Methods and systems for tracking movement of microscopic worms and worm-like organisms, and software therefor
Methods and systems for tracking one or more worms or worm-like organisms over a sequence of video frames in virtual real time. The methods and systems can include a robust organism model that accounts for shape changes that occur from one frame to another, such as peristaltic progression, longitudinal deformation, lateral deformation, and bending action. Other features disclosed include: features that allow a user to correct tracking errors, such as splitting a single organism track into two tracks, joining two organism tracks into a single track, switching locations of physical features (such as heads and tails of worms), deleting undesired tracked organisms, and manually tracing organism outlines for model fitting; features that allow a user to set tracking parameters by selecting one or more organisms having desired characteristics; features for automatedly resolving interactions between/among multiple organisms; and features for handling multiple tracking hypotheses, among others.
1. A method of tracking one or more worms, the method being implemented by a computer-based worm-tracking system and comprising:
receiving a plurality of frames of a video containing one or more worms;
displaying, sequentially on an electronic display, the plurality of frames of the video;
calculating, periodically using information in ones of the plurality of frames of the video, tracking information for one or more of the worms so as to create one or more tracked worms, wherein said calculating tracking information includes:
estimating a shape of each tracked worm using a geometric model that includes shape elasticity that accounts for changes in worm length among the ones of the plurality of frames; and
estimating the location and configuration of each tracked worm by applying a midpoint radial shift operator that compensates for cumulative radial positioning error at a central location of each tracked worm among the ones of the plurality of frames; and
displaying the tracking information overlaid correspondingly onto the ones of the plurality of frames displayed sequentially on the video display.
2. A method according to claim 1 , wherein said calculating tracking information includes applying a motion model that accounts for a bending angle of each tracked worm.
3. A method according to claim 1 , wherein said calculating tracking information includes fitting a worm model to each tracked worm by calculating a fitting force.
4. A method according to claim 3 , wherein said calculating tracking information further includes calculating an interaction force between at least two of the one or more worms and fitting the worm model to one of the at least two worms as a function of a sum of the fitting force and the interaction force, wherein the interaction force accounts for worm self-overlap.
5. A method according to claim 1 , wherein said calculating tracking information results in tracking each tracked worm, wherein said tracking includes:
tracking each tracked worm from one of the plurality of frames to a subsequent one of the plurality of frames as a function of a positional tolerance; and
starting a new worm track when the positional tolerance is exceeded.
6. A method according to claim 1 , wherein said calculating tracking information results in tracking each tracked worm, wherein said tracking includes:
tracking each tracked worm from one of the plurality of frames to a subsequent one of the plurality of frames as a function of a shape tolerance; and
starting a new worm track when the shape tolerance is exceeded.
7. A method according to claim 1 , further comprising allowing a user to identify whether or not to track any of the one or more worms at a frame boundary.
8. A method according to claim 7 , further comprising allowing the user to identify a number of frames to continue tracking a frame-boundary worm until tracking of the frame-boundary worm is terminated.
9. A method according to claim 1 , further comprising:
tracking fitness of multiple model-fit hypotheses over multiple successive ones of the plurality of frames; and
eliminating one or more of the multiple model-fit hypotheses over time as a function of the fitness.
10. A method according to claim 1 , further comprising:
allowing a user to identify a worm-tracking error; and
automatedly fixing the worm tracking error as a function of the user identifying the worm-tracking error.
11. A method according to claim 10 , wherein said allowing a user to identify a worm-tracking error includes allowing the user to connect two worm tracks together into a single worm track.
12. A method according to claim 10 , wherein said allowing a user to identify a worm-tracking error includes allowing the user to swap head and tail positions of each tracked worm.
13. A method according to claim 10 , wherein said allowing a user to identify a worm-tracking error includes allowing the user to split a worm track into two separate worm tracks.
14. A method according to claim 10 , wherein said allowing a user to identify a worm-tracking error includes allowing the user to delete each tracked worm.
15. A method according to claim 10 , wherein said allowing a user to identify a worm-tracking error includes allowing the user to manually trace a worm.
16. A method according to claim 1 , wherein said calculating tracking information includes:
tracking at least one worm of the one or more tracked worms in forward order relative to the video; and
tracking the at least one worm in backward order relative to the video.
17. A method according to claim 16 , wherein said tracking at least one worm in forward order includes tracking forward from an initial-worm-detection frame, and said tracking the at least one worm in backward order includes tracking backward from the initial-worm-detection frame.
18. A method according to claim 1 , further comprising analyzing at least one frame for the presence of any worms having an omega bend.
19. A method according to claim 18 , wherein said analyzing for the presence of any worms having an omega bend includes determining whether or not any of the worms has an acute bending angle over a plurality of frames.
20. A method according to claim 1 , further comprising:
setting an image threshold for the plurality of frames;
receiving a user selection of one or more of the worms; and
automatedly adjusting a worm-characterization filter as a function of the one or more user-selected worms so as to generate a set of worm-characterization parameters;
wherein said calculating tracking information includes calculating tracking information as a function of the set of worm-characterization parameters.
21. A method according to claim 1 , further comprising, prior to said calculating:
performing image thresholding on image data of the video to obtain thresholded-image data; and
applying one or more morphological filters to the thresholded image data to obtained segmented image data;
wherein the tracking information is derived from the segmented image data.
22. A method according to claim 1 , wherein each of said calculating tracking information and said displaying the tracking information occur substantially simultaneously with said displaying the plurality of frames.
23. A method according to claim 22 , wherein said displaying the plurality of frames of the video includes displaying the plurality of frames at a normal frame-rate that allows a viewer to perceive, in realtime, realtime motion of each tracked worm.