Manual

Manual rectangular:

In this mode, for each measurement, select (by clicking and dragging) a rectangular region around the object (D above). When you un-click the mouse, the entire rectangle will be highlighted (E) and the entire area within the rectangle will be measured. The reported area will be the area of the rectangle. However, since the background is still set to 0.702, only the pixels in the spot, where the image is different from the background, will contribute to the signal and density results. Thus, if the background is set correctly, the reported signal and density should be the same regardless of the exact size of the rectangle.

In Manual Rectangular mode, it is sometimes helpful to select several background areas first, and adjust the background level, until selecting a rectangle on the background gives a density close to zero.

Alternatively, click on ``Auto bkgd'' to allow the program to automatically calculate the background value each time (see below). Make sure to include a small portion of the background with each measurement if you do this.

Densitometry will continue as long as rectangles are selected. It is not necessary to click OK on the Result Window (this actually has no effect). To stop, click Cancel.

Fixed size rectangular:

This mode is similar to Manual Rectangular mode, except that all measurements are guaranteed to be the same area. Before the first measurement, select a rectangular region. An information box will appear saying ``Ready to begin densitometry on (size) regions''. From then on, as long as densitometry is active, clicking on a single point will cause a region of size and shape identical to the first region to be be analyzed, centered at the point where the mouse was clicked.

As with Manual Rectangular, if Auto Bkgd. is checked, the background level will be calculated automatically each time. In this case, it is important to make the rectangle large enough to include a small amount of the background, so that the calculation is accurate (ideally, the area of background and the area of signal should be the same).

Manual Irregular:

This mode gives the maximal amount of control to the user. In this mode, you must manually draw an outline around the feature to measure. When you un-click the mouse, an Information Box appears saying ``Click on any point inside selection''. This means the program needs to know which part of the area you sketched is the inside. Clicking on the inside causes the inside region to be temporarily colored blue (F in the figure above). When you are finished selecting the area, the blue will disappear, and the outlined area will be automatically measured.

This mode would typically be used for morphometry and area measurements, in which case it is desirable to avoid including background pixels in the selection.

If you want to measure multiple disjoint areas simultaneously, change the SELECT_MODE const in xmtnimage20.cc from SINGLE to MULTIPLE and recompile imal. The only difference will be that, instead of one blue area (F above), you can have any number of them. When you are done selecting blue areas, click the main Cancel button. The densitometry results will be reported for the union of all the sets of points.

Color images should be converted to grayscale before performing densitometry. Color densitometry is not yet supported.

If ``Auto calculate background'' is selected, you must include an approximately equal number of foreground and background pixels so the fuzzy k-means algorithm can calculate the correct background. It is recommended to keep ``auto background'' off and use a fixed background value when measuring irregular areas.

During this mode, pressing Ctrl-V will permanently mark the selected area.

Fixed Shape Irregular:

Same as Manual Irregular, except the shape needs to be defined only once. Subsequent mouse clicks will measure the same shape centered at the new location.

Note: The irregular shape is automatically prevented from being placed too close to the edge of the image.

During this mode, pressing Ctrl-V will permanently mark the selected area.

Auto find spots (2D Gel Densitometry):

This method is specifically designed for 2D gels. It performs a watershed partitioning to identify the spots. Spots that are incompletely separated are usually separated and measured correctly.

2D gel partitioned with Auto Find Spots function.

This method is a perfect match for automatic background selection (Sec. 8.11.3). For example, in the image below, the large spot in the lower left has been identified as a single spot, as shown by the black partition lines. If Automatic Background Calculation is selected, Imal will take all the spots inside this irregular region and use fuzzy logic to sub-partition them into foreground and background pixels. The fuzzy logic takes into account the fact that part of each background pixel also contains a variable amount of the foreground. These background pixels are then combined into a highly accurate background value, which is then subtracted from the signal in the spot. This results in a measurement that is unaffected by variations in the background level in different parts of the gel.

There are a few tricks that can be used to improve the results:

1. The method is very sensitive, and often finds spots that are not visible to the naked eye. If there is noise in the image, it may find a large number of very small spots. These can be eliminated by applying the noise filter prior to densitometry.
2. Converting the image to 8 bits/pixel also reduces the number of spots that are detected.
3. Imal sorts the spots in increasing size. You may notice that the last entry in the list is much larger than the others. This is not a real spot, but a product of the partitioning method. It represents the area of the entire image.
4. The partitioning method is also an excellent match for the automatic background selection (Sec. 8.11.3), which uses fuzzy logic to further partition each watershed region into foreground and background. This increases the accuracy of the background estimate by about a factor of ten, to well below sub-pixel-value levels.