OncoFEM is a software tool to perform numerical simulations of tumours based on medical image data or academic geometries, providing a possible tumour evolution. The software is written to speed up the development towards an increasing demand for patient-specific simulations, with the ultimate goal of supporting clinicians in their treatment planning, i. e. medication, surgical interventions, classifying of severness. The structure and workflow of OncoFEM is kept general, to be open for the inclusion of different types of tumours, organs or tissues. Nevertheless its initial implementation is written for the simulation of diffusive astrocytomas (brain tumour), such as Glioblastoma multiforme (GBM). The software divides into the preprocessing of medical images and a simulation core module. A pre-processing entity is already implemented, that homogenises MRI input data and segments the tumour and heterogeneous compartments of the brain. Numerical calculations can be performed by a combination of a macroscopic base model with process models on the microscale, that mimic the cell behaviour of cells and cell cohorts. For demonstration already the implementation of a two-phase model in the continuum-mechanical framework of the Theory of Porous Media is chosen, according to the tumour microenvironment based on Wolf et al. [1]. In OncoFEM, the problem is modelled with a porous approach of a solid extracellular matrix and an intercranical fluid, wherein mobile cancer cells are resolved and measured with a molar concentration. The processes on the microscale are assumed with a logistic Verhulst equation for the mobile cancer cells that can be coupled to a solid growth term. In case of growing tumour mass fluid will be accumulated in the affected areas and a swelling can be observed. The defined set of governing equations is then solved with the finite element method using the software package FEniCS [2].

The uploaded virtual box is a virtual machine of a linux mint 21.2 cinnamon, 64 bit system and 8 GB RAM. The machine contains the pre-installed version of OncoFEM version 1.0, that corresponds to its related publication. The virtual machine need to be imported in Oracle VM VirtualBox.

Username: oncofem password: 0000

The pre-installed version of oncofem is implemented in a conda environment that is activated with the terminal command

$conda activate oncofem

The Software can be found in /home/oncofem/Software/OncoFEM with the sub-folder for the tutorials, that can be run with

$python3 tut_0*

Of course, the tutorials will only run, if the system meets the necessary requirements. The tutorials (1, 2, 3, 6, 7) where performed on a local machine (intel cpu i7-9700k with 3.6 GHz, 128 GB RAM). The tumor segmentation (tutorial 4, 5) have been tested on a different machine with a gpu (Nvidia a40, 48 GB VRAM, 32 core AMD epyc type 7452) and the one_file_tumor_segmentation.py. For testing, the discretisation in space is decreased, compared to the results published in the respective paper.

The actual development can be found on github.

[1] Kayla J. Wolf et al., Dissecting and rebuilding the glioblastoma microenvironment with engineered materials, Nature Reviews Materials, Springer Science and Business Media LLC, 2019, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7347297/

[2] Anders Logg et al., Automated Solution of Differential Equations by the Finite Element Method, Springer Berlin Heidelberg, 2012, DOI: 10.1007/978-3-642-23099-8

The software provides a tutorial to learn the basic functionalities. More information can be found in the respective paper.

For usage, the user should be familiar in operating with unix based systems. OncoFEM is a python-based software. Therefore, scripts, e. g. from the tutorial, can be created and executed with the python command:

$python3 *

In order to run the tutorial please open a terminal and run the following commands:

$conda activate oncofem $cd Software/OncoFEM/tutorial $python3 tut_01_quickstart.py $python3 tut_02_academic_example.py

Be aware, that due to the limited power in a virtual box, it is recommended to work on academic test examples, rather than full brain geometries. Due to that, the first quickstart example runs with a reduced discretisation in space