How CFD Opens the Door for Personalized Medicine

Airflow velocity simulation through a constricted nasal passage. The simulations compare the results from four meshing techniques (top left — tetrahedral, top right — hexcore, bottom left — polyhedral, bottom right — poly-hexcore).

Simulations give medical practitioners the ability to diagnose patients and understand the effectiveness of treatment plans. Since these simulations can be patient-specific, they are essential to personalized medicine.

For instance, computational fluid dynamics (CFD) can help medical practitioners develop targeted and controlled drug delivery devices. These delivery devices can ensure that a drug is able to reach a specific site in a specific patient’s body.

CFD can also be used to provide highly detailed information about a patient’s nasal airflow. Professor Kiao Inthavong of the Royal Melbourne Institute of Technology (RMIT) University is assessing how these CFD simulations can be used in a medical environment to diagnose an obstruction. Eventually, these computer simulations could help develop an effective surgery protocol based on personalized medicine methodologies.

As CFD is still an emerging tool in the medical and clinical environment, Inthavong needs to demonstrate that the results from these in silico simulations compare with the in vitro and in vivo results.
Inthavong’ studies start with the conversion of patients’ medical scans into 3D watertight geometry. The challenge is that these scans usually generate messy geometry that require manual cleanup. A new ANSYS Fluent workflow simplifies this conversion process.

How to Use Patient Scans in CFD Simulations

The first step in converting medical scans into watertight geometry is to create a dome around the geometry of the patient’s face. This dome is used to accurately predict airflow around the nostrils. 

The nasal passages then split the airflow into two paths. Each passage constricts to a millimeter-wide gap — based on the patient’s obstruction. The airflows then re-join before traveling toward the lungs.

Cross section of the nasal geometry (left). The patient’s constricted nasal passage (middle). The nasal passage leading toward the lungs (right).

The cleaned geometry is then imported into Fluent where Inthavong uses the watertight meshing workflow to quickly wrap a refinement domain around the model.
Inthavong tested four different mesh topologies of the geometry:

  • Tetrahedral.
  • Polyhedral.
  • Hexcore.
  • Poly-hexcore.

These tests verify the results and help to optimize the numerical protocol. The various mesh topologies were then imported into the solver with appropriate boundary conditions, to create patient-specific results.

Comparing Mesh Topologies of a Nasal Passage

Inthavong compared the meshes to verify the airflow results and to determine which mesh topology has the highest computational efficiency.

Mesh cross sections of a constricted nasal passage (top left – tetrahedral, top right – hexcore, bottom left – polyhedral, bottom right - poly-hexcore)

The mesh topologies were assessed based on:

  • Accuracy.
  • Mesh count.
  • Computational time.
  • RAM usage.

Inthavong’s mesh topologies all show similar qualitative results to the experiments. The simulation results from each topology method are also consistent with each other.

The computational efficiency, however, favored the poly-hexcore mesh. Inthavong’s results show that for this patient-specific geometry, the poly-hexcore mesh was significantly faster and used less RAM than any other mesh. It even produced results twice as fast as the slowest mesh. Using this information, Inthavong can optimize the research.

The next step is to determine how these CFD simulations can be used to develop personalized medicine techniques and procedures/p>

To learn about other nasal flow simulations performed by Inthavong, read A Healthy Outlook at RMIT University. To learn more about the watertight geometry workflow, read the application brief available here.