Table 1.
List of the cancer samples and immunogens tested during the current study.
Figure 1.
Representation of color deconvolution using the old and the new optical density (OD) vectors.
A: Color deconvolution using the old OD vectors. B: Color deconvolution using the new OD vectors. C: Scatter plot comparing the intensities on the complimentary image with the old OD vectors (blue) and the new OD vectors (red). D: Plot comparing the number of pixels with the intensity value of 255. An improvement between 2 to 10 fold is shown using 7 different samples. Each data plot represents an individual sample with its respective pixel count of the intensity value of 255.
Figure 2.
Representative histogram profile and score of a cytoplasmic and nuclear stained image using IHC Profiler.
A: Profiling of the DAB stained cytoplasmic image sample. The histogram profile corresponds to the pixel intensity value vs. corresponding number counts of a pixel intensity. The log given below the histogram profile shows the accurate percentage of the pixels present in each zone of pixel intensity and the respective computed score. B: Profiling of the DAB stained nuclear stained image sample. The red spots on the DAB image indicate the threshold selection of the nucleus areas using the threshold function of ImageJ. The representative histogram profile corresponds to the number of pixels vs. the corresponding value at which the pixel of the respective intensity is present.
Figure 3.
Flow chart demonstrating the computing steps involved in the working algorithm.
Figure 4.
Impact of magnification on image scoring.
A: Analysis of a 10X image area where a significant amount of stroma and fatty tissue is present. After color deconvolution, the score assigned by IHC profiler on the DAB image was determined as low positive. B: Scoring analysis of the same tissue area where image captured was by using a 20X lens in the marked area, focusing more on the actual tumor mass resolute a score of positive. C: Scoring analysis of the same tissue area wherein the image was captured using a 40X lens, focusing more on eliminating the stromal and fatty tissue region increases the percentage of the positive pixels in the positive and high positive zones.
Table 2.
Comparison chart showing automated vs. manual scoring.
Table 3.
Agreement of scores between manual vs. IHC Profiler assessment.
Table 4.
Variability of scores in between the pathological opinions.
Table 5.
Variability of scores in between the pathological opinions.
Table 6.
Variability of scores in between the pathological opinions.
Table 7.
Comparison of IHC profiler with available IHC image analysis tools.