Session Item

The benefits of adaptive radiation treatment at MR-Linacs come at the cost of tedious and repetitive expert (re)contouring. The aim of this work was to compare AI-based methods for the autosegmentation of organs at risk (OARs), i.e., rectum and bladder, exploiting prior knowledge available at the irradiation stage in MR-guided radiation therapy (MRgRT) of prostate cancer patients. Two approaches were tested: patient-specific (PS) U-Net models and networks predicting deformation vector fields (DVFN).

Patients from two institutes operating 0.35 T MR-Linacs were included in the study. Cohort 1 (C1), comprising 73 planning images with manual delineations, was used for training of the benchmark U-Net model (BM). The second cohort (C2) included 19 patients with expert-contoured MRIs. One planning MRI and 5-33 fraction MRIs were available for each C2 patient (19 planning, 240 fraction images). PS models were generated for each patient by fine-tuning the BM using a planning image and validating it with the corresponding fraction data. 10 C2 patients were selected randomly for the PS hyperparameter optimization. The DVFN architecture was based on a U-Net with a spatial transformer layer and aimed to predict deformation fields between the planning and fraction images. Training was performed on the same 10 C2 patients (10 planning and 120 fraction images). The loss function was based on the L2 image similarity and the multi-scale dice similarity coefficient (DSC) between structures of interest. All models shared the same test set, i.e., fraction data from the remaining 9 C2 patients. The models were evaluated with DSC, the average (HDavg) and the 95th percentile (HD95) Hausdorff distance.

Table 1 shows the evaluation of PS and DVFN models and compares them to the BM. The PS network with DSC for bladder/rectum of 0.93/0.90 performs better than both the DVFN (0.79/0.77) and the BM (0.91/0.87). The same was observed for HDs. Figure 1 shows slices that illustrate the performance of the methods under consideration. The PS approach can correct pronounced BM mistakes and determine rectum ends properly due to the inclusion of planning knowledge. The main challenge of the PS method was that it favors the planning shape while the DVFN had difficulties to predict large deformations.

The methods presented in this work were developed for data from fractionated MRgRT, where prior knowledge from the planning phase can be used at the time of irradiation. DVFN appears to predict limited deformations and is therefore less suitable for organs undergoing substantial volume changes. Conversely, PS networks on average improve segmentation compared to the BM. PS is particularly beneficial for patients with unusual anatomies and helps to adjust the top and the bottom ends of the rectum. However, it could propagate errors in initial contouring and over-favour the planning shape. Therefore, PS and BM should be used as complementary methods.

Work funded by the Wilhelm Sander-Stiftung