11702

A pattern recognition system for prostate mass spectra discrimination based on the CUDA parallel programming model

S. Kostopoulos, D. Glotsos, K. Sidiropoulos, P. Asvestas, D. Cavouras, I. Kalatzis
Department of Medical Instruments Technology, Technological Education Institute of Athens, Greece; School of Engineering and Design, Brunel University West London, Uxbridge, Middlesex, UB8 3PH, UK
J. Phys.: Conf. Ser. 490 012144

@inproceedings{kostopoulos2014pattern,

   title={A pattern recognition system for prostate mass spectra discrimination based on the CUDA parallel programming model},

   author={Kostopoulos, Spiros and Glotsos, Dimitris and Sidiropoulos, Konstantinos and Asvestas, Pantelis and Cavouras, Dionisis and Kalatzis, Ioannis},

   booktitle={Journal of Physics: Conference Series},

   volume={490},

   number={1},

   pages={012144},

   year={2014},

   organization={IOP Publishing}

}

Download Download (PDF)   View View   Source Source   

1626

views

The aim of the present study was to implement a pattern recognition system for the discrimination of healthy from malignant prostate tumors from proteomic Mass Spectroscopy (MS) samples and to identify m/z intervals of potential biomarkers associated with prostate cancer. One hundred and six MS-spectra were studied in total. Sixty three spectra corresponded to healthy cases (PSA < 1) and forty three spectra were cancerous (PSA > 10). The MS-spectra are publicly available from the NCI Clinical Proteomics Database. The pre-processing comprised the steps: denoising, normalization, peak extraction and peak alignment. Due to the enormous number of features that rose from MS-spectra as informative peaks, and in order to secure optimum system design, the classification task was performed by programming in parallel the multiprocessors of an nVIDIA GPU card, using the CUDA framework. The proposed system achieved 98.1% accuracy. The identified m/z intervals displayed significant statistical differences between the two classes and were found to possess adequate discriminatory power in characterizing prostate samples, when employed in the design of the classification system. Those intervals should be further investigated since they might lead to the identification of potential new biomarkers for prostate cancer.
No votes yet.
Please wait...

* * *

* * *

HGPU group © 2010-2024 hgpu.org

All rights belong to the respective authors

Contact us: