20 November 2014 in International Journal of Computer Assisted Radiology and Surgery
1 July 2021 in Cancer Research
Introduction: Tumor Treating Fields (TTFields) is a treatment modality for glioblastoma multiforme (GBM) and other malignant tumors, utilizing low-intensity, intermediate frequency (100-300kHz) alternating electric field delivered through noninvasive transducer arrays. TTFields are currently being tested in a phase III clinical trial for the treatment of advanced Non-Small Cell Lung Cancer (NSCLC). Previous studies have demonstrated that the treatment efficacy increases when the dose of TTFields delivered to the tumor is at least 1 V/cm. Thus, personalized simulations to estimate the dose are performed as part of the patient treatment planning. For NSCLC treatment, these simulations require the segmentation of torso tissues. We are currently developing a semi-automatic software for the segmentation of these tissues. Achieving highly accurate segmentation is time consuming and not always necessary. Therefore, in this study we assess the level of segmentation accuracy that is required for TTFields dose estimation. Methods: We introduce a novel pipeline for evaluating the sensitivity of TTFields treatment planning errors in the segmentation of torso tissues. To this end, we incorporated an expert-segmented torso volume and artificially placed NSCLC tumors of different sizes in the vicinity of the tested torso tissue. We employed relaxations to the segmentation algorithm and derived the TTFields distribution with and without the simulated segmentation errors. We then computed the treatment planning error as the per-voxel average deviation of the TTFields dose computed over the erroneous segmentation from the reference one. The dose was calculated over the simulated tumor location and over the entire lung volume. Results: We have evaluated our sensitivity analysis pipeline on 9 different test cases. Based on previous analysis, we have set our tolerable target error to be 0.1 V/cm. Our results show that 6 out of the 9 tests were within the target, whereas the other 3 exceeded the target, requiring algorithmic refinements. Conclusions: We have presented a method for estimating the sensitivity of lung cancer TTFields treatment planning to segmentation errors. This method is used to evaluate the required accuracy of the semi-automatic torso segmentation tool that we are currently developing. Our method can be further extended to other types of tumors in different regions of the body. Citation Format: Hadas Ben Atya, Oren Peles, Reuben Shamir, Zeev Bomzon. Lung cancer TTFields treatment planning sensitivity to errors in torso segmentation [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 3070.