Computer Vision News - July 2016

Every month, Computer Vision News reviews a successful project. Our main purpose is to show how diverse image processing applications can be and how the different techniques can help to solve technical challenges and physical difficulties. This month we review a Chromosome Classification software developed by RSIP Vision’s engineers to determine a person’s karyotype and detect the presence of pathologies. Do you have a project in computer vision and image processing? Contact our consultants . Background One of the most classic challenges in computer vision is chromosome classification: this area has seen more than 30 years of software development and, in a certain way, all the history of computer vision is reflected in this task. Testing and classification of chromosomes plays a key role in our ability to understand genetic pathologies. Along the years, the most prominent approach used to meet this challenge made use of statistical methods comparing the patient’s chromosomes to the ideal set of healthy chromosomes. This is not very trivial, since the aspect of chromosomes is variable depending on the development of the nucleus and the phase of the cell’s mitosis. Algorithms which were used for years would accordingly detect the bands found on the surface of the chromosome and compare them to same dictionary or some ideal modelling. Of course, statistic similarity can be a very strong tool and has historically provided successful classification packages. Solution But nowadays, as everyone knows, our first guess to do anything in computer vision will be Deep Learning , and we have to examine how deep learning can help us do classification in this project. Of course, we can expect that by training on a large enough sample of chromosomes we will be able to get a good detection. But this is true only in ideal cases. In real-life situations 38 Computer Vision News Project Karyotyping: Chromosome Classification Set of chromosomes (some of them overlapping) “ Pick the most obvious and readily available opportunities ” Project