Search

Background: This study aimed at investigating the feasibility of developing a deep learning-based auto-segmentation model for the heart trained on clinical delineations. Material and methods: This study included two different datasets. The first dataset contained clinical heart delineations from the DBCG RT Nation study (1,561 patients). The second dataset was smaller (114 patients), but with corrected heart delineations. Before training the model on the clinical delineations an outlier-detection was performed, to remove cases with gross deviations from the delineation guideline. No outlier detection was performed for the dataset with corrected heart delineations. Both models were trained with a 3D full resolution nnUNet. The models were evaluated with the dice similarity coefficient (DSC), 95% Hausdorff distance (HD95) and Mean Surface Distance (MSD). The difference between the models were tested with the Mann-Whitney U-test. The balance of dataset quantity versus quality was investigated, by stepwise reducing the cohort size for the model trained on clinical delineations. Results: During the outlier-detection 137 patients were excluded from the clinical cohort due to noncompliance with delineation guidelines. The model trained on the curated clinical cohort performed with a median DSC of 0.96 (IQR 0.94-0.96), median HD95 of 4.00 mm (IQR 3.00 mm-6.00 mm) and a median MSD of 1.49 mm (IQR 1.12 mm-2.02 mm). The model trained on the dedicated and corrected cohort performed with a median DSC of 0.95 (IQR 0.93-0.96), median HD95 of 5.65 mm (IQR 3.37 mm-8.62 mm) and median MSD of 1.63 mm (IQR 1.35 mm-2.11 mm). The difference between the two models were found non-significant for all metrics (p>0.05). Reduction of cohort size showed no significant difference for all metrics (p>0.05). However, with the smallest cohort size, a few outlier structures were found. Conclusions: This study demonstrated a deep learning-based auto-segmentation model trained on curated clinical delineations which performs on par with a model trained on dedicated delineations, making it easier to develop multi-institutional auto-segmentation models. [ABSTRACT FROM AUTHOR]

Introduction:Minimizing the planning target volume (PTV) while ensuring sufficient target coverage during the entire respiratory cycle is essential for free-breathing radiotherapy of lung cancer. Different methods are used to incorporate the respiratory motion into the PTV. Material and methods:Fifteen patients were analyzed. Respiration can be included in the target delineation process creating a respiratory GTV, denoted iGTV. Alternatively, the respiratory amplitude (A) can be measured based on the 4D-CT and A can be incorporated in the margin expansion. The GTV expanded by A yielded GTV + resp, which was compared to iGTV in terms of overlap. Three methods for PTV generation were compared. PTVdel(delineated iGTV expanded to CTV plus PTV margin), PTVσ(GTV expanded to CTV and A was included as a random uncertainty in the CTV to PTV margin) and PTV∑(GTV expanded to CTV, succeeded by CTV linear expansion by A to CTV + resp, which was finally expanded to PTV∑). Results:Deformation of tumor and lymph nodes during respiration resulted in volume changes between the respiratory phases. The overlap between iGTV and GTV + resp showed that on average 7% of iGTV was outside the GTV + resp implying that GTV + resp did not capture the tumor during the full deformable respiration cycle. A comparison of the PTV volumes showed that PTVσwas smallest and PTVΣlargest for all patients. PTVσwas in mean 14% (31 cm3) smaller than PTVdel, while PTVdelwas 7% (20 cm3) smaller than PTVΣ. Conclusions:PTVσyields the smallest volumes but does not ensure coverage of tumor during the full respiratory motion due to tumor deformation. Incorporating the respiratory motion in the delineation (PTVdel) takes into account the entire respiratory cycle including deformation, but at the cost, however, of larger treatment volumes. PTVΣshould not be used, since it incorporates the disadvantages of both PTVdeland PTVσ. [ABSTRACT FROM AUTHOR]