Comparing human and algorithmic tracking of subviral particles in fluorescence microscopic image sequences

Abstract

Abstract:Tracking of subviral particles with automated methods enables the analysis of intracellular processes exhibited by viruses. A linear assignment problem solver and a Kalman-filter have been added to an existing particle tracking algorithm. First results produced with simulated image sequences showed that the improved algorithm is able to improve tracking results by closing gaps in the particle's trajectories. Here we report on the evaluation of the LAPKalman algorithm using real fluorescence-microscopic images. The results from the original and improved algorithm have been compared to the results of manual tracking. Evaluation results indicate that the improved algorithm is capable to reconstruct missing parts of particle tracks in difficult conditions. However, the evaluation of the algorithms and the manual tracking is a complex task because of the low image contrast and high object density with intersecting tracks in the live-cell images.

Fractal Dimensions of Subviral Particle Movement

“…The automated tracking by the previously presented Kienzle- [2] and LAP-Kalman-algorithms [3] provides a good basis for the analysis of subviral particles in fluoroscence image sequences. As recent investigations showed, the tracks match or even excel the trajectories, manually evaluated by experts at the Institute for Virology, Philipps-University, Marburg [4]. The next step to assist the research for new medicines by automation is to develop and establish methods to classify the detected tracks.…”

Automatic Classification of the Movements of Directed and Undirected Subviral Particles

“…where MSE1 is the mean squared error for Equation 5and MSE2 is the mean squared error for Equation (4). The resulting value is between -1 and 1, based on the type of motion.…”

“…Manual analysis of the subviral particles is too timeconsuming and could lead to non-reproducible results [4]. For that reason, algorithms have been developed to detect and track the particles automatically.…”