- The SESA algorithm uses the simplest method for image segmentation, which is called ‘linear thresholding’. The algorithm identifies all the ‘objects’ in a scan based on their contrast against the background and then calculates the total surface area covered by such objects. Therefore, this method is not based on absorbance but uses the object covered area, so that it is not affected by background intensity changes (such as shadow effects caused by, e.g., condensation on the microtiter plate lid and air bubbles in the culture medium) and can measure microbial growth with high accuracy at very low cell concentrations. However, when more than 20% of the total image area is covered by objects, the SESA algorithm accuracy starts to decline. The SESA algorithm gives faster results compared to conventional OD measurements. The SESA value is calculated as:
SESA value = log10(Σ(object covered area))
Only the best-focus Z-stack layers are analysed by the growth kinetic algorithm based on object-based image analysis (OBIA), which is the segmentation extracted average length (SEAL) algorithm. In general terms, OBIA employs two main processes: segmentation and classification. OBIA groups image pixels into homogeneous ‘objects’, which can have different shapes and intensity scale. The ‘objects’ are also associated with statistics that can be used for classification of such ‘objects’ and include geometry, context and texture.
- The SEAL algorithm is specifically designed to detect filamentation of rod shaped bacteria. The SEAL algorithm performs segmentation to identify all the ‘objects’ in a scan and determines their average length. The SEAL algorithm is limited when bacterial cells or filaments are overlapping and may lead to inaccurate determination of bacterial length at high cell concentrations. The SEAL value is calculated as
SEAL value =Σ(object length) / number of objects
The oCelloScope segmentation analysis is also based on OBIA and is performed on the best-focus layer of the Z-stacks to identify and quantify up to twenty different morphological features (e.g., area, eccentricity, symmetry, …) of individual ‘objects’ (or group of objects), such as microbes, spores and cells. Moreover, it is also possible to monitor the development of such morphological features over time with the segmentation kinetics analysis.
For samples that are sedimented to a monolayer at the bottom of the microwell (Fig. 2), all the cells will be in focus along the horizontal plane. Therefore, the image Z-stack will contain the all-in-focus image as well as the adjacent out-of-focus images along the vertical axis. The generation of video and data analysis are the same as shown for samples in suspension.