Introduction:

Traditionally, colon polyps’ measurements have been empirically estimated visually and with biopsy forceps, but neither method is inaccurate. Other methods have been studied but have not had the accuracy expected. The research reported here was undertaken to address this issue by building an algorithm for measure polyps from photographs taken through a colonoscope.

Materials and methods:

The study was done in three phases. First, an algorithm was built in MATLAB, and photos taken with a colonoscope were stored in the JPG format. In the second phase, images of objects with known sizes were checked against the algorithm to verify its accuracy. After verification of the algorithm’s accuracy, photographs of colon polyps were measured using the algorithm. In phase 3, images of polyps previously sent to three experts were measured with the algorithm. All photographs were taken with an Olympus Exera II Colonoscope.

Results:

For objects smaller than 5 mm, the algorithm overestimated sizes by 0.11 to 0.08 mm. For those greater than 5 mm, it overestimated sizes by 0.25 mm to 1.76 mm in those of 22 mm. The experts seriously overestimated sizes. They estimated that 7 mm polyps measured 12 mm, 8 mm polyps measured 15 mm, and 9 mm polyps measured 18mm.

Conclusion:

The algorithm developed is sufficiently accurate for measuring colon polyps and is easy to obtain and relatively easy to use. It could become a tool for overcoming the difficulty of measuring polyps during a colonoscopy.

        · abstract in Spanish     · text in English | Spanish     · English ( pdf ) | Spanish ( pdf )