Table 3. Overview of quantification approaches for fibred endoscopic imaging.

Organ (System)	Quantifying	References	Methodology	Comments
Circulatory	Red blood cell velocity.	Savoire et al. (2004)	Thresholding and line-fitting (M-estimators) translated (through trigonometry) to RBC velocity.	Inventive use of known and quantifiable artefact in raster scanning imaging systems for deriving physiological information. Preliminary results with uncertain clinical relevance.
		Perchant et al. (2007)	ROI tracking and alignment through (i) scanning distortion compensation, and (ii) global affine registration, for blood velocity estimation through spatio-temporal correlation.	Feasibility study. Preliminary results with uncertain clinical relevance.
Oropharyngeal	Epithelial cells in vocal chords.	Mualla et al. (2014)	Watershed segmentation (borders) and local minima detection (location).	Empirical, ad-hoc approach employing off-the-shelf image analysis methods. Limited data can potentially lead to poor generalisation of the proposed methodology.
Gastro-intestinal	Intestinal crypts in Inflammatory Bowel Disease. (eCLE)	Couceiro et al. (2012)	Detecting (local maxima), segmenting (ellipse fitting on edge detection) and quantifying (number, connectivity).	Empirical, ad-hoc approaches employing off-the-shelf image analysis methods Heuristic parameter estimation, hard thresholds and limited data can potentially lead to poor generalisation of the proposed methodologies.
	Intestinal crypts in colorectal polyps.	Prieto et al. (2016)	Contrast enhancement, thresholding (Otsu’s) and morphological filters (erosion, centre of mass, circularity).
	Goblet cells in villi. (eCLE)	Boschetto et al. (2015a)	Detecting (matched filters), segmenting (Voronoi diagrams) cells and identifying (hard threshold) goblet cells within the villi.
	Intestinal villi. (eCLE)	Boschetto et al. (2015b)	Detect via morphological filters (top-hat, morphological reconstruction and closing) and quad-tree decomposition.
		Boschetto et al. (2016b)	Subdivide to superpixels, extract features and classify through Random Forests to generate a binary segmentation map.	Employing established data driven approaches with reasonable size of data, resulting on better generalisation potential.
Pulmonary	Alveoli sacs in mice distal lung.	Namati et al. (2008)	Segmenting (optimum separation thresholding) and quantifying (8-point connectivity) alveolar sacs.	Limited data and uncertain translatability to human alveoli sacks due to their large size relative to the limited field of view.
	Stained mesenchymal stem cells in rat lungs.	Perez et al. (2017)	Contrast stretch, denoise (opening), threshold and count (connected component analysis).	Empirical, ad-hoc approach employing off-the-shelf image analysis methods.
	Stained bacteria in distal lung.	Karam Eldaly et al. (2018)	Outlier detection using a hierarchical Bayesian model along with a MCMC algorithm based on Gibbs sampler.	More elaborate approaches, adopting model-based and data-driven methodologies.
	Stained bacteria and cells in distal lung.	Seth et al., (2017, 2018)	Bacterial and cellular load using spatio-temporal template matching with a radial basis functions network.	They have potential for good generalisation and translation to clinical applications.