IEEE 2021 NSS MIC

Please note! All times in the online program are given in New York - America (GMT -04:00) times!

New York - America ()
Jan 29, 2022, 8:10:31 AM
Your time ()
n/a
Tokyo - Asia ()
Jan 29, 2022, 10:10:31 PM
Our exhibitors and sponsors – click on name to visit booth:

To search for a specific ID please enter the hash sign followed by the ID number (e.g. #123).

X-ray and CT

Session chair: Cho , Seungryong (Korea Advanced Institute of Science & Technology (KAIST), Daejeon, South Korea); Taguchi , Katsuyuki (Johns Hopkins University, Baltimore, USA)
 
Shortcut: M-12
Date: Friday, 22 October, 2021, 7:00 AM - 8:45 AM
Room: MIC - 2
Session type: MIC Session

Contents

Click on an contribution to preview the abstract content.

7:00 AM M-12-01

Preliminary investigation of directional-TV-based image reconstruction from limited-angular-range data with two orthogonal arcs (#678)

Z. Zhang1, B. Chen1, D. Xia1, E. Y. Sidky1, X. Pan1, 2

1 The University of Chicago, Department of Radiology, Chicago, Illinois, United States of America
2 The University of Chicago, Department of Radiation and Cellular Oncology, Chicago, Illinois, United States of America

Abstract

Computed tomography (CT) image reconstruction from data collected over a limited-angular range (LAR) is challenging, yet yields practical interest and significance. In the work, we propose an innovative imaging configuration with two orthogonal arcs of LARs and investigate image reconstruction by using the directional-total-variation (DTV) algorithm. Results show that the proposed configurations, along with the DTV algorithm, may improve image reconstruction accuracy over a single arc of LAR with the same total angular coverage. Results also indicate that the reconstruction performance can be affected by the distribution of angular range over the two orthogonal arcs. The work may provide insights to designing innovative configurations with LARs for CT imaging to yield reduced radiation dose and imaging time, and/or for collision-avoidance.

AcknowledgmentThis work was supported in part by NIH R01 Grant Nos. EB026282, EB023968, and Grayson-Jockey Club Research. The computation of the work was performed in part on the computer cluster funded by NIH S10-OD025081, S10-RR021039, and P30-CA14599 awards. The contents of this paper are solely the responsibility of the authors and do not necessarily represent the official views of NIH.
Keywords: limited-angular-range (LAR), directional total variation, primal-dual algorithm, computed tomography, two-orthogonal-arc configuration
7:15 AM M-12-02

A feasibility study of digital tomosynthesis system using a carbon nanotube array (#787)

H. Kim1, J. Soh2, U. Jeong2, S. Cho2, M. Ito3, Y. - J. Jung3, T. - H. Kim3, S. Cho2, 1

1 Korea Advanced Institute of Science and Technology, KAIST Institute for Artificial Intelligence, Daejeon, Republic of Korea
2 Korea Advanced Institute of Science and Technology, Nuclear and Quantum engineering, Daejeon, Republic of Korea
3 LG Electronics, LG Electronics Future Technology Center, Seoul, Republic of Korea

Abstract

Digital tomosynthesis is an imaging modality that provides quasi-three-dimensional images of anatomy by acquiring a limited number of projections at finite source positions.  Effective data sampling scheme in a compact digital tomosynthesis system is an active research area. This study presented a prototype digital tomosynthesis system using a moving carbon nanotube array and its tailored reconstruction algorithm. Our tomosynthesis system was configured with a fixed detector and a source array that consists of seven carbon nanotube-based X-ray sources. The system acquired projection data at different source positions by moving the source array. To handle data truncation and non-ideal source problems, we adopted a weighted data fidelity scheme for image reconstruction. The weighting functions were achieved from pre-shot data of uniform PMMA phantom. We also exploit the directional total variation minimization technique to produce better image quality. The combined object function was iteratively minimized by using the Chambolle-Pock algorithm. Physical hand and ankle phantoms were scanned to test the proposed system and reconstruction algorithms. Reconstruction results successfully demonstrated the feasibility of the proposed system.

Acknowledgment

This work was supported by LG Electronics.

Keywords: Tomosynthesis, Carbon nanotubes, Image reconstruction
7:30 AM M-12-03

Deep Learning-based Fully Automated Scan Range Detection in Chest CT Imaging (#924)

Y. Salimi1, A. AkhavanAllaf1, I. Shiri1, Z. Mansouri3, A. Saberimanesh1, A. Sanaat1, M. Pakbin4, D. Askari5, S. Sandoughdaran6, E. Sharifipour7, H. Arabi1, H. Zaidi1, 2

1 Geneva University Hospital, Division of Nuclear Medicine and Molecular Imaging, Department of Medical Imaging, Geneva, Genève, Switzerland
2 University of Geneva, Geneva University Neurocenter, Geneva, Genève, Switzerland
3 Shahid Beheshti University of Medical Sciences, Department of Biomedical Engineering and Medical Physics, Tehran, Iran (Islamic Republic of)
4 Qom University of Medical Sciences, Qom, Iran, Imaging Department, Qum, Iran (Islamic Republic of)
5 Shahid Beheshti University of Medical Sciences, Department of Radiology Technology, Tehran, Iran (Islamic Republic of)
6 Shahid Beheshti University of Medical Sciences, Department of Radiation Oncology, Tehran, Iran (Islamic Republic of)
7 Qom University of Medical Sciences, Neuroscience Research Center, Qum, Iran (Islamic Republic of)

Abstract

This study aimed to develop an automated scan range selection for minimal patient irradiation in CT examinations within the chest region. A total number of 20,820 chest CT images acquired for various indications were collected, the 3D lung masks were generated using a Deep Neural Network (DNN) developed by our group. Consequently, 2D projected localizer images and masks were computed in lateral (lat) and anterior-posterior (AP) directions. We developed a deep learning algorithm to predict the lung mask from 2D localizer images. Thereby, the scan range is automatically determined without the technologist’s intervention. Lastly, the impact of over-ranging on patients’ effective dose was investigated through personalized dosimetry of the given cohort. A significant over scanning range (31±24 mm) was observed in clinical setting for more than 95% of cases. The average Dice coefficient for 2D lung segmentation was 0.96 and 0.97 for AP and lateral projections, respectively. The proposed approach resulted in errors of 0.08±1.46 and -1.5±4.1 mm in the superior and inferior directions, respectively. The effective dose (ED) was reduced by 21% in the unseen external dataset when using the proposed automated scan range selection.

Acknowledgment

This work was supported by the Euratom research and training programme 2019-2020 Sinfonia project under grant agreement No 945196.

Keywords: CT dosimetry, deep learning, scan range, chest imaging
7:45 AM M-12-04

Pre-experiments using photon-counting CT with machine learning models for drug delivery system monitoring (#190)

T. Toyoda1, S. S. Djara Dima1, M. Sagisaka1, J. Kataoka1, M. Arimoto2, J. Kotoku3, M. Taki4, A. Oyama3, S. Kobayashi2, H. Kawashima2, D. Sato2, K. Yoshiura2, S. Terazawa5, S. Shiota5, H. Ikeda6, M. Ueda7

1 Waseda University, Tokyo, Japan
2 Kanazawa University, Ishikawa, Japan
3 Teikyo University, Tokyo, Japan
4 Rikkyo University, Tokyo, Japan
5 Hitacihi Metal Ltd, Tokyo, Japan
6 Insutitute of Space and Astronautical Science, Kanagawa, Japan
7 Okayama University, Okayama, Japan

Abstract

X-ray computed tomography (CT) has been widely used in medical diagnostic imaging.
However, conventional, energy-integrated CT requires a high radiation dose and can only provide monochromatic images that cannot eliminate various artifacts.  In contrast, photon-counting CT (PC-CT) can provide low-dose multicolor CT imaging, which enables the identification of multiple contrast agents. Therefore, as the next step of PC-CT utilization, we propose a new concept of treatment monitoring based on PC-CT. For example, the visualization and confirmation of the therapeutic effect of drug delivery systems (DDSs) have been mentioned. In this study, we performed in vitro imaging of gold nanoparticles (AuNPs), which are used as anticancer drugs in DDSs, as a preliminary validation for clinical application of DDSs monitoring. Furthermore, ultra-low concentration AuNPs imaging was also challenged.  However, in the PC-CT system, the lack of photon statistics, which is also caused by image reconstruction in the limited energy band, severely affected the discrimination between ultra-low concentration AuNPs and water. Thus, to complement the statistics and improve material decomposition, we applied two types of machine learning (ML) techniques, that is, U-Net and Noise2Noise to PC-CT images. These ML models were trained using low- and high-dose image pairs created in simple steps. Consequently, we successfully reproduced PC-CT images with a high peak signal-to-noise ratio, which enabled the extraction of ultra-low concentration AuNPs from water. As a future perspective, we will try to demonstrate DDSs monitoring in mice.

AcknowledgmentThis research was supported by JSPS KAKENHI Grant Number JP20K20923 and 20H00669. 
Keywords: Photon counting CT, MPPC, Drug delivery system, Machine learning, Deep-learning
8:00 AM M-12-05

Performance estimate of MPPC-based PC-CT system and initial results of CT image contrast (#793)

D. Sato1, M. Arimoto1, 2, K. Yoshiura1, T. Mizuno1, H. Kawashima1, S. Kobayashi1, J. Kataoka2, H. Kiji2, T. Toyoda2, S. Dima2, H. Ikeda3, S. Terazawa4, S. Shiota4

1 Kanazawa University, Kanazawa city, Japan
2 Waseda University, Shinjukuku, Japan
3 Institute of Space and Astronautical Science, Sagamihara city, Japan
4 Hitachi Metal Ltd., Minatoku, Japan

Abstract

X-ray computed tomography (CT) is widely used for three-dimensional nondestructive X-ray imaging of the internal structure of the human body or industrial materials. For modern technology in the medical field, dualenergy CT (DE-CT) with two types of X-ray effective energy has generally been used. However, the X-ray signals of DECT are integrated and read out in the form of a current. Thus, contamination with dark noise significantly degrades such imaging qualities as contrast, which causes a large radiation dose to patients. In addition, little energy information on DE-CT results in poor material discrimination of target materials. Recently, the photoncounting CT (PC-CT) system has been developed for future CT technology. Because the PC-CT system can detect individual X-ray photons, the dark-noise effect is expected to be highly suppressed. The multiple energy data obtained by PC-CT provide fruitful information on the energy dependence of the CT values, leading to high potential for material discrimination. Thus, an MPPC-based PC-CT system combined with high-speed scintillators has been proposed, and a 64-channel CT array system was developed recently. In this study, the details of the performance estimate of the MPPC-based PC-CT system were investigated in terms of energy information and photoncounting capability. The initial results of the CT image contrast compared with the clinical DE-CT system are presented. They show that the proposed PC-CT system achieved a similar or superior contrast-to-noise ratio value to that of the clinical DE-CT.

AcknowledgmentWe gratefully acknowledge financial support from JSPS KAKENHI Grant Numbers JP19H04483 and JP19K22924, the Naito Foundation, the Uehara Memorial Foundation, the Casio Science Promotion Foundation, and the JSPS Leading Initiative for Excellent Young Researchers program (M.A.).
Keywords: Photon-counting CT, MPPC, Dual-energy CT
8:15 AM M-12-06

Bridge-Assisted Micropillar Structure for High-Aspect Ratio X-ray Grating Fabrication (#1450)

A. Pil-Ali1, S. Adnani1, C. Con2, Z. H. Warsi1, K. S. Karim1, 2

1 University of Waterloo, Department of Electrical and Computer Engineering, and Centre for Bioengineering and Biotechnology, Waterloo, Ontario, Canada
2 KA Imaging Inc., Waterloo, Ontario, Canada

Abstract

Coded-aperture and Talbot-Lau are among x-ray phase contrast imaging techniques that work based on x-ray absorption masks (gratings). X-ray gratings play the central role in these techniques, where visibility -- image quality -- is highly dependent on the quality of gratings. Although fabrication process of high-aspect ratio structures for line-gratings is well-developed, there are technological challenges in realizing 2D-gratings through fabrication of fine high-aspect ratio micropillars that restrict x-ray imaging systems' sensitivity and functionality. The main challenge is micropillars mechanical stability during, and after, fabrication. Whether SU-8 based LIGA or silicon-based deep-RIE process is used to fabricate high-aspect ratio micropillars, during developing, cleaning, or electroplating, micropillars that are not stable enough could collapse, bend, or deform. In this work, we introduce bridge-assisted micropillars to improve their mechanical stability. The addition of auxiliary supporting bridge structures help micropillars to stand alone, which also eliminate the need for critical point drying step. To demonstrate proof-of-concept of bridge-assisted micropillars, two samples are fabricated through both SU-8 based LIGA and silicon-based deep-RIE process. The result of this work helps researchers further expand 2D phase-contrast x-ray imaging systems' sensitivity.

Acknowledgment

This work was supported in part by Natural Sciences and Engineering Research Council of Canada (NSERC) Collaborative Research and Training Experience (CREATE) program, and University of Waterloo.

Keywords: X-ray Phase Contrast Imaging, high-resolution, High-aspect ratio, X-ray Absorption Grating, Micropillar
8:30 AM M-12-07

Large-area Perovskite Detectors for Low-Dose X-ray Imaging (#1057)

A. Datta1, K. Hansen1, J. Lu2, R. Street2, S. Motakef1

1 CapeSym, Inc., R&D, Natick, Massachusetts, United States of America
2 PARC, Palo Alto, California, United States of America

Abstract

With the advent of flat panel detector technology, flat panel x-ray imagers (FPXIs) are now widely used in digital X-ray imaging. These imagers are currently being heavily used during the COVID-19 crisis to assess the peculiar lung-damage that is unique to this virus. Commercial FPXIs are primarily based on scintillators or are based on amorphous-Se semiconductor films that directly convert absorbed x-rays to an electronic signal.  The indirect FPXIs have high detective quantum efficiency (DQE) over the energy range of interest (up to 140kVp), but due to the isotropic propagation of light in the scintillators, these systems lack the necessary spatial resolution. On the other hand, a-Se based panels have excellent spatial resolution and DQE and require relatively low X-ray doses of X-ray. But, due to the low absorptivity of a-Se at higher energies, the use of direct conversion FPXIs is currently limited only to soft-tissue applications such as mammography. This presentation will demonstrate direct conversion X-ray detection with a detector structure based on novel photo-detecting polycrystalline semiconductor Methylammonium lead iodide (MAPbI3) characterized by a high attenuation coefficient and excellent charge transport properties. These detectors not only provide low dark current and high X-ray sensitivity but also can be deposited over large areas for manufacturing FPXIs with repeatable performance. We will present the spatial uniformity information of the detectors for areas as large as 10cm2. These sensors were also deposited on large area aSi:H active pixel arrays and timePIX3 ASICs as a stepping stone for translating this technology to clinical medical imaging applications.

Acknowledgment

This work was supported by US National Institute of Health under the contract number EB028208.

Keywords: X-ray Detectors, Direct radiography, Flat Panel detectors

Our exhibitors and sponsors – click on name to visit booth: