Improved segmentation of meristem cells by an automated optimisation algorithm

• Published in: LACNEM 2017 : 7th Latin American Conference on Networked and Electronic Media : 6-7 November 2017, Valparaiso, Chile
• Main Authors: Rojas, O, ero, MG, Mene´ndez, JM, Jones, A, Dewitte, W, Murray, JAH
• Format: Conference Proceeding
• Language: English
• Published: Stevenage The Institution of Engineering & Technology 06.11.2017
• Subjects: Algorithms; Ground truth; Image contrast; Image segmentation; Iterative methods; Mathematical analysis; Mathematical morphology; Microscopy; Optimization; Parameters; Similarity
• ISBN: 1785618253; 9781785618253
• DOI: 10.1049/ic.2017.0035

This paper presents a new technique based on mathematical morphology for the automatic segmentation of meristem cells in microscopy images. The meristem is a plant tissue that gives rise to all its structures. Meristem cells in confocal microscopy are seen as dark areas surrounded by bright ones. Images are characterised by regions of very low contrast and absolute loss of edges when deeper into the meristem. Edges are blurred, discontinuous, sometimes indistinguishable, and the intensity level inside the cells is similar to the background of the image. The optimal parameters for the segmentation are found by means of an iterative process called the Parametric Segmentation Tuning (PST). It compares the segmented images, obtained in successive steps, using an optimisation function to determine which pair allows to get the best segmentation. The technique was validated comparing a ground truth image with the output segmentation image to verify if the optimal parameters are close to those found with the proposed technique. A Receiver Operating Characteristic (ROC) analysis is used to verify the trend of the parameter variation towards the point of maximum similarity between the segmented and the ground truth images. Other similarity indexes were found to produce the same results.