Accurate model-based segmentation of gynecologic brachytherapy catheter collections in MRI-images

Highlights

• Segmentation and catheter identification in MRI images for brachytherapy.

• ONE CLICK user interaction per catheter.

• Coupling of mechanical model and image features.

• Outlier detection and correction.

• 93% accuracy and 0.29  mm precision error.

Abstract

The gynecological cancer mortality rate, including cervical, ovarian, vaginal and vulvar cancers, is more than 20,000 annually in the US alone. In many countries, including the US, external-beam radiotherapy followed by high dose rate brachytherapy is the standard-of-care. The superior ability of MR to visualize soft tissue has led to an increase in its usage in planning and delivering brachytherapy treatment. A technical challenge associated with the use of MRI imaging for brachytherapy, in contrast to that of CT imaging, is the visualization of catheters that are used to place radiation sources into cancerous tissue. We describe here a precise, accurate method for achieving catheter segmentation and visualization. The algorithm, with the assistance of manually provided tip locations, performs segmentation using image-features, and is guided by a catheter-specific, estimated mechanical model. A final quality control step removes outliers or conflicting catheter trajectories. The mean Hausdorff error on a 54 patient, 760 catheter reference database was 1.49  mm; 51 of the outliers deviated more than two catheter widths (3.4  mm) from the gold standard, corresponding to catheter identification accuracy of 93% in a Syed–Neblett template. In a multi-user simulation experiment for evaluating RMS precision by simulating varying manually-provided superior tip positions, 3σ maximum errors were 2.44  mm. The average segmentation time for a single catheter was 3 s on a standard PC. The segmentation time, accuracy and precision, are promising indicators of the value of this method for clinical translation of MR-guidance in gynecologic brachytherapy and other catheter-based interventional procedures.

Introduction

Gynecological malignancies, including those of the cervix, endometrium, ovaries, and external female genitalia, are a leading cause of mortality in women worldwide. In the United States, with an estimated 105,890 new cases and a mortality of 29%, gynecological malignancies continue to present a medical challenge (American Cancer Society, 2015). Chemoradiation, which consists of concurrent chemotherapy and external-beam radiation, followed by brachytherapy (Fig. 1) remains the standard-of-care for treatment of gynecologic cancers. Compared to external-beam radiation, brachytherapy allows for a higher total dose of radiation to a smaller area in less time, as the radiation sources are placed in direct contact with the cancerous tissue typically under CT- or X-ray-guidance (Han and Viswanathan, 2016). In high-dose-rate (HDR) interstitial brachytherapy, intersitial applicators with catheters that are approximately 20 cm long and 2  mm in diameter are inserted percutaneously through a standardized template surgically sutured to the patient’s perineum (Fig. 2). The catheters are used as channels for bringing radiation seeds in close proximity to the targeted tissue and delivering high-dose radiation to the cancer.

In a survey by the American Brachytherapy Society, the utilization of MRI increased from 2% to 34% between 2007 and 2014 (Grover et al., 2016). This is not surprising, given the ability of MRI to provide better imaging of the tumor and adjacent soft tissues (compared to CT), and hence its routine use in the radiologic diagnosis of pelvic cancers (Jolesz, 2014). However, the artifacts created in MRI scans using typical brachytherapy catheters are considerably more difficult to interpret compared to CT (see Fig. 1(a) and (b) for CT imaging catheter artifacts, and Fig. 1(c) for MRI imaging catheter artifacts). Artifacts created by catheters used in pelvic brachytherapy are very distinct in X-ray or CT images and therefore, amenable to automatic segmentation using the standard image-processing techniques of commercial brachytherapy treatment planning products.¹${}^{,}$² This observation is primarily because voxels on these catheters correspond to a narrow range of high Hounsfield values that are distinct from human tissue in CT images. In contrast, the grayscale range of catheters in MRI scans overlaps with that of human tissue, and one part of the catheter can appear to be substantially different from another part of the same catheter while being difficult to distinguish from surrounding tissue (Zand et al., 2007). Today, there are no automatic solutions for the segmentation of brachytherapy catheters from MRI images (Song, Cho, Iordachita, Guion, G, Kaushal, Camphausen, Whitcomb, 2012, DiMaio, Kacher, Ellis, Fichtinger, Hata, Zientara, Panych, Kikinis, Jolesz, 2005). Even manual segmentations from MRI are time-consuming, tedious, and error-prone because of the large numbers and high density of catheters in the images, and not used in clinical research or practice today.

We contribute to the state of research in the following ways:

• An innovative, customized algorithm: Use of a mechanical catheter bending model to constrain an image-coupled segmentation process; an automatic error-correction step to analyze intersecting catheters and automatically correct the results.

• A novel application area with a large database for medical image-analysis: Accuracy and precision of the proposed method are illustrated on a database of 760 catheters in 54 patients including difficult cases with up to 40 catheters for MRI-guided high dose rate gynecologic brachytherapy, a novel and growing clinical application area. This is the first report using this database.

• Performance accuracy and precision improvement using stronger model assumptions. We used a mechanical model to augment the Bezier-model (Pernelle et al., 2013).

• Precision evaluation: A multiple user input simulation to estimate precision; also the first of its kind for MRI-guided high dose rate gynecological brachytherapy.

In summary, we present here a novel catheter segmentation method, with promising accuracy and precision numbers, that has been validated using a large MRI database containing hundreds of catheters.