Objective Histologic Stain Quality and Variability Analysis through Digital Imaging: The Effect of Staining Automation
Matthew J Swadley, Marvin L Jones, Alton B Farris. Emory University, Atlanta, GA
Background: Histologic staining variability often poses difficulty in pathologic diagnosis and can be a source of frustration for practicing pathologists. However, assessment is often based upon subjective eyeball evaluation of glass slides. Digital image analysis provides the ability to objectively quantify such evaluations, and allow for regimented quality control over this portion of the hospital laboratory.
Design: Twenty-four daily Masson's Trichrome control slides (12 hand-stained, 12 automated) from available nearest consecutive days prior to and after the implementation of automated staining were digitally scanned with an Aperio Scanscope XT. Two 10X screenshots of representative vasculature were captured from each whole slide file in Aperio Imagescope. Using the open source GNU Image Manipulation Program (GIMP), the optical color components of hue, saturation, and intensity value (HSV) were systematically recorded for 10 points of perivascular fibrous tissue of each vessel. Combined HSV measurements were averaged for each day and analyzed for mean deviation and day-to-day variability.
Results: Automation resulted in statistically significant differences in both stain appearance and day-to-day variability. Mean perivascular stain hue and saturation differed significantly (p = 0.002 and 0.0006, respectively), while value showed no significant change.
|Hue (95% CI)||250.4 (237.7-263.1)||222.7 (218.9-226.6)|
|Saturation (95% CI)||43.5 (36.9-50.1)||65.1 (57.9-72.5)|
|Value (95% CI)||64.3 (60.0-68.6)||63.7 (60.2-67.2)|