Problems with installation and configuration
- The error "??? Invalid MEX-file 'some file name': The
specified module could not be found." is caused by some libraries
missing on your operating system. To solve this problem, install the
compatible C++ compiler, which can be found on
MathWorks website (e.g.
for version 2011a; make sure you select your operating system).
- Sometimes MicrobeTracker may run out of memory, especially on 32-bit
systems during long serrions. The memory problem will be mentioned in the
error message. Unfortunately, some MATLAB functons used in the suite have
memory leaks, and memory requirements imcrease with runtime. To comtinue, restart
MATLAB and try splitting your task into smaller ones, ideally requiring less
memory, (e.g. avoid using batch processing, and in case of large image stacks,
do cell detection and signal addition separately).
Problems with cell detection
- The program processes a large number of regions per cell detected with areas 1-20 pixels (as seen on the screen). Multiple small "phantom cells" detected (individual frames of 1st frame of a timelapse). The reason is that the threshold was not detected correctly. This usually happens is the illumination is not uniform. Try increasing thresFactorM and thresFactorF parameters (to 1.1-1.4) which increase the threshold, though this may lead to missing some cells and to crushing on some frames. If this happens only on some frames and it is affordable to skip them, set maxRegNumber to a lower number (1000-2000).
- Neighbor cells overlap. Most common reason is a large value of attrCoeff. Reduce it to 0-0.3 (algorithm 4) or 0-0.02 (algorithms 2 and 3), set repCoeff to less than attrCoeff·3.
- Cell outlines miss the poles. Check the
scaleFactor, try increasing
- Cell outlines form extensions from the poles in darker areas of the image.
- Cells divide too early or too late. Adjust
splitThreshold (usually in the range 0.15-0.5). If
neighbor cells don't split in the Individual Frames mode, set split1
to 1 so that it tries splitting. If individual cells split in the Individual Frames
joinangle (typically to 0.6-1.2) and joindist
(typically to 4-8) so that the program tries to join cells.
- The programs performs a timelapse well, but misses certain cells starting from one frame.
Usually this is due to large drift of the images or to an out-of-focus frame. Try aligning frames, though if the drift
is too strong, it may fail. Check is there are out-of-focus frames, you can
consider removing these frames from the stack. The reason for the problem can also be significant growth of the
cells between frames. The only solution in this case is taking reducing the
time intervals between frames.
- Signal profile does not show distinct peaks. Probably the background was
not subtracted. Make sure the signal images are loaded in the right channel
when performing background subtraction and integration of the signal.