[Zeck Vector T3 Pdf Download
Gifford Brickley
The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.
With vision impairment affecting millions of people world-wide, various strategies aiming at vision restoration are being undertaken. Thanks to decades of extensive research, electrical stimulation approaches to vision restoration began to undergo clinical trials. Quite recently, another technique employing optogenetic therapy emerged as a possible alternative. Both artificial vision restoration strategies reported poor spatial resolution so far. In this article, we compared the spatial resolution inferred ex vivo under ideal conditions using a computational model analysis of the retinal ganglion cell (RGC) spiking activity. The RGC spiking was stimulated in epiretinal configuration by either optogenetic or electrical means. RGCs activity was recorded from the ex vivo retina of transgenic late-stage photoreceptor-degenerated mice (rd10) using a high-density Complementary Metal Oxide Semiconductor (CMOS) based microelectrode array. The majority of retinal samples were stimulated by both, optogenetic and electrical stimuli using a spatial grating stimulus. A population-level analysis of the spiking activity of identified RGCs was performed and the spatial resolution achieved through electrical and optogenetic photo-stimulation was inferred using a support vector machine classifier. The best f1 score of the classifier for the electrical stimulation in epiretinal configuration was 86% for 32 micron wide gratings and increased to 100% for 128 microns. For optogenetically activated cells, we obtained high f1 scores of 82% for 10 microns grid width for a photo-stimulation frequency of 2.5 Hz and 73% for a photo-stimulation frequency of 10 Hz. A subsequent analysis, considering only the RGCs modulated in both electrical and optogenetic stimulation protocols revealed no significant difference in the prediction accuracy between the two stimulation modalities. The results presented here indicate that a high spatial resolution can be achieved for electrical or optogenetic artificial stimulation using the activated retinal ganglion cell output.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
The RGC neural activity was recorded using a Complementary Metal-Oxide-Semiconductor microelectrode array (CMOS-MEA) while stimulating the ex vivo retina of adult rd10 transgenic mice. We employed a support vector machine classifier to infer the spatial resolution achievable through either one of the two stimulation methods from the RGC spiking recorded in the same retinae to optogenetic and electrical stimuli.
Prior to the sample preparation, the surface of the recording chip (CMOS MEA, see below) was cleaned with Tickopur (R60, 5%, 80C, Dr. H. Stamm GmbH Chemische Fabrik, Berlin, Germany) and rinsed with distilled water. The active area was covered with a few microliters of poly-L-lysine (1 mg/ml, Sigma Aldrich GmbH, Vienna, Austria) in order to ensure tight tissue adhesion. The coating process proved to not affect the electrical stimulation, as shown in previous work (Eickenscheidt and Zeck, 2014; Stutzki et al., 2016).
In the current study, we used a commercial CMOS-MEA5000 system (Multi Channel Systems MCS GmbH, Reutlingen, Germany) for simultaneous stimulation and recording and CMOS-MEA chips as described previously (Bertotti et al., 2014; Corna et al., 2021). The chips comprise a recording array with 65 by 65 recording sites each of them separated by 16 μm and a second stimulation array comprising 32 by 32 stimulation sites with a center-to-center distance of 32 μm. The array of recording sites is interspersed with stimulation electrodes. To increase the stimulation strength (i.e., capacitive stimulation current), the top oxide was omitted leaving the chip surface with a native oxide only (Corna et al., 2021).
Spike sorting for the identification of single RGC units was performed using the CMOS-MEA-Tools software (Multi Channel Systems MCS GmbH, Reutlingen, Germany), which applies a sorting algorithm based on independent component analysis (Leibig et al., 2016).
Equation 1. The formula for the relative change in firing rate (RFR) of the RGCs identified. FRphase1 and FRphase2 stand for firing rate values in phase 1 and firing rate values in phase 2, respectively.
The statistical analysis of the data was performed using a classification method, the support vector machines1 (Drucker et al., 1997; Pisner and Schnyer, 2019). Due to its relative ease on the resources during the training phase and its performance, it began to gain popularity in the bioengineering field (Pisner and Schnyer, 2019).
One way to visualize the outcomes of the classification task is to use a confusion matrix (Tharwat, 2018), such as in Table 1. Different metrics typically used to assess the quality of the results, such as accuracy, precision, or recall are further extracted from this matrix.
The values computed are typically given in percentages and correspond to the percentage of either correctly or wrongly classified data points. The confusion matrix is used to calculate the f1 score (Eq. 3).
Equation 3. f1 score formula. Here, P denotes the value of the precision and R the value of the recall, respectively. TP refers to the number of true positives, FP the value of the false positives, and FN the value of the false negatives. For the definition of TP, FP, and FN see Table 1.
In this study, we investigated the spatial resolution achieved through electrical or optogenetic stimulation of RGCs in adult transgenic rod degenerated mice (rd10-ChR2). The retina was interfaced with the RGC layer facing the recording and stimulation sites of CMOS MEA chips (Figure 1A). This high density of the recording sites allows electrical imaging of somatic and axonal activity in a large RGC population (Figure 1B) as well as the option to electrically stimulate the retinal neurons. Within the 10 different retinal portions stimulated and analyzed in the following we identified artificial activation in over 800 RGCs.
The spiking activity of two exemplary RGCs to 30 repetitions of the same stimulus pattern (32 μm grid width) is shown in Figure 3A. The spiking in the interstimulus interval is not shown. The two cells were selected to demonstrate their selective activation by one single phase but not the second. Note, that the two cells are activated within each phase of the four sinusoidal stimuli without fading (i.e., decrease in firing rate). A second example (Figure 3B) from the same retina shows an even stronger firing and selective activation of the RGC in only one of the grating phases, i.e., grating reversals. The stronger response is probably caused by the extended stimulation area, now covering 4 32 single electrodes (i.e., 128 1,024 μm2). For these RGCs, no weak activation in the non-preferred phase was detected.
In addition to the RGC soma our algorithm identifies the corresponding axon. Exemplary axons are shown in Figures 3C,D. We note that these axons cross the stimulation electrodes and may potentially be activated. However, the selective stimulation shown in Figures 3A,B for four RGCs demonstrate that axons are not activated here. The avoidance of axonal stimulation in epiretinal configuration for low-frequency (40 Hz) stimuli strengthens the result of previous reports, where the stimulus shape was a small square (Corna et al., 2021) or a single circular electrode (Weitz et al., 2015).
The high prediction accuracy is quite remarkable, provided that both gratings with narrow stripes stimulate the RGCs. We asked, if the high prediction value is mainly determined by one or a few stimulated RGCs. The percentage of contributing cells was above 20% of the whole population irrespective of the stimulation frequencies (2.5 and 10 Hz).
These values are in agreement with a previous study on optogenetic stimulation of RGCs (Reh et al., 2021). However, we also had recordings where the classifier showed poor performance. To identify a potential source, we plotted the most informative features, the ones with the highest coefficients of the SVM, considering data from optogenetic stimulation with a phase reversal at 10 Hz and a spatial frequency of 50 μm (Figure 4B). From Figure 4B; we see that the coefficients are not uniformly distributed, the classifier assigns weights to the incorrectly classified data.
To test the robustness of the model, we used the StratifiedKFold() method, with 10 folds, while also shuffling the data to ensure a random selection. The results (Figures 4C,D) indicate that the classifier performs well, with only small standard deviation values, ranging between 5% and 13%. The mean f1-score for electrical stimulation with a 32 μm wide grating stimulus was 86.4% (n = 5 experiments) and higher for larger gratings.
Finally, we evaluated four recordings were for both, optogenetic and electrical stimulation a sufficiently large number of retinal ganglion cells (>30 RGCs) was responsive to both stimulation modalities. Recordings where either modality activated too few cells were not compared as this biased the evaluation. We first restricted the evaluation to the subset of RGCs activated by electrical and optogenetic stimuli (Figure 5A). The average prediction accuracy (f1 score) based on this subset was: 87.7% for electrical stimulation with 32 μm gratings and 89.8% for optogenetic stimulation with 30 μm gratings. These average values are not statistically different (Wilcoxon-Rank-Sum test); however, the low number of retinal samples compared here (n = 4) prevents a rigorous interpretation. The number of RGCs identified upon electrical or optogenetic stimulation in one single retina is not identical. We, therefore, compared the prediction for the four retina using for each retinae the entire responsive RGC population (Figure 5B). The prediction results increased only slightly for both electrical and optogenetic stimulation when considering the full RGC population activated by each modality (average f1 score = 97.8% for electrical stimuli with 32 μm gratings vs. average f1 score = 90.1% for optogenetic stimulation with 30 μm gratings). On average, high discrimination f1 scores were detected irrespective of the stimulation method, when considering retinae stimulated by both modalities.
795a8134c1