Copenhagen, Denmark
Onsite/Online

ESTRO 2022

Session Item

Saturday
May 07
16:55 - 17:55
Poster Station 2
08: Advances in radiotherapy planning & techniques
Madalyne Day, Switzerland
1600
Poster Discussion
RTT
CTV delineation for high-grade gliomas: is there agreement with tumor cell invasion models?
Wille Häger, Sweden
PD-0327

Abstract

CTV delineation for high-grade gliomas: is there agreement with tumor cell invasion models?
Authors:

Wille Häger1, Iuliana Toma-Dasu1, Marta Lazzeroni1, Mehdi Astaraki2

1Stockholm University, Department of Physics, Stockholm, Sweden; 2Royal Institute of Technology, Department of Biomedical Engineering and Health Systems, Huddinge, Sweden

Show Affiliations
Purpose or Objective

High-grade gliomas (HGGs) infiltrate normal tissue diffusively. The clinical target volume (CTV) accounting for the microscopic spread of clonogenic tumor cells for HGGs is typically delineated by extending the gross tumor volume (GTV) margin by 2 cm while avoiding anatomical barriers. Despite a wide margin for CTV delineation, tumor recurrence is persistent and prognosis remains poor. Mathematical models have been proposed to help determining the tumor invasion outside the observable GTV.

The purpose of this study is to investigate the agreement of the conventional CTV delineation with respect to the predicted tumor invasion.

Material and Methods

Data consisting of 108 T1-weighted MRI scans of HGG cases were analyzed. The images were segmented into white and gray matter, anatomical barriers, and tumor tissue, using an automated artificial intelligence-based algorithm. Tumor invasion was simulated using a Fisher-Kolmogorov equation which describes tumor cell concentration by accounting for cell proliferation and cell diffusion. Simulations were run until the 8000 cells/mm3 isocontour (the MRI visibility threshold) encompassed the segmented GTV. A CTV was delineated by isotropically expanding the GTV contour by 2 cm followed by trimming to avoid anatomical barriers. The simulated tumor was compared to the CTV using two methods: 1) Determining the highest cell concentration isoline (cenc) completely encompassed by the conventionally delineated CTV; and 2) Determining the Hausdorff distance dH(c) between the CTV and the volume encompassed by isosurfaces with given cell concentrations, c.

Results

Figure 1 shows an example of simulated tumor invasion expressed in terms of cell concentration as compared to the CTV delineation. There is poor agreement in terms of isotropy.

The values of the highest cell concentration isoline fully encompassed by the CTV, cenc, ranged from 0 cells/mm3 to 2.39·105 cells/mm3 (the normal tissue tumor cell carrying capacity). For 50% of cases cenc>1000 cells/mm3 (median cenc =902 cells/mm3). Fourteen out of 108 cases (13%) were deemed as outliers with cenc>1.27·104 cells/mm3 (>1.5IQR). In 15% of the cases, cenc was <1 cells/mm3, likely indicating an overextension of the CTV delineation.

A minimum value for median Hausdorff distance for all analyzed cases was seen for the volume encompassed by the isosurface of cell concentration in the range 200–400 cells/mm3. The absolute value of dH was on average in the same magnitude as the CTV extension of 2 cm but the simulated tumors did not conform to the CTV delineation due to their anisotropic spread. 


Figure 1: A simulated tumor invasion shown in terms of cell concentration. V0 and V correspond to the segmented and simulated GTVs, respectively. 


Conclusion

Both analyses of cenc and dH(c) indicate poor agreement between the CTV delineation and the simulated tumor invasion. Tumor invasion modelling could therefore be a valuable tool for assisting the CTV delineation of HGGs.