Michael J Doughty1. 1. Department of Vision Sciences, Glasgow-Caledonian University, Glasgow, Scotland. m.doughty@gcu.ac.uk
Abstract
BACKGROUND: To assess the use of a public domain software Image J (NIH Image, Bethesda, MD) to make dimensional measures of human bulbar conjunctival cells. METHODS: Impression cytology samples were obtained from the nasal bulbar conjunctiva, and color images were taken at 200× magnification and projected and an overlay prepared or the image uploaded into Image J. The final image magnification for the overlays was approximately 2× that on the computer screen. For either overlays or screen images, linear measures were made from 30 or 25 cells of the cell longest dimension (LONG) and of the longest dimension of the nucleus (NUCLONG). The predicted variability in measures, from the calculated average values for any particular image, was systematically assessed, and the overall average results were compared. RESULTS: Image J measures were within ±2% agreement with overlays, with LONG and NUCLONG measures increasing with the grade of squamous metaplasia. The net difference in LONG measures was within -2.65 to + 2.93 μm, and NUCLONG measures were within -0.87 and +1.39 μm. There was a slight tendency for NUCLONG measures to be systematically overestimated on smaller nuclei when using Image J on the color images. CONCLUSIONS: The manual use of Image J on on-screen images can provide reasonably accurate objective measures of bulbar conjunctival cells, as compared with a higher magnification manual overlay technique. This should be suitable for comparisons between samples and to assess the effects of any disease-related changes or any interventions.
BACKGROUND: To assess the use of a public domain software Image J (NIH Image, Bethesda, MD) to make dimensional measures of human bulbar conjunctival cells. METHODS: Impression cytology samples were obtained from the nasal bulbar conjunctiva, and color images were taken at 200× magnification and projected and an overlay prepared or the image uploaded into Image J. The final image magnification for the overlays was approximately 2× that on the computer screen. For either overlays or screen images, linear measures were made from 30 or 25 cells of the cell longest dimension (LONG) and of the longest dimension of the nucleus (NUCLONG). The predicted variability in measures, from the calculated average values for any particular image, was systematically assessed, and the overall average results were compared. RESULTS: Image J measures were within ±2% agreement with overlays, with LONG and NUCLONG measures increasing with the grade of squamous metaplasia. The net difference in LONG measures was within -2.65 to + 2.93 μm, and NUCLONG measures were within -0.87 and +1.39 μm. There was a slight tendency for NUCLONG measures to be systematically overestimated on smaller nuclei when using Image J on the color images. CONCLUSIONS: The manual use of Image J on on-screen images can provide reasonably accurate objective measures of bulbar conjunctival cells, as compared with a higher magnification manual overlay technique. This should be suitable for comparisons between samples and to assess the effects of any disease-related changes or any interventions.