Relevance Vector Machine Image Reconstruction Algorithm for Electrical Capacitance Tomography With Explicit Uncertainty Estimates

Journal Article · 16 January 2020 · IEEE Sensors Journal

DOI: https://doi.org/10.1109/jsen.2020.2966990 · OSTI ID:1802068

Ospina-Acero, Daniel  ^[1]; Marashdeh, Qussai M.  ^[2]; Teixeira, Fernando L.  ^[3]

---

1. The Ohio State Univ., Columbus, OH (United States); Tech4Imaging LLC, Columbus, OH (United States)
2. Tech4Imaging LLC, Columbus, OH (United States)
3. The Ohio State Univ., Columbus, OH (United States)

We present a Relevance Vector Machine (RVM) based algorithm for electrical capacitance tomography (ECT) applications that can concurrently provide image reconstruction results and uncertainty estimates about the reconstruction. To illustrate the RVM operation in ECT, we simulate typical ECT scenarios, making explicit the connection between the reconstructed pixel values and the corresponding uncertainty estimates in each case. We compare the RVM reconstruction performance with that of the Iterative Landweber Method (ILM) and the least absolute shrinkage and selection operator (LASSO) in all the considered scenarios. The results show that, in addition to the key advantage of providing uncertainty measures, RVM can achieve similar reconstruction results with either lower or similar computational complexity.

• Research Organization: Tech4Imaging LLC, Columbus, OH (United States)
• Sponsoring Organization: USDOE Office of Science (SC), Engineering & Technology. Office of Small Business Innovation Research (SBIR) and Small Business Technology Transfer (STTR) Programs
• Grant/Contract Number: SC0010228; SC0011936; SC0018758
• OSTI ID: 1802068
• Journal Information: IEEE Sensors Journal, Journal Name: IEEE Sensors Journal Journal Issue: 9 Vol. 20; ISSN 1530-437X
• Publisher: IEEECopyright Statement
• Country of Publication: United States
• Language: English