Algorithm Benchmarking for Foveal Avascular Zone Quantification from OCT Angiography

  • Typ:Bachelorarbeit
  • Betreuung:

    M. Sc. Simon Hoffmann

  • Bearbeitung:

    Johannes Breitling

  • Motivation:
    Diabetic Retinopathy (DR) causes major impacts in the life of patients. DR progression is accompanied by vascular alterations. Optical Coherence Tomography Angiography (OCT Angiography), a fast, non-invasive, high-resolution angiographic imaging modality, allows assessing the foveal avascular zone (FAZ) area, its perimeter and circularity. The large number of algorithms for FAZ parameter extraction leads to inconsistencies. Introduced disagreements in FAZ parameters will be analyzed.
    In a first step, a systematic literature review will be conducted. The focus will be set on foveal avascular zone boundary definitions and applied algorithms for segmentation and quantification. Reviewing the algorithm verification approaches rounds it off. In a second step, you will reimplement a selection of algorithms. An implementation in Matlab is preferred. You will conduct a benchmarking study. OCTA-500, FAZID and OCTAGON datasets are (publicly) available. Discuss the database you are working with, your implementations and findings critically. In a third step, inconsistencies introduced by the variety of algorithms are assessed. Disagreements will be examined among algorithms and compared to differences of healthy controls and DR patients.