Automated Quantification Of Proliferation With Automated Hot-Spot Selection In Phosphohistone H3/Mart1 Dual-Stained Melanomas
Staging of melanoma includes quantification of a pro- liferation index, i.e., presumed melanocytic mitoses of H&E stains are counted manually in hot spots. Yet, its reproducibility and prognostic impact increases by im- munohistochemical dual staining for phosphohistone H3 (PHH3) and MART1, which also may enable fully automated quantification by image analysis.
To ensure manageable workloads and repeatable measurements in modern pathology, the study aimed to present an automated quantification of proliferation with automated hot-spot selection in PHH3/MART1- stained melanomas.
Formalin-fixed, paraffin-embedded tissue from 153 consecutive stage I/II melanoma patients was immuno- histochemically dual-stained for PHH3 and MART1, and whole slide images were captured. An algorithm that automatically detects the number of PHH3/MART1-pos- itive cells was developed using commercial software. In preprocessing, the intensity band of the HSI color model and color deconvolution including a standard deviation filter highlighted PHH3 positivity. Threshold- ing functions classified the image into brown PHH3, red MART1, and blue hematoxylin. Finally, postprocessing al- gorithms pinpointed PHH3/MART1-positive cells based on size, MART1 surrounding, color intensity, and nuclearirregularity. Based on the labels of image analysis, a hot spot was automatically selected by a processing step where circles that detect PHH3/MART1-positive cells produce a heat map according to their cluster. The number of PHH3/MART1-positive cells was counted both automatically and manually in the global tumor area and in a manually and automatically selected hot spot, i.e., a fixed 1-mm2 square.
The mean difference between manual and automated global counts was 2.9 cells/mm2 (P = 0.0071) and 0.23 cells per hot spot (P = 0.96) for automated counts in manually and automatically selected hot spots. In 77% of cases, manual and automated hot spots overlapped. Fully manual hot-spot counts yielded the highest prog- nostic performance with an adjusted hazard ratio of 5.5 (95% CI, 1.3–24, P = 0.024) as opposed to 1.3 (95% CI,
0.61–2.9, P = 0.47) for automated counts with automated hot spots. In conclusion, the automated index and auto- mated hot-spot selection were highly correlated to their manual counterpart, but altogether their prognostic impact was noticeably reduced. Because correct rec- ognition of only one PHH3/MART1-positive cell seems important, extremely high sensitivity and specificity of the algorithm is required for prognostic purposes. The automated analysis may thus still aid and improve the pathologists’ detection of mitoses in melanoma and possibly be useful in other malignancies and future research studies.
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Authors who publish with this journal agree to the following terms:
1. Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
2. Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
3. Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
4. In case of virtual slide publication the authors agree to copy the article in a structural modified version to the journal's VS archive.