Microtubule-based transport is critical for trafficking of organelles, organization of endomembranes, and mitosis. The driving force for microtubule-based transport is provided by microtubule motors, which move organelles specifically to the plus or minus ends of the microtubules. Motor proteins of opposite polarities are bound to the surface of the same cargo organelle. Transport of organelles along microtubules is discontinuous and involves transitions between movements to plus or minus ends or pauses. Parameters of the movement, such as velocity and length of runs, provide important information about the activity of microtubule motors, but measurement of these parameters is difficult and requires a sophisticated decomposition of the organelle movement trajectories into directional runs and pauses. The existing algorithms are based on establishing threshold values for the length and duration of runs and thus do not allow to distinguish between slow runs and pauses, making the analysis of the organelle transport incomplete. Here we describe a novel algorithm based on multiscale trend analysis for the decomposition of organelle trajectories into plus- or minus-end runs, and pauses. This algorithm is self-adapted to the characteristic durations and velocities of runs, and allows reliable separation of pauses from runs. We apply the proposed algorithm to compare regulation of microtubule transport in fish and Xenopus melanophores and show that the general mechanisms of regulation are similar in the two pigment cell types.