OCT Image Analytics

The advent of optical coherence tomography (OCT), now essentially indispensable, engendered a quantum leap in ophthalmological practices. Specifically, OCT has enabled the ophthalmologists to visualize minute structural changes occurring deep inside the eye without resorting to the traditional invasive procedures. A variety of vision ailments, such as myopia, retinitis pigmentosa, age-related macular degeneration (AMD) and central serous chorioretinopathy (CSC) are screened using OCT. However, in managing intraocular diseases, traditionally, ophthalmologists inspect OCT images in a qualitative manner. Accordingly, to assess its condition, an ophthalmologist would browse through multiple OCT B-scans taken at different vertical locations, and pay attention to the structural changes. In effect, the physician mentally synthesizes the underlying 3D context with a view to forming a medical opinion. This procedure is repeated multiple times in order to monitor treatment response. However, such qualitative assessment suffers from several limitations. Firstly, the clinical opinion could vary from expert to expert, especially, if the structural changes are minute. Further, mentally tallying tiny structural changes from multiple images could be stressful. In view of this, clinicians prefer quantification of the aforementioned structure changes. In reality, however, quantification is still not widely practiced because, with limited assistance from technology, quantification is mostly performed manually, which is time consuming, stressful and susceptible to human error. Further, manual quantification is not practical where the structures are either too complex or too numerous. Against this backdrop, clinicians seek automated quantification tools for faster and accurate disease diagnosis and management. In response, we focus on developing automated image analytic tools which not only facilitates ophthalmologists to perform fast, accurate and stress-free disease management, but also helps improve their understanding of various pathologies.