Automated macrophage counting in DLBCL tissue samples: a ROF filter based approach

Table 1 Summary of segmentation methods’ properties

	(S1)	(S2)	(S3)	(S4)	(S5)	(S6)
Software type
proprietary			∙	∙
freeware extension of proprietary	∙	∙
freeware					∙	∙
Input
.png format	∙	∙	∙	∙	∙	∙
.ndpi(s) format			∙	∙
Output
count	∙				∙
area		∙	∙	∙		∙
annotated image			∙	∙	∙	∙
feature mask	∙	∙			∙	∙
logfile	∙	∙
Evaluation subregion
prescribed					∙	∙
manual detection				∙
automated detection	∙	∙	(∙)
Threshold adaptation
manually			∙	∙
automated	∙	∙			n/a	n/a
Feature detection
none			∙	∙
rule-based	∙	∙
by training set					∙	∙

Abbreviations: (MC) — manual count, (S1) — automated macrophage count from ROF filter based segmentation approach, (S2) — cumulative macrophage area from ROF filter based segmentation approach, (S3) — cumulative macrophage area from Tissue Studio software, (S4) — cumulative macrophage area from Halo software, (S5) — automated macrophage count from Mask R-CNN machine learning approach, (S6) — cumulative macrophage area from Mask R-CNN machine learning approach, (GE) — normalized gene expression values from nCounter platform

ISSN: 1480-9222