Ett flexibelt system för att automatiskt följa celler över tiden optimerat för olika tillämpningar genom användarkontrollerad återkoppling
Tidsperiod: 2013-01-01 till 2015-12-31
Budget: 2 452 000 SEK
Obtaining quantitative data from live cell images is the key to understanding molecular and cellular processes. Automated microscopes enable the collection of images at a rate which outpaces researchers´ ability to visually inspect them. Automated image analysis has thus become necessary to extract data. Tracking in life science applications exhibits a vast diversity of studied specimens as well as imaging techniques. For an automated image analysis method to be able to accommodate for this diversity it has to be very flexible. Flexibility often comes at a price in terms of algorithmic parameters. These have to be optimized for each new application, which usually requires extensive knowledge of underlying algorithms.Our aim is to develop method to reduce time and algorithmic insight required for an investigator to preform automated tracking in life science applications. To achieve this we will (1) Use existing state-of-the-art algorithms for object detection and tracking (2) Use the investigator´s ability to manually classify a subset of detected objects and tracks as correct or incorrect. (3) Use a computer´s ability to optimize parameters based on iterative feedback from the investigator´s classification, and (4) Use the computer´s ability to quickly detect and track large numbers of individual cells based on the optimized parameter settings. We will test and validate developed tools on three collaborative projects and disseminate them to the broader community.