Application of the computer vision system for evaluation of pathomorphological images

What is it about?

Work of medical personal with images is one of sought-after skill with insufficient amount of specialists that realized in their overloading and possible. In that connection purpose of our work was implementation of the computer vision system for evaluation of pathomorphological images as pathologists are most scarce specialist in modern medicine. We performed programmed and manual study of pathomorpological slides (immunohistochemical and cytological) with application of machine vision systems for counting of selected objects and comparison with previously manual estimation. The software was written in the Python 2.7 programming language using the OpenCV library for other purposes was modified. Two features were used to determine the nuclei and cells: the characteristic color range and the ratio of the area of the object to the square of its perimeter. We obtain average relative error of the suggested soft version about 9.2%, so accuracy of detection of cancer markers is 90.8% that is sufficient for the initial examination of a patient with screening examination of large number of patients even in so difficult images as immunohistochemistry.

Featured Image

Why is it important?

In pathomorphological images by grouping pixels into distinct segments (referred to as objects) using the pixel intensity/color information or image texture biological meaningful objects and structures may be represented and used for an object-based analysis. The resulting image is translated into an electronic format, suitable for processing by software, after which these tools analyze the image, search for objects on it, determine the type of object, etc., depending on the task. There are no simple and available systems for image analysis till now.

Perspectives