A generative adversarial network‐based (GAN‐based) architecture for automatic fiducial marker detection in prostate MRI‐only radiotherapy simulation images

Abstract

Purpose Clinical sites utilizing magnetic resonance imaging (MRI)‐only simulation for prostate radiotherapy planning typically use fiducial markers for pretreatment patient positioning and alignment. Fiducial markers appear as small signal voids in MRI images and are often difficult to discern. Existing clinical methods for fiducial marker localization require multiple MRI sequences and/or manual interaction and specialized expertise. In this study, we develop a robust method for automatic fiducial marker detection in prostate MRI simulation images and quantify the pretreatment alignment accuracy using automatically detected fiducial markers in MRI. Methods and Materials In this study, a deep learning‐based algorithm was used to convert MRI images into labeled fiducial marker volumes. Seventy‐seven prostate cancer patients who received marker implantation prior to MRI and CT simulation imaging were selected for this study. Multiple‐Echo T1‐VIBE MRI images were acquired, and images were stratified (at the patient level) based on the presence of intraprostatic calcifications. Ground truth (GT) contours were defined by an expert on MRI using CT images. Training was done using the pix2pix generative adversarial network (GAN) image‐to‐image translation package and model testing was done using fivefold cross validation. For performance comparison, an experienced medical dosimetrist and a medical physicist each manually contoured fiducial markers in MRI images. The percent of correct detections and F1 classification scores are reported for markers detected using the automatic detection algorithm and human observers. The patient positioning errors were quantified by calculating the target registration errors (TREs) from fiducial marker driven rigid registration between MRI and CBCT images. Target registration errors were quantified for fiducial marker contours defined on MRI by the automatic detection algorithm and the two expert human observers. Results Ninety‐six percent of implanted fiducial markers were correctly identified using the automatic detection algorithm. Two expert raters correctly identified 97% and 96% of fiducial markers, respectively. The F1 classification score was 0.68, 0.75, and 0.72 for the automatic detection algorithm and two human raters, respectively. The main source of false discoveries was intraprostatic calcifications. The mean TRE differences between alignments from automatic detection algorithm and human detected markers and GT were <1 mm. Conclusions We have developed a deep learning‐based approach to automatically detect fiducial markers in MRI‐only simulation images in a clinically representative patient cohort. The automatic detection algorithm‐predicted markers can allow for patient setup with similar accuracy to independent human observers.

End‐to‐end validation of fiducial tracking accuracy in robotic radiosurgery using MRI‐only simulation imaging

“…A limitation of this study is that it relies on the assumption that fiducial markers can be identified with good accuracy in MRI images. Much work has been done on the selection of optimal MRI sequences, optimal fiducial markers and detection methods for fiducial markers in MRI 10,18,19,35–37 . The goal of many of these studies was to identify fiducial markers for identification at simulation with the eventual goal of voxel burning the fiducial locations in DRR images for pretreatment positioning and alignment using C‐arm linacs.…”

“…Robotic radiosurgery platforms use fiducial tracking algorithms to detect fiducial markers for tumor localization on orthogonal X‐ray images for IGRT‐based tracking 14,17 . Fiducial markers appear as small dark signal voids on MRI and cannot be directly used with robotic radiosurgery fiducial marker tracking software 18–20 . For the clinical use of MRI‐only simulation with standard C‐arm linear accelerator (linac)‐based treatments, fiducials identified on MRI are voxel burned at the same location in their corresponding sCT images 21,22 .…”

Technical note: A method to synthesize magnetic resonance images in different patient rotation angles with deep learning for gantry‐free radiotherapy

“…Pix2pix GAN is a classical deep learning network for image-to-image translation, with a comparable performance to that of other methods. [28][29][30] However, it does not perform well for predicting different rotation positions from supine scanned images, which may explain why it is not a simple image-to-image translation task. The same architecture of pix2pix GAN has been developed for pseudo-CT synthesis from the corresponding MRI by our research group.…”

Generative adversarial networks and its applications in the biomedical image segmentation: a comprehensive survey

“…In MRI images, fiducial markers display as tiny-signal voids and are usually challenged to localized in images. The proposed approach relies on deep learning to automatically detect tinysignal fiducial features in scan images [152]. Segmenting the prostate accurately by using MRI images is a challenging research area in prostate cancer diagnosis.…”