Specific spectral sub-images for machine learning evaluation of optical differences between carbon ion and X ray radiation effects.

Autor:	Negoita RD; Applied Sciences Doctoral School, National University of Science and Technology Politehnica Bucharest, 313 Splaiul Independentei, Bucharest, 060042, Romania., Ilisanu MA; Doctoral School of Automatic Control and Computers, National University of Science and Technology Politehnica Bucharest, 313 Splaiul Independentei, Bucharest, 060042, Romania.; Holographic Imaging and Processing Laboratory, Physics Department, Faculty of Applied Sciences, National University of Science and Technology Politehnica Bucharest, 313 Splaiul Independentei, Bucharest, 060042, Romania., Irimescu IN; Applied Sciences Doctoral School, National University of Science and Technology Politehnica Bucharest, 313 Splaiul Independentei, Bucharest, 060042, Romania.; Tehnoplus Medical SRL, 1 Odobesti str, Bucharest, Romania., Popescu RC; Department of Life and Environmental Physics, Horia Hulubei National Institute for R&D in Physics and Nuclear Engineering, Reactorului 30, P.O. Box MG-6, 077125 Magurele, Romania.; Department of Bioengineering and Biotechnology, Faculty of Medical Engineering, National University of Science and Technology Politehnica Bucharest, G. Polizu Street, 1-7, 011061 Bucharest, Romania., Tudor M; Department of Life and Environmental Physics, Horia Hulubei National Institute for R&D in Physics and Nuclear Engineering, Reactorului 30, P.O. Box MG-6, 077125 Magurele, Romania.; Faculty of Biology, University of Bucharest, 91-95 Splaiul Independentei, 050095 Bucharest, Romania., Mihailescu M; Holographic Imaging and Processing Laboratory, Physics Department, Faculty of Applied Sciences, National University of Science and Technology Politehnica Bucharest, 313 Splaiul Independentei, Bucharest, 060042, Romania.; Research Centre in Fundamental Sciences Applied in Engineering, National University of Science and Technology Politehnica Bucharest, 313 Splaiul Independentei, Bucharest, 060042, Romania., Scarlat EN; Holographic Imaging and Processing Laboratory, Physics Department, Faculty of Applied Sciences, National University of Science and Technology Politehnica Bucharest, 313 Splaiul Independentei, Bucharest, 060042, Romania., Pleava AM; CAMPUS Research Centre, National University of Science and Technology Politehnica Bucharest, 313 Splaiul Independentei, Bucharest, 060042, Romania., Dinischiotu A; Faculty of Biology, University of Bucharest, 91-95 Splaiul Independentei, 050095 Bucharest, Romania., Savu D; Department of Life and Environmental Physics, Horia Hulubei National Institute for R&D in Physics and Nuclear Engineering, Reactorului 30, P.O. Box MG-6, 077125 Magurele, Romania.
Jazyk:	angličtina
Zdroj:	Heliyon [Heliyon] 2024 Aug 02; Vol. 10 (15), pp. e35249. Date of Electronic Publication: 2024 Aug 02 (Print Publication: 2024).
DOI:	10.1016/j.heliyon.2024.e35249
Abstrakt:	Advances in radiotherapy, particularly the exploration of alternative radiation types such as carbon ions have updated our understanding of its effects and applicability on chondrosarcoma cells. Here we compare the optical effects produced by carbon ions (CI) and X-rays (XR) radiations on chondrosarcoma cells nuclei and set an automated method for evaluating the radiation-induced alterations without the need of chemical marking. Hyperspectral images (HSI) of SW1353 chondrosarcoma line carry detectable optical changes of the cells irradiated either with CI or XR compared to non-irradiated ones (REF). The differences between the spectral profiles of CI, XR and REF nuclei classes led to partitioning the HSIs into spectral sub-images. The changes are detected by support vector machine (SVM) classifiers whose performances are evaluated by the most used point metrics: sensitivity ( SEN ), accuracy ( ACC ), and precision ( PREC ), applied on spatial feature values. Specific interaction mechanisms by radiation type reveal distinct subintervals where HSIs changes are more prominent, and the classifiers perform at best. For CI the best classifiers are obtained for sub-images in the interval (424-436 nm), while for XR the best classifiers are obtained for sub-images in the interval (436-445 nm). The classifiers work better with texture features than roughness features in both cases. The classifier with the best SEN point metric in the testing phase is the most suitable to measure the irradiation efficiency irrespective of the radiation type. The altered nuclei are easier to discriminate when irradiated with CI than with XR. The study proves that SVM with optical data offers a rapid, automated, and label-free method for evaluating radiation-induced alterations in chondrosarcoma nuclei, thereby enabling effective analysis of extensive data. Competing Interests: The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. (© 2024 The Authors.)
Databáze:	MEDLINE
Externí odkaz:	Zobrazit plný text záznamu