April 2014
Volume 55, Issue 13
ARVO Annual Meeting Abstract  |   April 2014
An adaptive binning approach in auto-fluorescence lifetime ophthalmology (FLIO) applied to healthy volunteers and patients with diabetes mellitus without diabetic retinopathy
Author Affiliations & Notes
  • Matthias Klemm
    Biomedical Engineering Group, Technische Universität Ilmenau, Ilmenau, Germany
  • Dietrich Schweitzer
    Experimental Ophthalmology, University of Jena, Jena, Germany
  • Footnotes
    Commercial Relationships Matthias Klemm, None; Dietrich Schweitzer, None
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science April 2014, Vol.55, 209. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Matthias Klemm, Dietrich Schweitzer; An adaptive binning approach in auto-fluorescence lifetime ophthalmology (FLIO) applied to healthy volunteers and patients with diabetes mellitus without diabetic retinopathy. Invest. Ophthalmol. Vis. Sci. 2014;55(13):209.

      Download citation file:

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

  • Supplements

Purpose: To improve detection of metabolic changes using FLIO before morphologic alterations are visible. A 3-exponential approximation is optimal for FLIO but requires 10000s of photons / pixel, which can’t be measured due to light exposure and time constraints. Thus, neighboring pixels are binned (static binning, SB) to increase photons / pixel. E.g. eyelashes or cataract cause inhomogeneous auto-fluorescence (AF) intensity. SB may not collect enough photons and cause wrong fluorescence lifetimes. Adaptive binning (AB) ensures required number of photons / pixel at highest possible spatial resolution. SB smoothes all image parts with same strength. AB smoothes dark parts stronger and bright parts less, preserving more detail.

Methods: In 37 healthy volunteers (62.6±11.9 years) and 34 diabetes patients without diabetic retinopathy (64.0±9.1 years) time-resolved retina AF was measured (scanning laser ophthalmoscope: 30° of fundus, 40µm resolution; excitation: diode laser with pico-second pulses, 446nm, 80MHz repetition rate; detection: spectral channels 490-560nm (ch1) and 560-700nm (ch2), time-correlated single photon counting method). All subjects had a crystalline lens. SB (factor 1) combines fluorescence in 3x3 surrounding pixels for each pixel. AB combines surrounding pixels with identical distances. Distance increases per iteration until a threshold (50k, 100k, 150k, 200k photons / pixel) is reached. A modified 3-exponential approach was applied to determine the fluorescence lifetimes. Regions of 100x70 pixels at similar locations in the macula were selected in each subject.

Results: Results are mean ± std. For healthy volunteers SB resulted in: τ1: 73±10ps, τ2: 355±62ps, τ3: 2580±297ps. AB (100k photons) resulted in: τ1: 69±6ps, τ2: 349±20ps, τ3: 2577±156ps. For diabetes patients SB resulted in: τ1: 79±12ps, τ2: 375±83ps, τ3: 3018±256ps. AB (100k photons) resulted in: τ1: 75±8ps, τ2: 353±28ps, τ3: 2796±163ps. The average goodness of fit value (χ2) of 1.3 is identical for volunteers and patients for both binning approaches. Results of ch2 are similar to ch1.

Conclusions: Determination of fluorescence lifetimes is improved. AB reduces std by nearly 50% in comparison to SB for volunteers and patients. Fluorescence lifetimes τ2 and τ3 are slightly reduced in patients by AB. AB is valuable to improve the FLIO method.

Keywords: 551 imaging/image analysis: non-clinical • 688 retina • 585 macula/fovea  

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.