The electroretinogram (ERG) is the only diagnostic tool available to objectively assess retinal function. Usually, this electrical potential is evoked by a flash of light and recorded using a single active electrode on the cornea. This response allows to measure the overall function of the retina. However, it does not enable the localization of the sources of the ERG signal. Moreover, since the signal is always evoked by a flash, it does not allow us to assess retinal function at rest.
The general objective of this research project focuses on the distinction between the periodic (or rhythmic) and aperiodic components of the electrophysiological signal of retinal activity and their contribution in terms of clinical diagnosis. Thus, the general hypothesis is that a clinical value for the early diagnosis of retinopathies is added, through the characterization and/or localization of these two components of the ERG signal (evoked or not).
The first part of this research project consists in developing an ERG acquisition methodology and technique to map the activity of the entire retina. Prior to this work, this was only possible for the macula using the multifocal ERG. The technique presented here allows to increase the quantity of recording electrodes, while limiting their total number around the eye (only 3), by using 11 gaze changes to create virtual electrodes (for a total of 33). By this addition, it becomes possible to reconstruct the bioelectrical activity of the sources of this signal (i.e., the retinal cells) by a method solving the inverse problem. Using an electro-anatomical model of the eye (with a Boundary Element Method) and a solution to this problem, it is then possible to calculate the functional topography of the 500 sources representing the retinal activity over 180° which explains the evoked electrical signals, recorded by the multiple virtual electrodes. In doing so, possible lesions present on the retina can be detected. This technique therefore allows a better diagnosis of the various retinal pathologies and a better follow-up in cases of degeneration. Specifically, the results show that by using a LORETA (Low Resolution Electromagnetic Tomography Algorithm), it is possible to reconstruct a central scotoma covering 10% of the retinal surface with signals whose signal-to-noise ratio is 50dB, or 55dB for the same scotoma that would be in the periphery.
The second part of this project consists in analyzing the intrinsic signal of the retina, by recording a resting-state ERG (i.e., non-evoked). To our knowledge, this signal has never been studied before. Given the similarity between the retina and the brain (in their respective development, function and cytoarchitecture), it is possible to compare the new ERG signal at rest to that of the electroencephalogram (EEG) at rest. When these two signals are recorded in an evoked context (by a stimulus or a task), the clinical diagnosis focuses mainly on the periodic component of the signal (i.e., its particular waves). For the resting version of these signals (ERG and EEG), the aperiodic component is also important in the analysis. On the one hand, it is only this component (measured by the Hölder exponent dominant in the signal) that is affected by changes in background light conditions, from dark to dim. On the other hand, the results presented here demonstrate that an invariant 90 Hz oscillatory component exists in the resting-state ERG signal and that it is present with or without using a flash (i.e., intrinsic to the retina), in contrast to the low frequency components of the flash ERG, which are absent when no flash stimulates the retina. Preliminary results on a small selection of pathological subjects indicate that this signal is not affected in the same way by pathologies as are the flash ERG or the multifocal ERG, suggesting a clinical diagnostic potential for the use of the restingstate ERG.
| Date | 9 Aug 2022 |
|---|
| Original language | French |
|---|
| Awarding Institution | - École de technologie supérieure
|
|---|
| Supervisor | Jean-Marc Lina (Supervisor) & Pierre Lachapelle (Co-supervisor) |
|---|
Gauthier, M. (Author),
Lina (Supervisor) & Lachapelle (Co-supervisor),
9 Aug 2022Student thesis: Doctoral thesis › Doctorate in Engineering: Engineering