MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Assessment of Spatial Dosimetric Resolution for Voxel-Based Analyses in Radiation Oncology: How Much Radiobiological Detail Can We Find Out From a Cohort of Patients Treated with a Given Radiation Therapy Technique?

R Mohan3*, S Monti1, A Stanzione2, R Pacelli2, Z Liao3, J Deasy4, G Palma1, L Cella1, (1) Italian National Research Council, Napoli, IT, (2) University Of Naples "Federico II", Napoli, IT, (3) MD Anderson Cancer Center, Houston, TX, (4) Memorial Sloan Kettering Cancer Center, New York, NY

Presentations

(Tuesday, 7/14/2020) 1:00 PM - 3:00 PM [Eastern Time (GMT-4)]

Room: Track 2

Purpose:
Voxel-based analyses (VBA) have been introduced to evaluate local dose-response patterns via voxelwise statistical analysis on Dose Maps (DMs) spatially normalized on a common anatomical reference. Here, a novel tool was introduced to assess the spatial correlations within DMs belonging to different datasets in order to highlight the resolution issues related to VBA.

Methods:
We spatially normalized on XCAT digital phantom [Segars et al. 2010] the DMs of thoracic cancer patients from 4 datasets and treated with different RT techniques: 3DCRT for Hodgkin Lymphoma (HL); IMRT, SBRT and protons for lung cancer. We considered the cardiac substructures provided with XCAT and the lung subregions (https://radiopaedia.org/cases/bronchopulmonary-segments-annotated-ct-1) segmented by a radiologist. For each dataset, we analyzed the connectogram [Irimia et al. 2012] representing the relevant links between each pair of substructures according to the Spearman correlation (Rs) between the related mean doses. The spatial correlation between the dose delivered to distinct anatomical subregions affects the spatial resolution of the significance map from VBA for a given radiation-induced effect.

Results:
A total of 390 patients were analyzed. The connectogram identified different dosimetric connectivity patterns depending on the considered RT technique and tumor. The lung-SBRT dataset showed the shortest correlation lengths, allowing for a valuable disentangling (at Rs²<0.2) of even close small organ substructures (as lobe segments or cardiac chambers and walls). Conversely, HL-3DCRT dataset exhibits high correlations even between large, distant structures (such as lobes of contralateral lungs). An intermediate behavior was found for lung-IMRT and proton datasets.

Conclusion:
The connectogram revealed as a valuable tool to provide an insight on the resolution limits inherent to a given dataset of DMs in the VBA context. As expected, the most promising features were observed in the SBRT dataset, which involves heterogeneous dose patterns with steeper gradients than conventional RT treatments.

Keywords

Not Applicable / None Entered.

Taxonomy

Not Applicable / None Entered.

Contact Email