Abstract
Purpose: :
Retinal ganglion cell (RGC) loss is one of the earliest and most important cellular changes in glaucoma. To date, calculating the number of apoptosing RGCs imaged using DARC (Detection of Apoptosing Retinal Cells) has been done manually. This method is time-consuming, has inter-operator and intra-operator variability, and is subject to bias. The purpose of this study was to develop an automated method of counting the number of RGCs that is accurate, time-efficient and non-operator dependent.
Methods: :
An algorithm was developed on Matlab software which allowed identification of apoptosing RGCs labelled with fluorescent annexin-5. Processing included a local-luminance and local-contrast "gain control", a "blob analysis" to differentiate between cells, vessels and noise, and the exclusion of non-cell structures using combined size and aspect ratio criteria. The algorithm then generated an automated count of the total number of accurately positively identified apoptosing RGCs. The manual count of apoptosing RGCs was used as the gold standard comparator in the analysis. 100 rat eyes imaged with DARC underwent assessment using manual counting by a blinded individual and automated counting using our newly developed technique. To test statistical correlation, Pearson’s R, Intraclass Correlation Coefficient (ICC) and Cronbach’s Alpha Reliability Coefficient were used.
Results: :
There was a strong agreement between the apoptosing RGC counts obtained through manual and automated counts, as shown by both Pearson's R and the ICC, CI (0.969-0.998), r= 0.996 and p<0.0001. Cronbach Alpha analysis also found an excellent consistency between the two tests, alpha= 0.996. The automated cell count was therefore consistent with the gold standard cell count.
Conclusions: :
The novel automated image analysis algorithm has enabled an accurate calculation of the total number of apoptosing RGCs that was highly comparable to the manual count. This innovative technique has the additional advantages of being reproducible, fast, non-labour intensive and cost-effective. As DARC enters the phase of clinical trials, this technique may have great implications for the large-scale implementation of DARC in the early diagnosis, monitoring and assessment of glaucoma and retinal neurodegeneration.
Keywords: imaging/image analysis: non-clinical • image processing • apoptosis/cell death