4273
MRI data transmission via fifth-generation (5G) cellular networks1Peter L. Reichertz Institute for Medical Informatics, TU Braunschweig and Hannover Medical School, Braunschweig, Germany
Synopsis
In the last years, research in MRI has expanded towards portable technology, e.g., trailer units carrying conventional scanners or bedside scanners with low-field strengh. The increased portability and shift away from the centralized hospital environment requires wireless data transmission with appropriate data rates to unfold its full potential. In this work, we made use of current fifth-generation (5G) cellular networks for transmitting DICOM data. We demonstrate the potential of reaching more than 50 Mbit/s upload speed using consumer-grade hard- and software. This holds great potential for cellular transmission of data from portable scanners, allowing to increase access to MRI technology.
Introduction
Currently, there are two trends aiming for higher portability of MRI technology: i) Conventional MRI scanners are customized towards mobile trailer units which enable availability outside the hospital and ii) portable, low-cost MRI scanners are developed [1], which enable higher availability inside the hospital, e.g., at the bedside [2]. Both approaches indent to bring MRI to more patients for targeting the “Global Unmet Need for MRI” [3].Portable MRI requires wireless data transmission with appropriate data rates for the transmission of the image data to a central server. Existing cellular networks (e.g., 3G, 4G) and wireless computer networks based on IEEE 802.11 standards are potential bottlenecks of data transfer. Furthermore, mobility implies stronger security requirements once the scanner leaves the local network.
5G cellular networks could serve as a data transmission modality allowing easy integration, high data rates, and safe data transmission. In the current Release 15 by the 3rd Generation Partnership Project (3GPP), 5G builts upon existing 4G infrastructure (“Non-Stand alone” (NSA)). Future releases will enable the “Stand alone” mode including higher data rates and enhanced features such as network slicing or edge computing. In this work, we analyze the capabilities of 5G NSA for transmission of DICOM data.
Methods
An open-source DICOM server (Orthanc v1.8.1 [4]) was set-up as a virtual container on a server (Ubuntu Linux Server, Intel Core i5, 32GB RAM, 2TB SSD) located within a virtual private network (VPN). An off-the-shelf business laptop (Manjaro Linux, Intel i7, 16GM RAM) acted as DICOM client (dcmsend() of DICOM Toolkit, https://dicom.offis.de/dcmtk). It was connected via Ethernet connection to a HUAWEI 5G CPE Pro 2 router (Balong 5000 chipset, theoretical 5G transmission rates: 3.6 Gbps (downstream), 250 Mbps (upstream)) which provided 5G access via a SIM card. An off-the-shelf internet data plan for business customers with unlimited volume was used (Deutsche Telekom AG, Business Mobil XL Plus). Figure 1 shows the experimental setup.For the experiments, three locations within a medium-sized city (approx. 250,000 inhabitants) in Germany with 5G availability and different characteristics were chosen: industrial area, residential area, and campus area. At each location, the client setup was applied indoors next to a window and the same batch of DICOM data was sent three times to account for data rate fluctuations.
The dataset was 1 GB of DICOM files acquired during a MRI phantom study (Magnetom Terra, Siemens Healthineers, Erlangen, Germany; "clinical mode") with a fluid-filled phantom simulating tissue. Immediately before starting the DICOM transmission, relevant parameters were read from the web-interface of the 5G router, namely signal to interference plus noise ratio (SINR), reference signal received quality (RSRQ), and reference signal received power (RSRP). As reference, the DICOM client was connected via ethernet to the same physical network as the DICOM server and the experiment was repeated.
Results
Table 1 shows the results with the distance to the cell tower being approximated using a free database of cellular towers (https://www.opencellid.org). In all experiments, DICOM data was transmitted successfully without data corruption or loss. Analyzing the average durations of 5G data transmission shows that upload rates were approximately 57.99 (industrial area), 55.93 (campus area), and 50.67 Mbit/s (residential area). While RSRP and RSRQ are similar for all three areas, the SINR shows interestingly a contrary behavior compared to the data transfer duration. A possible explanation for this is a different number of clients of the cells, which we expect to be lowest for the industrial area, followed by the campus and residential area, respectively.Conclusion
In this work, we made use of 5G NSA cellular networks for transmitting DICOM data. Using consumer-grade hardware, we demonstrated the potential of reaching more than 50 Mbit/s upload speed. This is somehow far away from the theoretical limit of the hardware (250 Mbit/s), but it should be considered that experiments were performed in realistic and indoor scenarios.Even in a residential area – with a presumably high number of cellular users due to the COVID19 pandemic – 1GB of DICOM data was transmitted in under three minutes. This holds great potential for cellular transmission of data from portable MRI scanners. Upcoming releases of the 5G standard will bring even higher data rates and and new technologies which could further increase its value.
Acknowledgements
The authors are thankful to Dr. Stefan Maderwald (Erwin L. Hahn Institute for MRI) and Albert Gnandt (Peter L. Reichertz Institute for Medical Informatics) for providing the DICOM data and support with the DICOM server setup, respectively. This project received funding from the Federal Ministry of Transport and Digital Infrastructure (grant # VB5GFWOTUB).References
[1] Cooley CZ, McDaniel PC, Stockmann JP, Srinivas SA, Cauley SF, Śliwiak M, Sappo CR, Vaughn CF, Guerin B, Rosen MS, Lev MH, Wald LL. A portable scanner for magnetic resonance imaging of the brain. Nat Biomed Eng. 2020 Nov 23. doi: 10.1038/s41551-020-00641-5. Epub ahead of print. PMID: 33230306.
[2] Sheth KN, Mazurek MH, Yuen MM, Cahn BA, Shah JT, Ward A, Kim JA, Gilmore EJ, Falcone GJ, Petersen N, Gobeske KT, Kaddouh F, Hwang DY, Schindler J, Sansing L, Matouk C, Rothberg J, Sze G, Siner J, Rosen MS, Spudich S, Kimberly WT. Assessment of Brain Injury Using Portable, Low-Field Magnetic Resonance Imaging at the Bedside of Critically Ill Patients. JAMA Neurol. 2020 Sep 8:e203263. doi: 10.1001/jamaneurol.2020.3263. Epub ahead of print. PMID: 32897296; PMCID: PMC7489395.
[3] Geethanath S, Vaughan JT Jr. Accessible magnetic resonance imaging: A review. J Magn Reson Imaging. 2019 Jun;49(7):e65-e77. doi: 10.1002/jmri.26638. Epub 2019 Jan 14. PMID: 30637891.
[4] Jodogne S. The Orthanc Ecosystem for Medical Imaging. J Digit Imaging. 2018 Jun;31(3):341-352. doi: 10.1007/s10278-018-0082-y. PMID: 29725964; PMCID: PMC5959835.
Figures
