ESTRO 38 Abstract book

S219 ESTRO 38

Conclusion MidV determination based on DIR is accurate. A high correlation was found in MidV-P for the evaluated methods and the differences in tumour motion were small. Different results were found for patients with poor soft tissue contrast between tumor and surrounding tissue. For these patients, the automatic approach can reduce the interobserver variation and can obtain more consistent results by using an enlarged focus region. The automated method is now clinically used to determine tumour motion and MidV-P. OC-0421 Stage migration in planning PET/CT scans in lung cancer patients referred to radiochemotherapy S. Kivistik 1 , L. Randle 1 , M. Vardja 1 , A. Aasa 1 , J. Jaal 1 1 Tartu University Hospital, Dept of Radiotherapy and Oncological Therapy, Tartu, Estonia Purpose or Objective With recent advances in immune-oncology and the introduction of checkpoint inhibitors into clinical practice, the treatment landscape of lung cancer has changed dramatically. Immune checkpoint inhibitors can be used as adjuvant therapy after completion of radiochemotherapy in stage III non-small cell lung cancer (nsclc) and as systemic treatments in stage IV nsclc as well as small-cell lung cancer (sclc). Cancer treatment costs are rising, therefore, it is extremely important to assure that patients receive most effective treatment at the right time based on the exact stage of the disease. In our institution, 18-fluoro-deoxyglucose positron emission tomography ( FDG - PET ) is mostly used for patients that are considered to have stage III disease based on previous computed tomography (CT) findings and are consequently referred to receive radiochemotherapy. The aim of this study was to assess the impact of radiotherapy planning PET-CT on stage migration in these patients. Material and Methods The stage migration after PET-CT was evaluated in 82 patients referred to our department to receive radiochemotherapy. All patients were previously assessed in multidisciplinary team meetings (including radiologist) and had stage III disease based on available CT scans. Results Out of 82 patients, 65 had nsclc (79%) and 17 sclc (21%). After planning PET-CT, metastatic stage IV disease was detected in 31 lung cancer patients (38%). The stage migration was seen in 22 patients (34%) with nsclc and in 9 patients (53%) with sclc. In nsclc, metastases were most frequently found in bones (55%), lymph nodes (55%), lungs (32%), pleura (23%), adrenal glands (18%) and liver (9%). In sclc, metastases were most frequently detected in lymph nodes (67%), bones (22%), lungs (11%), pleura (11%) and liver (11%). Radiochemotherapy was given to 45 patients (88%) with locoregional disease. Some patients did not receive radiochemotherapy due to refusal (n=1), treatment in another hospital (n=1), or not acceptable mean lung dose (n=4). Conclusion In the situation, where PET-CT is not routinely used for staging of all lung cancer patients and it is mainly utilized for radiotherapy planning, significant stage migration can be detected. Big proportion (38%) of lung cancer patients that were previously considered to have stage III disease showed metastatic lesions after planning PET-CT. This study confirms that proper selection of patients with PET- CT is mandatory to guarantee optimal use of cancer care resources.

several manual actions have to be performed which are time consuming, and interobserver variation can occur. An automatic procedure can reduce the workload and variability. Aim of this study was to compare the results of the manual approach with the automatic approach using the Deformable Image Registration (DIR) module in RayStation. Material and Methods For 23 patients, the automatic procedure was applied and compared to the manually determined MidV. The tumour was delineated at one of the 10 reconstructed phases of the 4D CT-scan. This delineation is then copied to the other phases and translated manually to the correct position. For the automatic approach, all phases were registered deformably to the 0% phase scan using the Anaconda algorithm. To improve the DIR in the close proximity of the tumour, a 2 cm expansion was used as focus region. After visually inspecting the DIRs (figure 1), the delineated tumour was propagated automatically to the other phases. The center of mass position of the translated or propagated contour was used for calculating the MidV-P and tumour motion.

Results For all patients, the DIRs were considered acceptable for calculating the MidV-P and tumour motion. The difference in MidV-P was 0.0 ±0.5 mm in LR direction, 0.4 ±1.0 mm in CC direction and 0.1 ±0.6 mm in AP direction (mean ±SD). For 22/23 patients the maximum difference in position was 2 mm and a high correlation was found in all directions. For 4/23 patients a difference in tumour motion larger than 2 mm was observed (figure 2). The mean difference in tumour motion was 0.6 ±0.8 mm in LR direction, 0.1 ±1.4 mm in CC direction and 0.2 ±0.9 mm in AP direction (mean ±SD). These variations are similar to the interobserver variation for the manual procedure. For one patient, manual determination of the MidV-P and tumour motion was difficult due to poor soft tissue contrast between tumour and surrounding tissue. This resulted in a difference in MidV-P in the CC direction of 3.6 mm and a difference in tumour motion in LR and CC direction of 2.4 and 4.4 mm respectively. By using the automatic approach with the enlarged focus region, the surrounding vessels are included in this region to obtain more consistent results compared to the manual approach.

Poster Viewing : Poster viewing 8: TP Developments

PV-0422 Consequences of respiratory motion variability in lung 4DMRI datasets