Calibration Alignment Sensitivity in Corneal Terahertz Imaging

Faezeh Zarrinkhat*, Mariangela Baggio, Joel Lamberg, Aleksi Tamminen, Irina Nefedova, Juha Ala-Laurinaho, Elsayed E.M. Khaled, Juan M. Rius, Jordi Romeu, Zachary Taylor

*Corresponding author for this work

Research output: Contribution to journalArticleScientificpeer-review

4 Citations (Scopus)
82 Downloads (Pure)

Abstract

Improving the longitudinal modes coupling in layered spherical structure contributes significantly to corneal terahertz sensing, which plays a crucial role in the early diagnosis of cornea dystrophies. Using a steel sphere to calibrate reflection from the cornea sample assists in enhancing the resolution of longitudinal modes. The requirement and challenges toward applying the calibration sphere are introduced and addressed. Six corneas with different properties are spotted to study the effect of perturbations in the calibration sphere in a frequency range from 100 GHz to 600 GHz. A particle-swarm optimization algorithm is employed to quantify corneal characteristics considering cases of accurately calibrated and perturbed calibrated scenarios. For the first case, the study is carried out with signal-to-noise values of 40 dB, 50 dB and 60 dB at waveguide bands WR-5.1, WR-3.4, and WR-2.2. As expected, better estimation is achieved in high-SNR cases. Furthermore, the lower waveguide band is revealed as the most proper band for the assessment of corneal features. For perturbed cases, the analysis is continued for the noise level of 60 dB in the three waveguide bands. Consequently, the error in the estimation of corneal properties rises significantly (around 30%).

Original languageEnglish
Article number3237
JournalSensors
Volume22
Issue number9
DOIs
Publication statusPublished - 1 May 2022
MoE publication typeA1 Journal article-refereed

Keywords

  • cornea
  • particle-swarm optimization
  • terahertz

Fingerprint

Dive into the research topics of 'Calibration Alignment Sensitivity in Corneal Terahertz Imaging'. Together they form a unique fingerprint.

Cite this