Semi-automatic segmentation of mitochondria on electron microscopy images using kalman filtering approach

Mohammadi Alamdari, Aynaz
Mitochondria are membrane bound organelles found in most eukaryotic cells. Mitochondria provide cell’s energy; hence they are called ‘power houses of the cell’. The structure of mitochondria can be illustrated in an electron micrograph. This structure has two membranes: inner and outer. There is a gap between these two membranes, called inter-membrane space. Folds of inner membrane inside the mitochondria form the cristae. To study the relation between mitochondria’s physical structure and its function, electron microscope tomography (EMT) is used to visualize mitochondria. EMT provides 3D structure of mitochondria in high resolution images. In the slices of tomographic images provided by EMT, mitochondria appear as elliptical structures. The cristae are also visualized in these images with various pathology and biological variations. One of the preferred method can be semi-automatic segmentation; since manual segmentation in medical images is time and energy consuming and tedious; moreover fully automatic methods also fail in medical images and cause incorrect results because of low quality of images and restrictions imposed by image acquisition. In this work, an endeavour is made to segment mitochondrial outer boundary using active contour, Kalman filter and optical flow. In the first slice of the images, a contour is provided by user. Then, for the other slices, position values and velocity values calculated using the active contour and optical flow (respectively) are combined with the Kalman filter to predict the points of the boundary in the next slice. In addition, a set of automatic and semi-automatic tools are developed to determine splitting and merging mitochondria, and to segment them.
Citation Formats
A. Mohammadi Alamdari, “Semi-automatic segmentation of mitochondria on electron microscopy images using kalman filtering approach,” M.S. - Master of Science, Middle East Technical University, 2016.