January 9, 2019. Article on fully automated interpretation of the dicentric chromosome assay for radiation quantification now available

Our paper, “RADIATION DOSE ESTIMATION BY COMPLETELY AUTOMATED INTERPRETATION OF THE DICENTRIC CHROMOSOME ASSAY”  is now published in the journal Radiation Protection Dosimetry.

Unfortunately, the journal has not made the article open access. We have made it available on our ADCIWiki website, as permitted by the copyright agreement.

The link to the pdf full text is:

http://adcidewiki.cytognomix.com/files/pubs/LiRPD2019.pdf

July 13, 2014. Presentations at the 2014 American Society of Human Genetics Conference

Cytognomix will be presenting several papers at the upcoming ASHG annual meeting (October 18-22, 2014, San Diego):

Using information theory to analyze and predict splicing mutations in rare and common diseases: performance and best practices. N.GCaminsky, E. Mucaki and P.K. Rogan

Reversing differences in chromatin accessibility that distinguish homologous mitotic metaphase chromosomes. W.A. Khan, P.K. Rogan, J.H.M. Knoll

Automated Dicentric Chromosome Identification by Machine Learning-based Image Processing. P.K. Rogan, Y. Li, A. Subasinghe, J. Samarabandu, R. Wilkins and J.H. Knoll

Towards the minimal breast cancer genome and its relevance to chemotherapy. S.N. Dorman, J.H. Knoll, K. Baranova, C. Viner, P.K. Rogan

The FANCM c.5791C>T nonsense mutation (rs144567652) induces exon skipping and is a risk factor for familial breast cancer. Paolo Peterlongo ,  Francesca Damiola, Eliseos Mucaki,  Valentina Dall’Olio ,Sara Pizzamiglio  , Irene Catucci ,  Anders Kvist , Paolo Verderio, Mara Colombo , Loris Bernard ,  Hans Ehrencrona, Laura Caleca, Valeria Pensotti , Sylvie Mazoyer, Peter K. Rogan ,Paolo Radice

 

Please contact us if you would like to meet or discuss this work.

April 22, 2014. New paper describing Automated Biodosimetry Software published

We have published an article in Radiation Protection Biodosimetry describing our patented Automated Dicentric Chromosome Identifier Software for both Desktop and Supercomputer systems. The citation is:

Peter K. Rogan,  Yanxin Li,  Asanka Wickramasinghe,  Akila Subasinghe, Natasha Caminsky,  Wahab Khan,  Jagath Samarabandu,  Ruth Wilkins, Farrah Flegal, and Joan H. Knoll.  AUTOMATING DICENTRIC CHROMOSOME DETECTION FROM CYTOGENETIC BIODOSIMETRY DATA. Radiat Prot Dosimetry.  first published online April 21, 2014 doi:10.1093/rpd/ncu133  (Rogan et al. Radiat Prot Dosimetry. 2014).

This paper was presented at the EPR Biodose 2013 meeting in Leiden, Netherlands.  The software identifies highly variable features in a large quantity images in relatively short time frame. Multiple technologies are employed, including SVM machine learning, gradient vector flow, parallelization, and other methods.

January 28, 2013. Platform presentation on Automated Dicentric Chromosome Identifier Software

“Automating Dicentric Chromosome Detection from Cytogenetic Biodosimetry Data” at  the International EPRBioDose 2013 Conference in Leiden, Netherlands (March 24-28).

Authors: Peter Rogan(1,2), Akila Subasinghe(1), Asanka Wickramasinghe(1), Yanxin Li(1), Jagath Samarabandu(1), Joan Knoll(1,2), Ruth Wilkins(3), Farah Flegal(4); (1)University of Western Ontario, (2)Cytognomix Inc., (3)Health Canada, (4)Atomic Energy of Canada Ltd., Canada.

Abstract:  We are developing a prototype software system with sufficient capacity and speed to estimate radiation exposures by counting dicentric chromosomes in metaphase cells from many individuals in the event of a
mass casualty. Top-ranked metaphase images are segmented by defining chromosomes with an active contour gradient vector field (GVF), and by determining centromere locations along the centerline. The centerline is
extracted by Discrete Curve Evolution (DCE) skeleton branch pruning and curve interpolation. Centromere detection minimizes the global width and DAPI-staining intensity profiles along the centerline. A second
centromere is identified by reapplying this procedure after masking the first. Dicentrics can be identified by applying a support vector machine-based classification, which uses features that capture width and intensity
profile characteristics as well as local shape features of the object contour at candidate pixel locations. The correct location of the centromere is also refined in chromosomes with sister chromatid separation. The
overall algorithm has both high sensitivity (85%) and specificity (94%). Results are independent of the shape and structure of chromosomes in different cells, regardless of which laboratory protocol is followed or the
specimen source. The requisite throughput is being achieved by recoding MATLAB software modules for different segmentation functions in C++/OpenCV, and integrating them in the prototype. Processing of

numerous images is accelerated by both data and task software parallelization with the Message Passaging Interface and Intel Threading Building Blocks as well as an asynchronous non-blocking I/O strategy. Relative
to a serial process, metaphase ranking, GVF, and DCE are respectively 100 and 300 fold faster on an 8-core I7-based desktop and on a 64-core shared memory cluster computer. Extrapolation from these benchmarks to
a 64-core system in which all of the software modules have been integrated indicates that it should be feasible to process metaphases for dicentric chromosomes from 1000 specimens in 20 hours.