Archived page: ADCIS Applications: Examples: Bone Analysis

Study of the porosity of the bone

Bone Analysis

The Department of Medicine and Therapeutics, Medical School at University of Aberdeen, Scotland, is actively pursuing research to study the biology of bone and bone disease. Rob van't Hof, one of the leading Professor of that Department, used the Aphelion™ ActiveX components to develop a turn-key software system capable of detecting cells responsible for the removal of bone, also called the bone resorption, or the osteoclast.

The Department of Medicine and Therapeutics is actively working on the Resorption Pit Assay. In this assay, osteoclasts are grown on small, thin slices of bone. If the osteoclasts are active, they will excavate a small resorption pit which can be visualized with a reflected light microscope as distinct dark objects. Using the Aphelion ActiveX components, a custom application was developed to quantify the amount of resorption.

Live Image

A live image is captured with a regular black and white, CCD camera, and the pits are identified according to several parameters, such as their gray-scale intensity, their size and shape. A straight threshold cannot be performed on the gray-scale image, as the background is not uniform. Frequency filtering techniques are then used to cope with this problem.

Threshold performed on the original image

Using FFT filtering or a convolution with a large kernel, pits can be detected. During the final step, any remaining noise is manually removed and the area of the resorption pits is measured. Results are then written to an Excel compatible file for later analysis.

Threshold after performing an FFT

Final result

Using the powerful Aphelion ActiveX components, such as the Frequency Filtering Toolkit which includes the FFT and the Object Filtering operator available in Object Processing Toolkit, the software which was developed by the Department of Medicine and Therapeutics is much faster than a previous software developed by the group, and gives more accurate results. In addition, as minimum user interaction is required to edit the resulting image, results are more reproducible, and again, more accurate.