under ongoing construction). - Last selected path is now memorized for almost all folder selection boxes (seperately for all typess of folders). - Several minor improvements and fixes ----------------------------- NeuroDeblur 1.2.13 (10/17/2024) - FIX: NeuroDeblur crashed when trying to load images of diferent sizes into the 'ColorCombiner' module. Now an error message indicating the problem is displayed. - FIX: fixed an issue introduced in one of the last versions causing that the context menu was not shown anymore after right clicking on an image displayed in a 'result window' or on a preview image shown in the 'preview image bar'. ----------------------------- NeuroDeblur 1.2.12 (09/26/2024) - Upgraded to support the latest CUDA drivers (NVIDIA GPU toolkit version 12.6). - Added an option to check for updates in the Help menu of the GUI. - The Color Channel Combiner now logs "clip range settings" for each channel in the parameters.txt file. - BUG FIX: Resolved an issue where the GUI window would not restore properly after being minimized (introduced in version 1.2.11). ----------------------------- NeuroDeblur 1.2.11 (09/25/2024) - Fixed a crash that occurred when NeuroDeblur attempted to use more than 60 processors. The maximum number of processors is now limited to 60. - Corrected an issue with the installer for version 1.2.10, which prevented NeuroDeblur from functioning properly in some cases. - Minor UI design adjustments. ----------------------------- NeuroDeblur 1.2.10 (08/24/2024) - Filenames copied to the clipboard from command lines now automatically include quotation marks, allowing for seamless use in shell windows when filenames contain spaces. - The installer is now digitally signed, removing the need for a safety warning message prior to installation. - Fixed an issue in version 1.2.9 where the "Check GPU compatibility" button in the Preferences dialog was non-functional. ----------------------------- NeuroDeblur 1.2.9 - Fixed multiple bugs introduced in version 1.2.8 due to switching to a new GUI toolkit with compatibility issues (please avoid using version 1.2.8). - Corrected errors in the documentation (accessed via the -h command) for multiple deconvolution toolkit commands. ----------------------------- NeuroDeblur 1.2.8 - Empty output folders generated during runtime are now reliably removed upon program exit. - Improved processing speed with parallelized PSF computation in light sheet mode. - Fixed an issue where the content of the 'original data' window was not updated correctly when selecting a different slice in the 'results window.' - Resolved a problem that prevented the reloading of deconvolved data if its folder was moved to a different location. ----------------------------- NeuroDeblur 1.2.7 (07/12/2024) - Added support for the latest Nvidia drivers (version 55.85 or later is required). Please update your driver before installation. - Enhanced spherical aberration correction for confocal and brightfield deconvolution modes. - The 'obj_spec_str' parameter for deconvolution tools (bfdeconv.exe, cfdeconv.exe, lsdeconv.exe, sfdeconv.exe) now includes the descriptor 'mode:{0,1}' to specify the z-axis scanning method. 'mode:0' scans the objective focus through the fixed sample, while 'mode:1' scans the sample through the objective focus. - Improved synchronization options between the deconvolution result window and the corresponding original image window, with separate controls for region and brightness/contrast synchronization. - Added a new option in the context menu (accessed by right-clicking the completed deconvolutions list) to display detailed command string descriptions. - Various bug fixes and code optimizations have improved speed and performance. - Minor improvements to the UI design. ----------------------------- NeuroDeblur 1.2.6 - Resolved an issue where the main window would disappear after being moved to a second monitor and maximized. - Fixed a critical bug in psfdeconv.exe where the program crashed when displaying the help text using `psfdeconv -help`. - When changing the output type (TIFF-folder or single file) under 'Preferences,' the program now prompts for confirmation to apply the change to the current job list. ----------------------------- NeuroDeblur 1.2.5 (05/07/2024) - Updated command-line parameter order for lsdeconv.exe to match other deconvolution tools (bfdeconv, cfdeconv, sfdeconv). The order is now: ..., rf, lambda_ex, lambda_em, ... - The license dialog now includes an option to request a 14-day demo license. -----------------------------