1
|
Thalhammer J, Schultheiß M, Dorosti T, Lasser T, Pfeiffer F, Pfeiffer D, Schaff F. Improving Automated Hemorrhage Detection at Sparse-View CT via U-Net-based Artifact Reduction. Radiol Artif Intell 2024; 6:e230275. [PMID: 38717293 DOI: 10.1148/ryai.230275] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 06/06/2024]
Abstract
Purpose To explore the potential benefits of deep learning-based artifact reduction in sparse-view cranial CT scans and its impact on automated hemorrhage detection. Materials and Methods In this retrospective study, a U-Net was trained for artifact reduction on simulated sparse-view cranial CT scans in 3000 patients, obtained from a public dataset and reconstructed with varying sparse-view levels. Additionally, EfficientNet-B2 was trained on full-view CT data from 17 545 patients for automated hemorrhage detection. Detection performance was evaluated using the area under the receiver operating characteristic curve (AUC), with differences assessed using the DeLong test, along with confusion matrices. A total variation (TV) postprocessing approach, commonly applied to sparse-view CT, served as the basis for comparison. A Bonferroni-corrected significance level of .001/6 = .00017 was used to accommodate for multiple hypotheses testing. Results Images with U-Net postprocessing were better than unprocessed and TV-processed images with respect to image quality and automated hemorrhage detection. With U-Net postprocessing, the number of views could be reduced from 4096 (AUC: 0.97 [95% CI: 0.97, 0.98]) to 512 (0.97 [95% CI: 0.97, 0.98], P < .00017) and to 256 views (0.97 [95% CI: 0.96, 0.97], P < .00017) with a minimal decrease in hemorrhage detection performance. This was accompanied by mean structural similarity index measure increases of 0.0210 (95% CI: 0.0210, 0.0211) and 0.0560 (95% CI: 0.0559, 0.0560) relative to unprocessed images. Conclusion U-Net-based artifact reduction substantially enhanced automated hemorrhage detection in sparse-view cranial CT scans. Keywords: CT, Head/Neck, Hemorrhage, Diagnosis, Supervised Learning Supplemental material is available for this article. © RSNA, 2024.
Collapse
Affiliation(s)
- Johannes Thalhammer
- From the Department of Physics, School of Natural Sciences (J.T., M.S., T.D., F.P., D.P., F.S.), Munich Institute of Biomedical Engineering (J.T., M.S., T.D., T.L., F.P., D.P., F.S.), Department of Diagnostic and Interventional Radiology, School of Medicine, Klinikum rechts der Isar (J.T., M.S., T.D., F.P., D.P.), Institute for Advanced Study (J.T., F.P., D.P.), and Computational Imaging and Inverse Problems, Department of Computer Science, School of Computation, Information, and Technology (T.L.), Technical University of Munich, Boltzmannstrasse 11, 85748 Garching, Germany
| | - Manuel Schultheiß
- From the Department of Physics, School of Natural Sciences (J.T., M.S., T.D., F.P., D.P., F.S.), Munich Institute of Biomedical Engineering (J.T., M.S., T.D., T.L., F.P., D.P., F.S.), Department of Diagnostic and Interventional Radiology, School of Medicine, Klinikum rechts der Isar (J.T., M.S., T.D., F.P., D.P.), Institute for Advanced Study (J.T., F.P., D.P.), and Computational Imaging and Inverse Problems, Department of Computer Science, School of Computation, Information, and Technology (T.L.), Technical University of Munich, Boltzmannstrasse 11, 85748 Garching, Germany
| | - Tina Dorosti
- From the Department of Physics, School of Natural Sciences (J.T., M.S., T.D., F.P., D.P., F.S.), Munich Institute of Biomedical Engineering (J.T., M.S., T.D., T.L., F.P., D.P., F.S.), Department of Diagnostic and Interventional Radiology, School of Medicine, Klinikum rechts der Isar (J.T., M.S., T.D., F.P., D.P.), Institute for Advanced Study (J.T., F.P., D.P.), and Computational Imaging and Inverse Problems, Department of Computer Science, School of Computation, Information, and Technology (T.L.), Technical University of Munich, Boltzmannstrasse 11, 85748 Garching, Germany
| | - Tobias Lasser
- From the Department of Physics, School of Natural Sciences (J.T., M.S., T.D., F.P., D.P., F.S.), Munich Institute of Biomedical Engineering (J.T., M.S., T.D., T.L., F.P., D.P., F.S.), Department of Diagnostic and Interventional Radiology, School of Medicine, Klinikum rechts der Isar (J.T., M.S., T.D., F.P., D.P.), Institute for Advanced Study (J.T., F.P., D.P.), and Computational Imaging and Inverse Problems, Department of Computer Science, School of Computation, Information, and Technology (T.L.), Technical University of Munich, Boltzmannstrasse 11, 85748 Garching, Germany
| | - Franz Pfeiffer
- From the Department of Physics, School of Natural Sciences (J.T., M.S., T.D., F.P., D.P., F.S.), Munich Institute of Biomedical Engineering (J.T., M.S., T.D., T.L., F.P., D.P., F.S.), Department of Diagnostic and Interventional Radiology, School of Medicine, Klinikum rechts der Isar (J.T., M.S., T.D., F.P., D.P.), Institute for Advanced Study (J.T., F.P., D.P.), and Computational Imaging and Inverse Problems, Department of Computer Science, School of Computation, Information, and Technology (T.L.), Technical University of Munich, Boltzmannstrasse 11, 85748 Garching, Germany
| | - Daniela Pfeiffer
- From the Department of Physics, School of Natural Sciences (J.T., M.S., T.D., F.P., D.P., F.S.), Munich Institute of Biomedical Engineering (J.T., M.S., T.D., T.L., F.P., D.P., F.S.), Department of Diagnostic and Interventional Radiology, School of Medicine, Klinikum rechts der Isar (J.T., M.S., T.D., F.P., D.P.), Institute for Advanced Study (J.T., F.P., D.P.), and Computational Imaging and Inverse Problems, Department of Computer Science, School of Computation, Information, and Technology (T.L.), Technical University of Munich, Boltzmannstrasse 11, 85748 Garching, Germany
| | - Florian Schaff
- From the Department of Physics, School of Natural Sciences (J.T., M.S., T.D., F.P., D.P., F.S.), Munich Institute of Biomedical Engineering (J.T., M.S., T.D., T.L., F.P., D.P., F.S.), Department of Diagnostic and Interventional Radiology, School of Medicine, Klinikum rechts der Isar (J.T., M.S., T.D., F.P., D.P.), Institute for Advanced Study (J.T., F.P., D.P.), and Computational Imaging and Inverse Problems, Department of Computer Science, School of Computation, Information, and Technology (T.L.), Technical University of Munich, Boltzmannstrasse 11, 85748 Garching, Germany
| |
Collapse
|
2
|
Nakai Y, Miyazaki O, Kitamura M, Imai R, Okamoto R, Tsutsumi Y, Miyasaka M, Ogiwara H, Miura H, Yamada K, Nosaka S. Evaluation of radiation dose reduction in head CT using the half-dose method. Jpn J Radiol 2023:10.1007/s11604-023-01410-5. [PMID: 36961648 PMCID: PMC10366020 DOI: 10.1007/s11604-023-01410-5] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Grants] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/25/2022] [Accepted: 02/27/2023] [Indexed: 03/25/2023]
Abstract
PURPOSE The present study introduced the half-dose method (HDM), which halves the radiation dose for conventional head computed tomography (CT), for postoperative hydrocephalus and follow-up for craniosynostosis at a children's hospital. This study aimed to evaluate the contribution of selective head CT scanning optimization towards the overall reduction of radiation exposure. MATERIALS AND METHODS We retrospectively assessed 1227 and 1352 head CT examinations acquired before and after the introduction of the HDM, respectively, in children aged 0-15 years. The radiation exposure was evaluated using the CT dose index volume (CTDI-vol), dose-length product (DLP), rate of HDM introduction, and effect of reducing in-hospital radiation dose before and after the introduction of the HDM. For an objective evaluation of the image quality, head CT scans acquired with HDM and full-dose method (FDM) were randomly selected, and the image noise standard deviation (SD) was measured for each scan. In addition, some HDM images were randomly selected and independently reviewed by two radiologists. RESULTS The HDM was introduced in 27.9% of all head CTs. The mean CTDI-vol of all head CTs was 21.5 ± 6.9 mGy after the introduction, a 14.9% reduction. The mean DLP was 418.4 ± 152.9 mGy.cm after the introduction, a 17.2% reduction. Compared to the FDM images, the noise SD of the HDM ones worsened by almost 0.9; however, none of the images were difficult or impossible to evaluate. CONCLUSION The HDM yielded diagnostically acceptable images. In addition, a change in protocol for only two diseases successfully reduced the patients' overall radiation exposure by approximately 15%. Introducing and optimizing the HDM for frequently performed target diseases will be useful in reducing the exposure dose for the hospital's patient population.
Collapse
Affiliation(s)
- Yoshitomo Nakai
- Department of Radiology, National Center for Child Health and Development, 2-10-1 Okura, Setagaya-Ku, Tokyo, Japan.
- Department of Radiology, Kyoto Prefectural University of Medicine, 465 Kawaramachi Hirokoji Kamigyo-Ku, Kyoto, Japan.
| | - Osamu Miyazaki
- Department of Radiology, National Center for Child Health and Development, 2-10-1 Okura, Setagaya-Ku, Tokyo, Japan
| | - Masayuki Kitamura
- Department of Radiology, National Center for Child Health and Development, 2-10-1 Okura, Setagaya-Ku, Tokyo, Japan
| | - Rumi Imai
- Department of Radiology, National Center for Child Health and Development, 2-10-1 Okura, Setagaya-Ku, Tokyo, Japan
| | - Reiko Okamoto
- Department of Radiology, National Center for Child Health and Development, 2-10-1 Okura, Setagaya-Ku, Tokyo, Japan
| | - Yoshiyuki Tsutsumi
- Department of Radiology, National Center for Child Health and Development, 2-10-1 Okura, Setagaya-Ku, Tokyo, Japan
| | - Mikiko Miyasaka
- Department of Radiology, National Center for Child Health and Development, 2-10-1 Okura, Setagaya-Ku, Tokyo, Japan
| | - Hideki Ogiwara
- Division of Neurosurgery, National Center for Child Health and Development, 2-10-1 Okura, Setagaya-Ku, Tokyo, Japan
| | - Hiroshi Miura
- Department of Radiology Japanese Red Cross Society, Kyoto Daini Hospital, 355-5 Haruobicho Kamanza-Dori Marutamachi-Agaru, Kamigyo-Ku Kyoto, Japan
| | - Kei Yamada
- Department of Radiology, Kyoto Prefectural University of Medicine, 465 Kawaramachi Hirokoji Kamigyo-Ku, Kyoto, Japan
| | - Shunsuke Nosaka
- Department of Radiology, National Center for Child Health and Development, 2-10-1 Okura, Setagaya-Ku, Tokyo, Japan
| |
Collapse
|
5
|
Bingyang B, Gang W, Zhiqing S, Li N, Zhou B, Xu S, Li D. A Preliminary Study of Personalized Head CT Scan in Pediatric Patients. Dose Response 2021; 19:1559325820985660. [PMID: 33746652 PMCID: PMC7940743 DOI: 10.1177/1559325820985660] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.3] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/15/2020] [Revised: 11/30/2020] [Accepted: 12/13/2020] [Indexed: 01/02/2023] Open
Abstract
Objectives: In the present study, we introduced a practical approach to quantify organ-specific radiation doses and investigated whether low-dose head circumference (HC)-based protocols for non-enhanced head computed tomography (CT) could reduce organs-specific radiation dose in pediatric patients while maintaining high image quality. Methods: A total of 83 pediatric patients were prospectively recruited. Without limits to the HC, 15 patients were selected as a convention group (CON group) and underwent non-enhanced head CT scan with standard-dose protocols (tube current-time products of 250mAs). Low-dose group (LD group), including remaining 68 pediatrics were divided into 3 subgroups based on the HC: 54.1-57.0 cm for LD200mAs group (HC-based protocols of 200mAs), 51.1-54.0 cm for LD150mAs group (HC-based protocols of 150mAs), 48.1-51.0 cm for LD100mAs group (HC-based protocols of 100mAs). Subjective and objective image quality was evaluated and measured by 2 experienced radiologists. Radimetrics was used to calculate organs-specific radiation dose, including the brain, eye lenses, and salivary glands. Results: In CON250mAs group, radiation doses in the brain and salivary glands were conversely correlated with HC, and pediatric patients with smaller HC received higher organs-specific radiation dose. Reducing tube current-time product from 250 to 100mAs could significantly reduce the organ-specific radiation dose. The subjective image quality score ≥ 3.0 is acceptable for diagnosis purposes. The signal to noise ratio (SNR) and the contrast to noise ratio (CNR) of bilateral thalamus and centrum semiovale in 3 LD subgroups were not statistically different compared with the CON group. Conclusion: Our research indicated that low-dose HC-based protocols of non-enhanced head CT scan can evidently reduce the organ-specific radiation doses, while maintaining high image quality. HC can serve as a vital tool to guide personalized low-dose head CT scan for pediatric patients.
Collapse
Affiliation(s)
| | - Wang Gang
- Third Hospital of Jilin University, Changchun, China
| | - Shao Zhiqing
- Jilin University First Hospital, Changchun, China
| | - Nan Li
- Jilin University First Hospital, Changchun, China
| | - BoXu Zhou
- Jilin University First Hospital, Changchun, China
| | - ShuJia Xu
- Jilin University First Hospital, Changchun, China
| | - Dan Li
- Jilin University First Hospital, Changchun, China
| |
Collapse
|