Table 1.

Troubleshooting guide.

Trouble ID	Troubleshooting	Analysis step	Possible cause	Solutions/Optimization criteria
				During Digital Image Analysis	During staining and image acquisition
1	Partial tissue detection	Global tissue detection	No homogeneous fluorescence signal on whole sections (Intensity, Brightness…)	Choose large thresholds of homogeneity and brightness (0–5 and 0–255)	Use mounting media and a specific IF coverslip (Thickness =  0.17 µm)
					Image acquisition must be processed 1 to 10 days after staining
2	Background or other tissue have been incorrectly classified in the mucosa ROI, for a large area of section	ROI classification	Poor separation between the specific signal and background	Use ROI reclassification based on the fluorescence signal intensity, compactness and density of tissue Manual correction step	Optimization of the exposure time and fluorescence minimal and maximal intensity during image acquisition
					All slides must be stained and scanned with the same procedure and same time
			Imprecise automatic learning due to a heterogeneous fluorescence pattern	Make a new automatic learning algorithm using more representative image objects (segmented part of a ROI) from each ROI
3	Background or other tissue have been incorrectly classified in the mucosa ROI for a small area of section	ROI classification	Imprecise automatic learning due to a heterogeneous fluorescence pattern	Make a new automatic learning algorithm using more representative image objects from each ROI
			Heterogeneous tissue staining or fluorescence pattern for the same ROI	Use ROI reclassification based on the tissue size and relationship to neighbour object
				Manual correction
4	Lymphoid structure has been incorrectly classified as mucosa	ROI classification	Imprecise automatic learning due to a heterogeneous fluorescence pattern	Make a new automatic learning algorithm using more representative image objects from each ROI CAUTION: DO NOT USE follicle objects with weak cell density for correct follicle area distinction
			Heterogeneous tissue staining or fluorescence pattern for the same ROI	Use ROI reclassification based on the tissue size and relationship to neighbour object
				Manual correction
5	Imprecise nuclei detection	Cellular detection	Heterogeneous staining or too blurry fluorescence signal of DAPI	Increase the threshold of DAPI detection	Optimization of staining (Antibody concentration and incubation time)
					Optimization of automated focus during scan, possibility to apply Extending Focus Imaging)
					Optimization of fluorescence acquisition (exposure time, intensity thresholds)
			Heterogeneous staining or too blurry fluorescence signal of membrane marker	Increase the threshold of membrane marker detection if the fluorescence membrane intensity is too high
				Decrease the threshold of membrane marker detection if the fluorescence membrane intensity is too low
				Use the detection mode “growth from nuclei”
6	False-positive cells detected	Cellular classification	Imprecise cell detection (cf Troubleshooting 5)	Cf Troubleshooting 5	Cf Troubleshooting 5
			Fluorochrome deposits	Increase the threshold of DAPI for cell classification	Optimization of staining (decrease concentration of detection, decrease time of denaturation by heat, increase wash number and time)
				Increase the minimal size of cells during membrane detection step
				Exclude aggregates of fluorochromes from the cellular analysis using nuclear filter module
			Tissue autofluorescence on the margin (mucosa and tumor) specific to paraffin-embedded tissue	Exclude area of cellular analysis with the use of new area classification (Mucosa bis, Tumor bis)	Increase the difference between background and specific staining during image acquisition (time exposure, fluorescence intensity thresholds)
7	Aberrant cellular detection links with specific membrane or nuclear structures (e.g., polynuclear cells)	Cellular detection	Aberrant distinction between the nuclear and membrane signals due to incorrect membrane detection by the algorithm	Decrease the resolution and magnification parameter analysis and only use membrane staining to detect cells

All of the steps are based on the fluorescence intensity and fluorescence intensity thresholds, so they were sensitive to any change. This table provides solutions to overcome these problems and quickly optimise the analysis.