A pilot study of a new method for visualizing three-dimensional dose distribution on skin surface images to assess radiation dermatitis

This study aimed to develop an intuitive system capable of objectively analyzing the direct correlation between radiation dose distribution and DR, by visualizing the absolute skin dose calculated by the TPS on a skin surface image acquired with a 3D camera. To our knowledge, there have been no similar publications on DR in the field of radiation oncology.

Heterogeneous dose distribution and partial hotspots are among the strongest predictors of DR9.16. However, dose predictors, such as mean doses and the dose-volume parameter9,16,21, are not sufficient to explain the RD pattern as a function of the heterogeneous dose distribution within the same irradiation field. To address this issue, we have developed this new, intuitive radiation dose-toxicity assessment system, by which the 3DSD can be extracted from the TPS dose distribution and mapped to the RD image. Using 82 3DSSIs from 19 patients who underwent radiotherapy, we verified that this new method is very useful in evaluating a direct correlation between the cutaneous dose distribution and the DR pattern.

As 3D cameras provide both photographs and 3D depth data of the body, they are suitable for both visual inspection and matching RD with 3D dose distribution. A 2D photograph is difficult to match with a 3D dose distribution because there is no angle and distance information. Therefore, we used 3D-based recording using skin surface data measured by a depth-sensing technique. We used an affordable and portable 3D camera that could easily create a highly accurate and reliable 3D skin surface model. Recently, some studies have reported the use of 3D cameras to measure various skin changes caused by radiotherapy22.23. These have shown that 3D cameras are superior to conventional photographs in the detailed assessment of the onset and development of skin changes, and in the objective measurement and documentation of DR. Currently, research on 3D cameras aims to objectively document DR. In this context, the RaTES matches and visualizes the dose distribution to 3DSSI as a first step towards the development of a comprehensive RD assessment tool. Our new approach can be used for the accurate prediction and prevention of RD based on the dose distribution of TPS.

Commercial TPS has limited accuracy in calculating skin surface doses. The accuracy of TPS skin dose calculation depends on the dose calculation algorithm and treatment technique and is therefore different for each facility.17,24,25. The variability of the dermal dose calculation suggests that it is necessary to check the correlation of the dermal dose calculated from the TPS of each establishment to also apply the dosimetric indices – or normal tissue complication probabilities (NTCP) – related to the seriousness of DR. The RaTES introduced in this study can be easily applied in institutions, and the contribution of the dose distribution to the occurrence of RD can be accurately assessed by matching the RD image with the skin dose distribution calculated by the TPS. institutional. This approach may also provide an opportunity to establish institution-optimized skin dose-volume constraints for treatment planning.

The advantage of this approach is a detailed dose distribution map for the degree of RD. Combined 3DSD with 3DSSI distribution showed that a heterogeneous dose distribution results in severe toxicity at hotspots compared to mild symptoms in low dose areas (Fig. 1). Additionally, dose information visualized on skin surface images can help identify non-dosimetric factors that affect the severity of DR. The radiation dose resulting from DR, such as hyperpigmentation, erythema, and moist desquamation, can vary greatly between individuals. Some studies have shown that concurrent chemotherapy, psoriasis, and smoking history are associated with the occurrence of DR21.26. The information obtained can help identify these clinical risk factors for DR.

Predicting and estimating the radiation dose-toxicity relationship is important for physicians and patients to make informed decisions, as DR affects patients’ quality of life27. Currently, dosimetric factors, such as dose-volume point, mean doses, and dose-volume histograms, predict DR. In IMRT/VMAT with a heterogeneous dose distribution, this approach is not effective in demonstrating a direct correlation between dose distribution and DR. If the contribution of the dose distribution pattern in TPS to RD can be objectively assessed, it is possible to predict and manage RD based on the patient’s treatment plan and provide appropriate patient guidance. Even if adjustment of the treatment plan is necessary due to severe DR, the dose distribution can be readjusted because the TPS dose and the location at which the DR is likely to occur can be easily identified. This study provided a basis for establishing a direct correlation between the skin dose distribution and the RD image.

Our results can be applied in various clinical contexts. Bernier et al. presented revised comprehensive consensus guidelines for DR in patients with head and neck cancer receiving epidermal growth factor receptor inhibitors in combination with radiation therapy and defined the degree of wet desquamation in percent in radiation fields28. These results allow the clinical application of these grading guidelines, as the RD image and the dose distribution can be matched and the irradiation area can be defined as the area receiving a specific dose.

Stereotactic radiosurgery, hypofractionated radiotherapy and particle therapy have recently been introduced into clinical practice. Establishing a predictive DR model for a new treatment technique is a challenge. Our approach allows an easy and simple assessment of the correlation between dose distribution and DR. This can help identify dosimetric factors associated with the occurrence of DR before establishing a toxicity model for a new treatment technique. In particle therapy, such as proton therapy, the skin may receive a relatively high dose compared to photon therapy, and the calculation of skin dose in TPS may be incorrect due to uncertainty of relative biological effectiveness (RBE). )9,29,30. RBE depends not only on linear energy transfer, but also on dose and tissue type. Differences in EBR lead to differences between the dose of TPS and the dose delivered to the patient. In vivo data is needed to clinically identify EBR differences31. The proposed RaTES provides the visualized information of the skin dose distribution corresponding to the RD image of the patient, which is useful for verifying the dose and the biological effect on the skin related to the occurrence of RD in the therapy. particles. Moreover, the difference in EBR can be indirectly confirmed by a cross-comparison with the dosimetric factors related to the DR of photon therapy.

The present study is the first step towards the development of comprehensive and automated assessment tools for DR and has a number of potential applications. In the future, the accuracy and reliability of predictive models could be improved by implementing an automated system that can assess RD rank and generate an NTCP tailored to the specific facility and treatment technique. The 3D camera can measure both RD and various skin changes, such as psoriasis, swelling and fibrosis22. Some studies have reported psoriasis as the only clinical predictor of DR21. We hope to develop an RD prediction system that takes into account both patient-related factors and dosimetric characteristics using a single 3D image acquisition.

There were some limitations to this study. First, the 3DSD and 3DSSI were matched using a rigid image registration algorithm. We used RD images acquired in the treatment position to minimize registration error. Further research on deformable image registration is needed to allow accurate matching, even if different from treatment position, because DR increases even after radiation therapy ends. Second, in the case of two or more plans created during the treatment period for the same treatment site, the doses were distorted and summed to calculate the skin dose accurately. However, changes in dose distribution due to anatomical changes occurring during treatment have not been taken into account. Third, our study focused on a methodology to match and visualize the skin dose on the patient’s skin image, and the spatial accuracy of 3DSSI and 3DSD was calculated as the mean registration deviation. However, the concordance between the RD pattern and the dose distribution was qualitatively assessed. Since the extent of occurrence of RD is visual information and not quantitative information (segmented data), it cannot be analyzed with similarity coefficients or statistical methods. Further study is needed to segment the RD area from the photograph. Finally, our results were based on a limited number of patients from a single institution, which may affect the robustness of the system and hinder the generalizability of our results. In particular, further studies are needed with larger numbers of patients in a prospective design to investigate the correlation between cutaneous dose distribution and DR using the proposed study approach.

This study visualized the dose distribution of TPS on a skin surface image of DR obtained using a 3D camera in patients undergoing radiotherapy. We have demonstrated that this new method can intuitively and objectively assess the correlation between skin dose distribution and DR directly, reliably and easily. This approach can also aid in the prediction and management of DR caused by heterogeneous dose distribution in IMRT/VMAT. This study forms the basis for the development and validation of a more accurate model of facility-specific normal tissue complication probability for DR, and may also help physicians make treatment decisions based on detailed evidence. However, the results should be interpreted with caution given the limitations of this study.

Comments are closed.