This site is devoted to mathematics and its applications. Created and run by Peter Saveliev.

Ophthalmology diagnosis

From Mathematics Is A Science

Jump to: navigation, search

This problem came from a medical doctor/researcher.

Some preliminary results after some experimenting with Pixcavator. In cell test 10, the amount of dark is about 43%. In cell test 11, I also found a fairly good contour. There was some guessing on my part. For example, in cell test 10, the change from white to black is very gradual, so there are many ways we can separate dark from light (see Objects in gray scale images). To figure out precisely how to use our software to solve the problem, we need to understand better what you want to get from the image. To do this right, we'd have to start with manual analysis. Next step is to outline in the image, by hand, the contours of the features you need to be counted or measured (see Pixcavator technical support).

eye atrophy - original atrophy captured by the green contours

eye atrophy - another image atrophy captured by the green contours

eye atrophy - another image, original very blurry image improved version, binary

The image was "posterized".

eye atrophy - another image atrophy captured by not quite well atrophy captured - two regions

In the last image, the upper right subtle atrophy is included.

The link is to a poster presentation made at the last meeting of Association for Research in Vision and Ophthalmology (ARVO) by Dr. Nalin Mehta. The title (shortened) is Evaluation of Choroidal Circulation Using Collapsed C-Scan Imaging. The goal is "to better differentiate between the various components and sub-types of age-related macular degeneration". Further:

En face (C-scan) OCT images, collapsed and summated in the antero-posterior axis, were acquired for all patients over approximately monthly intervals throughout their treatment regimen. These images were standardized and analyzed using Pixcavator 3.1 (Intelligent Perception Co., Huntington, WV), an image analysis program which first captures the contours of the choroidal vascular pattern, differentiating the same from background scatter phenomenon, and then quantifies this pattern in proportion to the entire scanned area, producing an Excel (Microsoft Corp., Sacramento, CA) spreadsheet with locations and measurements of these vascular structures.


Other image analysis examples