Detection of Out-of-Focus Regions

Precise focusing is one of the important parameters to obtain high-quality images. Unfortunately, whole slide images produced by commercial scanners may result in out-of-focus areas. The main reason for the blurred images is the improper focusing on the tissue. Although the scanners are able to calculate the precise focusing parameter for a specific point, the regions between two focus points can be out-of-focus due to the 3d structure of the tissue. The amount of the blurring will depend on the difference between the estimated and real distances in the z-axis for this region. As a result, for very small offset values the blurring will be negligible. On the other hand, for larger offset values, the blurring will become more noticeable. During the last decades, several quality assessment approaches were proposed for whole slide images (WSI). Most of these approaches aim to identify the quality of the acquired image in terms of focusing.

In this plugin, we aim to allow users to find out-of-focus regions using Sedeen Viewer. The plugin is developed in C++. In order to avoid analyzing the regions outside the tissue area, the plugin starts with finding the tissue map at 1x magnification of the entire slide by applying Otsu Thresholding. After tissue detection, the plugin reads 64x64 tiles inside the tissue and calculates a focus score for each tile based on the entropy and contrast of the tile. The user is able to define the sampling rate during this analyzes. A higher sampling value means a denser map and causes more computation. The plugin also includes a cache mechanism and this function allows the user to use the previous analysis.