Automated Lung Measurements from Microscope Images:
Denver Children’s Hospital
Original image of a stained section of a rat lung
showing lung alveoli and air spaces.
The yellowing at the right edge is due to
uneven microscope illumination.
Calibration
A Results Table is automatically generated
expressing the data workup according to the
researcher's specifications.
expressing the data workup according to the
researcher's specifications.
1. Thresholding the green slice of the RGB image, shows heavy background
interference due to uneven microscope illumination.
2. Using a rolling ball background subtraction algorithm, the entire image can
be thresholded according to a histogram analysis routine that determines
threshold points based on uniform histogram features.
3. Either the septal dimensions or the air spaces can be measured with
ImageJ's Analyze Particles routine using a simple inversion of the image.
4. Making the image binary permits binary processing to greatly reduce artifacts.
1. Small scale details in the image of the septa cancause many artifacts when the image is Skeletonized.
2. Smoothing algorithms (such as erode-dilate cycles)
simplify the skeletonized image.
Results
A Results Table is automatically generated
expressing the data workup according to the
researcher's specifications.
expressing the data workup according to the
researcher's specifications.