Main Logo
Logo

Society for Pediatric Radiology – Poster Archive


Samuel Brady

Is this you? Register and claim your profile. Then, you can add your biography and additional Information.

Showing 3 Abstracts.

In this study, we trained two convolutional neural networks to automatically identify the third vertebral level and segment the abdominal muscle in contrast enhanced abdominal CT images. In the future, these models will be used to determine reference ranges for skeletal muscle mass in children by age for the purpose of identifying patient characteristics associated with differences in skeletal muscle mass. Read More

Meeting name: SPR 2020 Annual Meeting & Postgraduate Course , 2020

Authors: Castiglione James, Gilligan Leah, Somasundaram Elanchezhian, Trout Andrew, Brady Samuel

Keywords: Neural Network, Segmentation, Deep Learning

This educational exhibit will review 1) challenges of CT imaging near metal, 2) current acquisition and reconstruction methods for reducing metallic artifacts, and 3) our initial experience using a GE Revolution CT system for Dual-energy scanning combined with metal artifact reduction (MAR) image reconstruction. Artifacts caused by metallic implants have limited clinical diagnoses for decades using single-energy CT (single kVp, polyenergetic beam) with standard image reconstruction. Low-energy photons in the beam are absorbed by metal, leaving only high-energy photons passing through (ie. beam hardening). Beam hardening due to metal, along with photon starvation and scatter, result in dark shading and bright/dark streaking, as well as lower signal-to-noise levels. Dual-energy CT (DECT) has demonstrated promise for beam hardening reduction because it enables reconstruction of a monoenergetic image, similar in theory to acquiring data with a monoenergetic beam. Recent developments in CT data reconstruction have also achieved better image quality near metal by mitigating shading and streaking artifacts. On our Revolution CT, MAR reconstruction is available solely in dual-energy mode. For our patients with metallic prostheses, we perform DECT and review monoenergetic images with and without MAR. MAR images typically show markedly reduced artifacts from metal and thereby improved image quality. Fig 1 displays 70 keV monoenergetic images both with and without MAR for a patient with a pacemaker. Streaking artifacts arising from the pacemaker were apparent throughout anatomy without MAR, while significantly reduced streaking and improved visualization of the aortic bifurcation is observed in the MAR reconstructed image. Images from a patient with pedicle screws and metallic rods in the spine are shown in Fig 2. Although present, shading and streaking was noticeably reduced with MAR allowing better visibility of the paraspinal soft-tissue structures and the main portal vein. On occasion, however, MAR yielded more severe artifacts for certain slices, such as in the thigh for a patient with a metallic femoral rod just above a total knee replacement (Fig 3). In summary, recent technical advancements incorporated into the Revolution CT system have improved image quality for many of our patients with metallic implants. Predicting a priori when MAR will be worse is not yet possible, so viewing monoenergetic images with and without MAR is recommended. Read More

Meeting name: SPR 2018 Annual Meeting & Postgraduate Course , 2018

Authors: Hornsby Richard, Brady Samuel, Mccarville Beth, Nett Elizabeth, Rupcich Franco, Blancq Terry, Artz Nathan

Keywords: metal artifact reduction, Dual-energy CT

At our institution, airway radiographs are routinely checked by the radiologist to ensure diagnostic image quality prior to the technologist completing the examination. These checks interrupt the workflow for both the technologist and radiologist. In this study, we develop and validate a deep learning algorithm to detect non-diagnostic lateral airway radiographs. Read More

Meeting name: SPR 2020 Annual Meeting & Postgraduate Course , 2020

Authors: Somasundaram Elanchezhian, Brady Samuel, Crotty Eric, Trout Andrew, Anton Christopher, Towbin Alexander, Coley Brian, Dillman Jonathan

Keywords: Deep learning, Airway, Xray