Poster Poster Program Therapy Physics

Metal Artifact Reduction for Radiotherapy CBCT By MV-to-Kv and Kv/MV-to-Kv Image Conversion

Abstract
Purpose

Cone beam CTs at typical kV energies exhibit metal artifacts in IGRT patients with metal prosthetics. MV-CBCT, available on some linacs, reduces artifacts at the expense of soft tissue contrast-to-noise (CNR). As a remedy, we propose an AI network to convert low artifact MV-CBCT volumes into synthetic kV-CBCTs, thereby achieving traditional CNR performance without the artifacts. We also examine an enhanced variant that combines MV-CBCT with artifacted kV-CBCT, when available, to synthesize superior kV-CBCT images.

Methods

A U-Net (MV-kV) was trained to convert MV-CBCT slices of a metal-containing subject to artifact-free kV-CBCT slices. A second U-Net (kV/MV-kV) was trained to derive similar synthetic kV slices but from two inputs, an MV-CBCT and a metal-artifacted kV-CBCT. Input/output data pairs for training/testing were collected using Truebeam scans of a custom pelvis phantom with 2.5” femoral sockets, allowing simulated hip implants to be interchanged with metal-free, natural anatomy. MV-CBCT (2.5 MV, 45 mGy) was implemented with a high-efficiency, 4-layer x-ray detector prototype. Artifact reduction was measured in terms of structural similarity (SSIM) to a metal-free reference kV-CBCT (140 kVp, 37 mGy) in the region of the bladder and prostate. The CNR between adipose and muscle in that region was also quantified.

Results

The proposed U-Nets substantially reduced metal artifacts, with SSIM improvements by >2x in scans with a unilateral titanium hip implant and >4x with bilateral aluminum and titanium. CNR exceeded the given MV-CBCT by >2.7x in all cases. The kV/MV-kV converter matched the CNR of the reference metal-free kV-CBCT in all cases as well, and improved visualization of the prostate as compared to MV-kV only.

Conclusion

AI-based MV-to-kV conversion promises to reduce CBCT metal-artifacts without the sacrifices in CNR normally afflicting MV-CBCT. When available, an additional channel of input from an artifacted kV-CBCT appears to further improve and stabilize the method.

People
Matthew W. JacobsonPresenting Author · Department of Radiation Oncology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Harvard Medical School Thomas C. Harris, PhDAuthors · Department of Radiation Oncology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Harvard Medical School Mathias LehmannAuthors · Varian Imaging Laboratory Pascal HuberAuthors · Varian Imaging Laboratory Roshanak EtemadpourAuthors · Department of Radiation Oncology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Harvard Medical School Francois de Kermenguy, PhDAuthors · Department of Radiation Oncology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Harvard Medical School Dianne M. Ferguson, PhDAuthors · Department of Radiation Oncology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Harvard Medical School Nicholas Lowther, PhDAuthors · Department of Radiation Oncology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Harvard Medical School Marios Myronakis, PhDAuthors · Department of Radiation Oncology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Harvard Medical School Vera BirrerAuthors · Varian Imaging Laboratory Gregory C. Sharp, PhDAuthors · Massachusetts General Hospital Pablo Corral Arroyo, PhDAuthors · Varian Imaging Laboratory Yue-Houng Hu, PhDAuthors · Department of Radiation Oncology, Dana-Farber Cancer Institute/Brigham and Women’s Hospital, Harvard Medical School Raphael BrueggerAuthors · Varian Imaging Laboratory Mathias StammeierAuthors · Varian Imaging Laboratory Bradford VecchioneAuthors · Department of Radiation Oncology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Harvard Medical School Rony FueglistallerAuthors · Varian Imaging Laboratory Ross I. Berbeco, PhDAuthors · Department of Radiation Oncology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Harvard Medical School

Related

Similar sessions

Poster Poster Program
Jul 19 · 07:00
Python-Based Automation Framework for Annual Machine QA Data Archiving In Qatrack+

Annual water-tank measurements help ensure beam characteristics remain consistent with commissioning baselines. However, the lack of a standardized processing workflow and decentralized data storage makes it difficult to analyze...

Syed Bilal Ahmad, PhD
Therapy Physics 0 people interested
Poster Poster Program
Jul 19 · 07:00
User Expectations and Current Availability of HDR Brachytherapy Audits In Europe

The aim of this work was to evaluate the need to implement more dosimetric audits in high‐dose‐rate brachytherapy (HDR-BT) in Europe and to identify which characteristics such audits should meet according to users.

Javier Vijande, PhD Laura Oliver Cañamás
Therapy Physics 0 people interested