Observer uncertainties of soft tissue‐based patient positioning in IGRT

Abstract Purpose There remain uncertainties due to inter‐ and intraobserver variability in soft‐tissue‐based patient positioning even with the use of image‐guided radiation therapy (IGRT). This study aimed to reveal observer uncertainties of soft‐tissue‐based patient positioning on cone‐beam computed tomography (CBCT) images for prostate cancer IGRT. Methods Twenty‐six patients (7–8 fractions/patient, total number of 204 fractions) who underwent IGRT for prostate cancer were selected. Six radiation therapists retrospectively measured prostate cancer location errors (PCLEs) of soft‐tissue‐based patient positioning between planning CT (pCT) and pretreatment CBCT (pre‐CBCT) images after automatic bone‐based registration. Observer uncertainties were evaluated based on residual errors, which denoted the differences between soft‐tissue and reference positioning errors. Reference positioning errors were obtained as PCLEs of contour‐based patient positioning between pCT and pre‐CBCT images. Intraobserver variations were obtained from the difference between the first and second soft‐tissue‐based patient positioning repeated by the same observer for each fraction. Systematic and random errors of inter‐ and intraobserver variations were calculated in anterior–posterior (AP), superior–inferior (SI), and left–right (LR) directions. Finally, clinical target volume (CTV)‐to‐planning target volume (PTV) margins were obtained from systematic and random errors of inter‐ and intraobserver variations in AP, SI, and LR directions. Results Interobserver variations in AP, SI, and LR directions were 0.9, 0.9, and 0.5 mm, respectively, for the systematic error, and 1.8, 2.2, and 1.1 mm, respectively, for random error. Intraobserver variations were <0.2 mm in all directions. CTV‐to‐PTV margins in AP, SI, and LR directions were 3.5, 3.8, and 2.1 mm, respectively. Conclusion Intraobserver variability was sufficiently small and would be negligible. However, uncertainties due to interobserver variability for soft‐tissue‐based patient positioning using CBCT images should be considered in CTV‐to‐PTV margins.


| INTRODUCTION
Prostate cancer is the most frequently diagnosed cancer in males. Approximately 1.3 million new cases of prostate cancer and 359 000 associated deaths were reported worldwide in 2018. 1 Common approaches for treating localized prostate cancer include active surveillance, radical prostatectomy, radiotherapy, and hormonal therapy. 2 Since external beam radiation therapy (a minimal dose of 72 Gy) showed similar biochemical relapse-free survival rates of localized prostate cancer as radical prostatectomy, 3 radiation therapy has an advantage for elderly people, who cannot undergo surgery due to complications. 2,3 In the current radiation therapy for prostate cancer, imageguided radiation therapy (IGRT) with cone-beam computed tomography (CBCT) images has been commonly used in clinical practice to increase the accuracy of patient positioning. 4 Zelefsky et al. reported that intensity modulated radiotherapy (IMRT) with image-guided patient positioning (IGPP) improved prostatespecific antigen (PSA) outcomes and toxicities of organs at risk (OAR). 5 Soft-tissue-based patient positioning with CBCT images [6][7][8][9] and intraprostatic fiducial marker-based patient positioning [10][11][12] have been used for more accurate target-based-patient positioning (TBPP). Soft-tissue-based patient positioning with CBCT images is a noninvasive approach and has the advantage of providing soft-tissue information such as circumstances of targets and critical organs. However, there are uncertainties due to inter-and intraobserver variability, [6][7][8][9] which may influence clinical outcomes and OAR toxicities.
Several studies have assessed inter-and intraobserver variability in soft-tissue-based patient positioning. [6][7][8][9] Some studies have investigated the accuracy of soft-tissue-based patient positioning with CBCT images compared to patient positioning based on fiducial markers for patients undergoing external beam radiotherapy of the prostate. 6,7 Jereczek-Fossa et al. used online CBCT positioning performed by a radiation oncologist immediately prior to treatment as reference for manual soft-tissuebased patient positioning reviewed by observers. 8 To our knowledge, no observer studies have evaluated inter-and intraobserver variability in soft-tissue-based patient positioning using prostate contours on CBCT images.
We hypothesized that the uncertainties due to inter-and intraobserver variability for soft-tissue-based patient positioning using CBCT images would not be negligible. In this observer study,  The uncertainties for soft-tissue-based patient positioning were evaluated by inter-and intraobserver variations. The interobserver variations were obtained from residual errors, which denote differences between soft-tissue positioning errors and reference positioning errors. The evaluation scheme of residual errors by each observer for a fraction is illustrated in Fig. 1. The reference positioning errors were obtained from prostate cancer location errors (PCLEs) of contour-based patient positioning between planning CT (pCT) and pretreatment CBCT (pre-CBCT) images. The PCLEs of contour-based patient positioning indicate the centroid distance of prostate contours on pCT and pre-CBCT images. The soft-tissue positioning errors were measured from the PCLEs of soft-tissue-based patient positioning in the observer study, which were performed between pCT and pre-CBCT images after an automatic bone-based registration by six observers. Then, the intraobserver variations were evaluated from the differences between the first and second soft-tissuebased patient positioning repeated by the same observer for each fraction. Finally, CTV-to-PTV margins (hereafter PTV margins) were calculated from the systematic and random errors of the inter-and intraobserver variations in AP, SI, and LR directions.

2.F | Interobserver variations
Inter-observer variations were evaluated from the residual errors, which denoted the differences between the soft-tissue positioning errors and reference positioning errors for each fraction. The systematic error (ɛ inter ) and random error (σ inter ) for interobserver variations were calculated from the root mean square (RMS) of the residual errors by N observers, respectively, as F I G . 1. The evaluation scheme of residual errors by each observer on pre-cone-beam computed tomography image of a fraction. and where N is the number of observers. ɛ inter;j and σ inter;j represent systematic and random errors of the residual errors for an observer j, respectively. The systematic error (ɛ inter;j ) and random error (σ inter;j ) for an observer j were given by respectively,where n is number of patients. m inter;i;j and m inter;j represent the mean residual error of a patient i by an observer j and the mean residual error of all patients by the observer j, respectively. σ inter;i;j represents the SD of the residual error of a patient i by an observer j. m inter;i;j and m inter;j are given by where F is the number of fractions. d inter;i;j;k represents the residual error at a fraction k of a patient i by an observer j, which denotes the difference between a soft-tissue positioning error and reference positioning error at each fraction.

2.G | Intraobserver variations
To further explore the effect of intraobserver variations in soft-tissue-based patient positioning on prostate IGRT, each observer repeated the soft-tissue-based patient positioning process for the same cases 3 months later. An intraobserver variation was evaluated as the difference between the first and second soft-tissue-based patient positionings repeated by an observer at each fraction of a patient. Five observers participated in the observer study with ten patients. The systematic error (ɛ intra ) and random error (σ intra ) for intraobserver variations were calculated from the RMS of the intraobserver errors by N observers, respectively, as and where N is the number of observers. ɛ intra;j and σ intra;j represent systematic and random errors of the intraobserver errors for an observer j, respectively. The systematic error (ɛ intra;j ) and random error (σ intra;j ) for an observer j were given by where n is number of patients. m intra;i;j and m intra;j represent the mean intraobserver error of a patient i by an observer j and mean intraobserver error of all patients by the observer j, respectively. σ intra;i;j represents SD of the intraobserver errors of a patient i by the observer j. m intra;i;j and m intra;j are given by where F is the number of fractions. d intra;i;j;k represents the intraobserver error at a fraction k of a patient i by an observer j, which denotes the difference between the first and second soft-tissuebased patient positioning repeated by an observer at each fraction of a patient.

2.H | PTV margin calculations
Planning target volume margins were calculated from the systematic and random errors of interobserver and/or intraobserver variations using the van Herk's margin formula, 14 as follows: This formula was derived based on a dose-population histogram to deliver at least 95% of a prescribed dose to 90% of a patient population.   The RMSs of the random errors for interobserver variations by the six observers were 1.8, 2.2, and 1.1 mm in AP, SI, and LR directions.

3.C | Intraobserver variations
The systematic and random errors of intraobserver variations calculated as the difference between the first and second soft-tissuebased patient positioning repeated by each observer for each fraction were <0.2 mm in AP, SI, and LR directions. These dose evaluation indices were defined based on ICRU report 83 16 as follows:

3.D | PTV margins
and where V PTV is the volume of PTV, and V PD98% is the volume receiving 98% of the prescribed dose. D2, D98, and D50 mean the minimum dose that covers 2%, 98%, and 50% of the PTV, respectively. Dose | 79 high rate of symptomatic infection with fiducial marker implantation and concluded that noninvasive approaches for prostate IGRT, such as CBCT, should be considered. 21 The present study has the advantage that the uncertainties of soft-tissue-based patient positioning were evaluated on pre-CBCT images without fiducial markers.
If same therapists perform soft-tissue-based patient positioning during the entire course of the treatment, the intraobserver variability would be <0.2 mm, which may be negligible. However, there could remain an uncertainty specific to each observer. Therefore, the uncertainty of the observer should be taken into account for calculating PTV margins. Nevertheless, the observer-specific uncertainties can also be evaluated by comparing with the contour-based patient positioning proposed in this study.
Recently, a commercial on-board CBCT equipped with an iterative image reconstruction function has been available. The iterative image reconstruction algorithm can improve the CBCT image quality by reducing noise and artifacts. 22

| CONCLUSION S
In conclusion, the results of the present study revealed the uncertainties for soft-tissue-based patient positioning by inter-and intraobserver variability on pre-CBCT images in IGRT for prostate cancer. Interobserver variations in AP, SI, and LR directions were 0.9, 0.9, and 0.5 mm, respectively, for systematic error and 1.8, 2.2, and 1.1 mm, respectively, for random error. The systematic and random errors of intraobserver variations were <0.2 mm in all directions. The PTV margins considering both interobserver and intraobserver variations in AP, SI, and LR directions were 3.5, 3.8, and 2.1 mm, respectively. Consequently, while intraobserver variability was sufficiently small and would be negligible, uncertainties due to interobserver variability for soft-tissue-based patient positioning using CBCT images should be considered in CTV-to-PTV margins.

ACKNOWLEDG MENTS
This work was supported by JSPS KAKENHI Grant Number JP19K16803. The authors are grateful to all members of Arimura Laboratory (http://web.shs.kyushu-u.ac.jp/~arimura), whose comments contributed significantly to this study. We would also like to thank Editage (www.editage.com) for English language editing.

CONFLI CT OF INTEREST
No conflict of interest.