Investigative Ophthalmology & Visual Science Cover Image for Volume 65, Issue 7
June 2024
Volume 65, Issue 7
Open Access
ARVO Annual Meeting Abstract  |   June 2024
Adaptive-balancing detection towards clinical use of vis-OCT
Author Affiliations & Notes
  • Roman Kuranov
    Northwestern University, Evanston, Illinois, United States
    Opticent Health, Evanston, Illinois, United States
  • David Andrew Miller
    Northwestern University, Evanston, Illinois, United States
  • Raymond Atkinson
    Northwestern University, Evanston, Illinois, United States
    Opticent Health, Evanston, Illinois, United States
  • Hao Zhang
    Northwestern University, Evanston, Illinois, United States
  • Footnotes
    Commercial Relationships   Roman Kuranov Opticent Health, Code E (Employment); David Miller None; Raymond Atkinson Opticent Health, Code E (Employment); Hao Zhang Opticent Health, Code I (Personal Financial Interest)
  • Footnotes
    Support  NH Grants U01EY0033001; R44EY026466
Investigative Ophthalmology & Visual Science June 2024, Vol.65, 5497. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Roman Kuranov, David Andrew Miller, Raymond Atkinson, Hao Zhang; Adaptive-balancing detection towards clinical use of vis-OCT. Invest. Ophthalmol. Vis. Sci. 2024;65(7):5497.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose : The advancement of the vis-OCT towards clinical use is hindered by the lack of cost-effective low-noise light sources. Subpixel-balanced detection reduces requirements to the vis-OCT sources but requires precise hardware pre-calibration of two spectrometers, which can be degraded over time. We developed subpixel calibration that does not require hardware pre-calibration and metadata for balancing detection.

Methods : Retinal vis-OCT data was acquired from a healthy volunteer after 20 days of the system’s initial installation. To enable balanced detection, we used two synchronized spectrometers. We used two published balanced detection and one unbalanced method to compare human vis-OCT images.
The noise-based cross-correlation (NCC) balancing uses a relative intensity noise-dominated dataset to identify pixel pairs that share the highest temporal correlation. To achieve subpixel mapping, a third-order polynomial fit is applied to the pixel map vector. This requires qualified non-clinical engineering pre-calibration proving and additional metadata that need to be stored for processing.
The adapted balance minimizes the mean amplitude and variance of the balanced A-line by selecting the coefficients of an interpolation vector using the Nelder-Mead simplex search algorithm. In this method, high-quality vis-OCT images are acquired from the fringes without any preliminary hardware metadata.

Results : For both balanced methods, the RIN noise is mostly eliminated (Figures 1a & 1b), while for the adaptive balance, noise is 11% lower than for the NCC, after 20 days. The noise for the unbalanced image (Figure 1c) is six times higher than the adaptive balanced level near the zero delay (0-100 µm) and 1.7 times higher at deeper depths (>450 µm). In the unbalanced image, the choroid and outer limiting membrane are not delineated.

Conclusions : We demonstrated that subpixel-balanced detection can suppress the RIN noise in vis-OCT imaging and provide a nearly shot-limited performance of the vis-OCT. The adaptive balance can provide the highest quality based on the actual data set without metadata. The balance methods that rely on the subpixel pre-alignment require qualified engineers’ intervention.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.

 

Figure 1. Vis-OCT images of the macular region using a) adaptive balance, b) pre-calibrated noise-based cross-correlation (NCC), and c) unbalanced. The image size is 3 mm (horizontal) by 0.64 mm (vertical).

Figure 1. Vis-OCT images of the macular region using a) adaptive balance, b) pre-calibrated noise-based cross-correlation (NCC), and c) unbalanced. The image size is 3 mm (horizontal) by 0.64 mm (vertical).

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×