BClara
BClara is a medical data visualization tool. The tool is focused on segmentation and 3D reconstruction of livers, tumors and hepatic veins. The goal is a fast volume rendering technique specifically designed for VR (Virtual Reality) use.
The main goal of the project is the visualization of medical data with increased accuracy for segmentation of small structures, such as hepatic veins, using the fast volumetric rendering technique, which is specially designed for use in VR (Virtual Reality). The BClara extends the functionality of the already existing open-source software Blender. The software enables the use of HW resources of a supercomputer cluster for the segmentation of medical data. The BClara addon enables the reading of medical data in two formats, the segmentation of organs, bones and soft tissues and their 3D reconstruction and subsequent visualization in VR.
Addon Description
The software performs segmentation of medical data including small structures (e.g. hepatic vessels) based on deep learning models using neural networks and the Clara Train SDK framework. Part of the software works on the user side of Blender (frontend), which is mainly for data loading and visualization. The computing part of the software designed for segmentation is deployed on an HPC cluster (backend). The user must set login and password generated by IT4Innovations infrastracture to connect to the server and use the HPC resources for the segmentation task. The output is an interactive scene made up of volumetric models.
Addon Installation
The addon is stable on the version of Blender 3.3.1. Available here: https://download.blender.org/release/Blender3.3/ Since we will modify the blender folder by adding dll files, you should download the zip version (for Windows blender-3.3.1-windows-x64.zip).
DLL files & pyopenvdb. You must manually copy all files from the folder bclara/dll to your downloaded and extracted blender folder which will look like this:
Addon Preferences. To comminucate with the server, you need to set a username and password, private key and project (provided by IT4I infrastracture), and install dependencies.
If the installation is succesfful, you will see the tab called BClara with the following user interface.
Installation of Clara to the cluster
If you want to use the BClara backend and inference on the cluster, you need an access to the IT4I infrastracture. You have an option to get a computing time, for example, by using Open Access Competition. For more information: https://www.it4i.cz/en/for-users/computing-resources-allocation Then we will help you with the deployment of Clara on the cluster. Please contact me at petra.svobodova@vsb.cz
How to use BClara
In the first tutorial, you can learn how to use BClara pipeline to load data (in DICOM or NIfTI format), connect to the cluster (need your login credentials), choose the segmentation model, segment particular tissues (automatically visualize the result), create and load volume models and customize them.
In the second tutorial, you will see how to create and visualize volume objects in the scene if you have already obtained patient data and mask (both in NIfTI format). The connection to the HPC cluster is not required.
Acknowledgement
This work was supported by the Doctoral grant competition VSB - Technical University of Ostrava, reg. no. CZ.02.2.69/0.0/0.0/19_073/0016945 within the Operational Programme Research, Development and Education, under project DGS/TEAM/2020-008 "Development of a tool for scientific data processing and visualization in VR with multi-user support".