Evaluation of Deep Learning Clinical Target Volumes Auto-Contouring for Magnetic Resonance Imaging-Guided Online Adaptive Treatment of Rectal Cancer.

Purpose: Segmentation of clinical target volumes (CTV) on medical images can be time-consuming and is prone to interobserver variation (IOV). This is a problem for online adaptive radiation therapy, where CTV segmentation must be performed every treatment fraction, leading to longer treatment times and logistic challenges. Deep learning (DL)-based auto-contouring has the potential to speed up CTV contouring, but its current clinical use is limited. One reason for this is that it can be time-consuming to verify the accuracy of CTV contours produced using auto-contouring, and there is a risk of bias being introduced. To be accepted by clinicians, auto-contouring must be trustworthy. Therefore, there is a need for a comprehensive commissioning framework when introducing DL-based auto-contouring in clinical practice. We present such a framework and apply it to an in-house developed DL model for auto-contouring of the CTV in rectal cancer patients treated with MRI-guided online adaptive radiation therapy. Methods and Materials: The framework for evaluating DL-based auto-contouring consisted of 3 steps: (1) Quantitative evaluation of the model's performance and comparison with IOV; (2) Expert observations and corrections; and (3) Evaluation of the impact on expected volumetric target coverage. These steps were performed on independent data sets. The framework was applied to an in-house trained nnU-Net model, using the data of 44 rectal cancer patients treated at our institution. Results: The framework established that the model's performance after expert corrections was comparable to IOV, and although the model introduced a bias, this had no relevant impact on clinical practice. Additionally, we found a substantial time gain without reducing quality as determined by volumetric target coverage. Conclusions: Our framework provides a comprehensive evaluation of the performance and clinical usability of target auto-contouring models. Based on the results, we conclude that the model is eligible for clinical use.

Online Adaptive MRI-Guided Radiotherapy for Primary Tumor and Lymph Node Boosting in Rectal Cancer.

The purpose of this study was to characterize the motion and define the required treatment margins of the pathological mesorectal lymph nodes (GTVln) for two online adaptive MRI-guided strategies for sequential boosting. Secondly, we determine the margins required for the primary gross tumor volume (GTVprim). Twenty-eight patients treated on a 1.5T MR-Linac were included in the study. On T2-weighted images for adaptation (MRIadapt) before and verification after irradiation (MRIpost) of five treatment fractions per patient, the GTVln and GTVprim were delineated. With online adaptive MRI-guided radiotherapy, daily plan adaptation can be performed through the use of two different strategies. In an adapt-to-shape (ATS) workflow the interfraction motion is effectively corrected by redelineation and the only relevant motion is intrafraction motion, while in an adapt-to-position (ATP) workflow the margin (for GTVln) is dominated by interfraction motion. The margin required for GTVprim will be identical to the ATS workflow, assuming each fraction would be perfectly matched on GTVprim. The intrafraction motion was calculated between MRIadapt and MRIpost for the GTVln and GTVprim separately. The interfraction motion of the GTVln was calculated with respect to the position of GTVprim, assuming each fraction would be perfectly matched on GTVprim. PTV margins were calculated for each strategy using the Van Herk recipe. For GTVln we randomly sampled the original dataset 20 times, with each subset containing a single randomly selected lymph node for each patient. The resulting margins for ATS ranged between 3 and 4 mm (LR), 3 and 5 mm (CC) and 5 and 6 mm (AP) based on the 20 randomly sampled datasets for GTVln. For ATP, the margins for GTVln were 10-12 mm in LR and AP and 16-19 mm in CC. The margins for ATS for GTVprim were 1.7 mm (LR), 4.7 mm (CC) and 3.2 mm anterior and 5.6 mm posterior. Daily delineation using ATS of both target volumes results in the smallest margins and is therefore recommended for safe dose escalation to the primary tumor and lymph nodes.

Effect of intrafraction adaptation on PTV margins for MRI guided online adaptive radiotherapy for rectal cancer.

PURPOSE: To determine PTV margins for intrafraction motion in MRI-guided online adaptive radiotherapy for rectal cancer and the potential benefit of performing a 2nd adaptation prior to irradiation. METHODS: Thirty patients with rectal cancer received radiotherapy on a 1.5 T MR-Linac. On T2-weighted images for adaptation (MRIadapt), verification prior to (MRIver) and after irradiation (MRIpost) of 5 treatment fractions per patient, the primary tumor GTV (GTVprim) and mesorectum CTV (CTVmeso) were delineated. The structures on MRIadapt were expanded to corresponding PTVs. We determined the required expansion margins such that on average over 5 fractions, 98% of CTVmeso and 95% of GTVprim on MRIpost was covered in 90% of the patients. Furthermore, we studied the benefit of an additional adaptation, just prior to irradiation, by evaluating the coverage between the structures on MRIver and MRIpost. A threshold to assess the need for a secondary adaptation was determined by considering the overlap between MRIadapt and MRIver. RESULTS: PTV margins for intrafraction motion without 2nd adaptation were 6.4 mm in the anterior direction and 4.0 mm in all other directions for CTVmeso and 5.0 mm isotropically for GTVprim. A 2nd adaptation, applied for all fractions where the motion between MRIadapt and MRIver exceeded 1 mm (36% of the fractions) would result in a reduction of the PTVmeso margin to 3.2 mm/2.0 mm. For PTVprim a margin reduction to 3.5 mm is feasible when a 2nd adaptation is performed in fractions where the motion exceeded 4 mm (17% of the fractions). CONCLUSION: We studied the potential benefit of intrafraction motion monitoring and a 2nd adaptation to reduce PTV margins in online adaptive MRIgRT in rectal cancer. Performing 2nd adaptations immediately after online replanning when motion exceeded 1 mm and 4 mm for CTVmeso and GTVprim respectively, could result in a 30-50% margin reduction with limited reduction of dose to the bowel.

Benchmarking daily adaptation using fully automated radiotherapy treatment plan optimization for rectal cancer.

Background/purpose: In daily plan adaptation the radiotherapy treatment plan is adjusted just prior to delivery. A simple approach is taking the planning objectives of the reference plan and directly applying these in re-optimization. Here we present a tested method to verify whether daily adaptation without tweaking of the objectives can maintain the plan quality throughout treatment. Materials/methods: For fifteen rectal cancer patients, automated treatment planning was used to generate plans mimicking manual reference plans on the planning scans. For 74 fraction scans (4-5 per patient) an automated plan and a daily adapted plan were generated, where the latter re-optimizes the reference plan objectives without any tweaking. To evaluate the robustness of the daily adaptation, the adapted plans were compared to the autoplanning plans. Results: Median differences between the autoplanning plans on the planning scans and the reference plans were between -1 and 0.2 Gy. The largest interquartile range (1 Gy) was seen for the Lumbar Skin D2%. For the daily scans the PTV D2% and D98% differences between autoplanning and adapted plans were within ± 0.7 Gy, with mean differences within ± 0.3 Gy. Positive differences indicate higher values were obtained using autoplanning. For the Bowelarea + Bladder and the Lumbar Skin the D2% and Dmean differences were all within ± 2.6 Gy, with mean differences between -0.9 and 0.1 Gy. Conclusion: Automated treatment planning can be used to benchmark daily adaptation techniques. The investigated adaptation workflow can robustly perform high quality adaptations without daily adjusting of the patient-specific planning objectives for rectal cancer radiotherapy.

Quantification of Esophageal Tumor Motion and Investigation of Different Image-Guided Correction Strategies.

PURPOSE: To accurately quantify esophageal tumor position variability and to optimize image guided correction strategies. MATERIAL AND METHODS: Esophageal cancer patients receiving chemoradiotherapy (41.4-50.4 Gy in 23-28 fractions combined with carboplatin plus paclitaxel) were included in a prospective cohort study (NCT02139488). Gold fiducial markers were inserted into the esophageal tumors during diagnostic endoscopic ultrasound. Four-dimensional (4D) planning computed tomography (CT) and daily 4D cone beam (CB) CT scans were acquired. Each CBCT was registered to the planning CT using different regions of interest (bone; 3D), and carina, diaphragm, clinical target volume (CTV), and fiducial markers (4D) for alignment and using the fiducial markers as the true tumor position. Subsequently, a planning target volume (PTV) margin accounting for residual uncertainties, including the average respiratory motion, was calculated for each of these registrations. RESULTS: Fifty-six patients with tumors located in the proximal (n = 1), mid (n = 7), or distal esophagus (n = 25) or at the gastroesophageal junction (n = 23) were included. The average peak-to-peak respiratory tumor motion was 0.20, 0.92, and 0.34 cm on the planning CT in left-right (LR), cranial-caudal (CC), and anterior-posterior (AP) directions, respectively. The required PTV margin with average motion amplitude, depending on the correction strategy used for image guidance, ranged from 0.8 cm to 1.0 cm, 1.1 cm to 1.6 cm, and 0.7 cm to 0.9 cm in LR, CC, and AP direction, respectively. A registration based on the CTV resulted in the smallest PTV margins (0.8, 1.1, and 0.7 cm in LR, CC, and AP direction, respectively). For bone registration the calculated PTV margins were 1.0, 1.3, and 0.7 cm in LR, CC, and AP directions, respectively. The registration based on the diaphragm increased PTV margins. CONCLUSIONS: Substantial and anisotropic position variability of esophageal tumors was observed during radiation therapy, and nonuniform margins should be considered. Cranial-caudal PTV margins need to be larger than those commonly used. Target positioning during image-guided radiotherapy could be improved with a CTV registration-based correction strategy.

Influence of EUS training and pathology interpretation on accuracy of EUS-guided fine needle aspiration of pancreatic masses.

BACKGROUND: Identification, staging, and fine needle aspiration of pancreatic mass lesions are probably the most technically demanding EUS skills. This study evaluated the effect of formal training on the diagnostic accuracy of EUS-guided fine needle aspiration (EUS-FNA) of pancreatic masses and the source of the variability in diagnostic accuracy between initial and later procedures. METHODS: Sixty-five patients with pancreatic masses underwent EUS-FNA between April 1998 (introduction of EUS-FNA) and August 1999, 20 of whom were examined by 3 endosonographers without prior experience with EUS-FNA. The initial experience of these 3 endosonographers (April to December 1998; group A patients), which included a formal training period of 2 months, and their later experience (January to August 1999; group B patients) were evaluated. Final diagnoses were determined by surgical pathology or clinical follow-up. All EUS-FNA samples were reviewed by 4 blinded pathologists to determine the contribution of pathologist interpretation to varying EUS-FNA accuracy. RESULTS: After a short training period, there was a significant improvement in EUS-FNA accuracy (33% vs. 91%; p = 0.004). After pathology review, good agreement was identified between original FNA interpretation and that on review (kappa = 0.78; 95% CI [0.5, 1.0]). There were differences between the mean cellularity score (2.8 vs. 1.8, p = 0.01) and mean number of passes (5.1 vs. 2.8, not significant) for correct versus incorrect FNA specimens. CONCLUSION: Significant improvements in EUS-FNA accuracy can be achieved with a short period of mentored training. EUS-FNA errors during the initial learning phase are primarily due to inadequate specimens. Interpretation of pancreatic EUS-FNA specimens remained consistent before and after training.

Initial experience with EUS-guided trucut needle biopsies of perigastric organs.

BACKGROUND: The aims of this study were to determine the feasibility, safety, and yield of a 19-gauge EUS-guided-trucut needle for obtaining biopsy specimens of perigastric organs. METHODS: The study was performed in swine under general anesthesia. EUS-guided trucut needle biopsy specimens were obtained from the spleen, liver, pancreas body, and left kidney. Biopsy specimens were assessed for size, fragmentation, and representation of the target organ. OBSERVATIONS: Twenty-eight biopsy specimens were obtained from the 4 target organs with two needles. Median biopsy length was 6 mm (spleen), 4 mm (liver), 6 mm (left kidney), and 2 mm (pancreas body). Of all the specimens, 75% to 100% had tissue representative of the target organ. EUS visualization of the needle was excellent and no complications were identified. CONCLUSIONS: Use of the trucut needle under EUS guidance to obtain biopsy specimens of perigastric organs appears safe and yields specimens that are representative of the target organ sampled. Further study of the utility and safety of this needle in humans is warranted.