Correlation between 3D scanner image and MRI for tracking volume changes in head and neck cancer patients

Abstract Introduction We investigated the correlation between optical surface imaging using a three‐dimensional (3D) scanner and magnetic resonance imaging (MRI) for suggesting feasibility in the clinical process of tracking volume changes in head and neck patients during radiation treatment. Methods Ten patients were divided into two groups depending on the location of their tumor (i.e., right or left side). With weekly imaging data, the change in volume based on MRI was evaluated during the treatment course. Four volumes of interest (VOIs) were calculated on the 3D surface image of the facial and cervical areas using an optical 3D scanner, and the correlation between volumetric parameters were analyzed. Results The target volume changed significantly overall for both groups. The changes parotid volume reduced by up to 3.8% and 28.0% for groups A (right side) and B (left side), respectively. In Group A, VOI 1 on the facial area and VOI 3 on the cervical area decreased gradually during the treatment course by up to 3.3% and 10.7%, respectively. In Group B, only VOI 4 decreased gradually during the treatment course and reduced by up to 9.2%. In group A, the change in target volume correlated strongly with right‐side parotid, VOI 1, and VOI 3, respectively. The parotid also showed strong correlations with VOIs (P < 0.01). The weight loss was strongly correlated with either PTV or parotid without statistical significance (P > 0.05). In group B (left side), the change in target volume correlated strongly with each volumetric parameter, including weight loss. For individual patient, PTV showed more correlation with VOIs on the cervical area than VOIs on the facial area. Conclusions An optical 3D scanner can be applied to track changes in volume without radiation exposure during treatment and the optical surface image correlated with MRI.


| INTRODUCTION
Head and neck (H&N) cancers are often treated using radiation therapy. 1 Advanced radiation techniques such as intensity-modulated radiation therapy (IMRT) and volumetric modulated arc therapy (VMAT) can deliver highly conformal doses to the target volumes while sparing the adjacent normal tissue. 2 During the course of treatment, patients exhibit significant anatomical changes related to tumors response and weight loss. [3][4][5] Anatomical changes in both tumor and normal tissues can result in underdosage or dose inhomogeneity for the target and overdosage for the organs at risk (OARs). Wang et al. reported that the parotid and submandibular glands (SMG) shrunk during radiation therapy. 6 Shreerang et al. investigated the weekly volume changes in target volumes and OARs using repeated computed tomography (CT) scans. 7 The most significant volumetric changes and dosimetric alterations in tumor volumes and OARs during a course of chemotherapy with IMRT occurred by week 2 of radiotherapy.
Adaptive radiotherapy (ART) is a novel approach in which the treatment plan is adjusted during the course of treatment to account for anatomical changes and improve therapeutic gain. Recently, clinical outcomes have been reported for H&N ART. [8][9][10] Schwartz et al. performed a trial investigating ART for oropharyngeal squamous cell carcinoma in order to examine toxicity and survival outcomes. 11,12 They concluded that ART can provide a dosimetric benefit with only one or two re-planning interruptions and that properly timed re-planning delivers achievable dosimetric improvement a majority of the time. ART is directly associated with image-guided radiotherapy (IGRT) using inroom megavoltage (MV) CT, CT-on-rails, or cone-beam CTs (CBCT) obtained prior to daily treatment. 13 Currently, adaptive magnetic resonance image-guided radiation therapy (MR-IGRT) has been reported and clinically implemented. 14,15 Acharay et al. reported the first successful clinical application of online adaptive MR-IGRT. 15 They concluded that the trigger for re-optimization should be considered with changes in anatomy because not all patients derive the same benefit from ART. However, there are several limitations to the high cost of installation, the operational experience required by the operator, and in the case of IGRT, additional radiation exposure.
Optical 3D surface scanning systems have been applied to radiation therapy because of their advantages of low cost, accuracy, speed, and flexible handling without unnecessary radiation exposure.
All studies on 3D scanning systems only analyzed the accuracy of the patient positioning. [16][17][18][19] There was no study to volumetric analysis using an optical 3D scanner. In this study, we investigated the feasibility of using optical 3D scanner surface imaging for predicting volume changes during the treatment course and evaluated the correlation between MRI and these images.

2.A | Patient selection and MRI imaging
In this study, ten prospective patients among the H&N cancer patients in our institution were randomly enrolled. It was approved by the institutional review board of Seoul National University College of Medicine/Seoul National University Hospital (IRB No.1505-055-671) for this study. All patients in this manuscript have signed informed consent forms and agreed to publish these case details.
Patients were divided into two groups depending on the location of their tumor (i.e., right or left side).  surface image were applied to other 3D surface images for one patient. We performed the rigid image registration using the polylines defined. Figure 3 shows a sample of 3D surface image registration.
The color map indicates the difference between the registered 3D surface images. In each patient group parotid and VOIs on the same side as the location of the tumor were evaluated. The percent weight loss was also evaluated during the treatment course. All changes were evaluated according to the differences in values obtained from weekly images and that obtained from the images of the first week.
Spearman's correlation analysis was performed to investigate the relationship among the above calculated variables. 22

3.B | Correlation analysis
Spearman's correlation coefficients (r) with absolute values of averaged volume were calculated for correlation analysis between the volumetric parameters, as summarized in Table 3 Table 4. The change in PTV correlated strongly with VOI 1 and VOI 3 in three patients in Group A. One patient only showed the strong correlation between PTV and VOI 3. As shown in Fig. 4, the 5th patient in Group A has no correlation between PTV and VOIs. The PTV also showed more correlation with VOI 4 on the cervical area than VOI 2 on the facial area in Group B.
T A B L E 2 Normalized mean value with standard deviation of the relative change calculated from each patient during the treatment course compared to the first week.

| DISCUSSION
We investigated the feasibility of optical 3D scanner surface imaging for predicting the volume changes of head and neck cancer patients during treatment. Ten patients with different types of H&N cancer were involved in this feasibility study. It is well-recognized that all patients are individuals and respond differently to radiation therapy, even if they have the same type of cancer in the same location. 23,24 Therefore, to achieve these individual therapeutic benefits, we considered the ART treatment plan approach that reflects each anatomical change. However, the ART process is time-and resource-  predictive factors using weekly CT scans and reported that parotid gland shrinkage during IMRT was not linear. 31 Accordingly, in this study, the averaged patient data, for patients with the same-side tumor, were adopted rather than patient-specific analysis in order to investigate the feasibility of using optical 3D scanner surface imaging to predict volume changes. Based on the correlations between the MR and 3D surface images, the volume changes of PTV and the parotid could be predicted with optical 3D scanner surface imaging.
Patient weight loss may be an indicator of adaptive planning. Lee et al. reported that major weight loss above 6% needs adaptive planning. 32 Duma et al. reported no correlation between weight loss and volume changes of regions of interest (ROIs), except for a strong positive correlation with shrinkage of the parotid. 33 In this study, weight loss showed a strong positive correlation with the volume changes of ROIs. After the final week of treatment, however, the weight loss was above 4-5%, and it was not a local reduction in the volume of the H&N region. The optical 3D scanner surface images, which were for local regional volume, also showed a strong positive

| CONCLUSION
In this study, we verified that the volume changes undergone by H&N cancer patients during treatment can be detected by surface imaging using optical 3D scanners. The optical 3D scanner could be applied to track changes in volume without additional radiation exposure.

CONFLI CT OF INTEREST
The authors declare no conflict of interest.