The identification of the carcinogenic risk of chemicals is currently mainly based on animal studies. The in vitro Cell Transformation Assays (CTAs) are a promising alternative to be considered in an integrated approach. CTAs measure the induction of foci of transformed cells. CTAs model key stages of the in vivo neoplastic process and are able to detect both genotoxic and some non-genotoxic compounds, being the only in vitro method able to deal with the latter. Despite their favorable features, CTAs can be further improved, especially reducing the possible subjectivity arising from the last phase of the protocol, namely visual scoring of foci using coded morphological features. By taking advantage of digital image analysis, the aim of our work is to translate morphological features into statistical descriptors of foci images, and to use them to mimic the classification performances of the visual scorer to discriminate between transformed and non-transformed foci. Here we present a classifier based on five descriptors trained on a dataset of 1364 foci, obtained with different compounds and concentrations. Our classifier showed accuracy, sensitivity and specificity equal to 0.77 and an area under the curve (AUC) of 0.84. The presented classifier outperforms a previously published model.

A comprehensive statistical classifier of foci in the cell transformation assay for carcinogenicity testing / G. Callegaro, K. Malkoc, R. Corvi, C. Urani, F.M. Stefanini. - In: TOXICOLOGY IN VITRO. - ISSN 0887-2333. - 45:3(2017), pp. 351-358. [10.1016/j.tiv.2017.04.030]

A comprehensive statistical classifier of foci in the cell transformation assay for carcinogenicity testing

F.M. Stefanini
Ultimo
2017

Abstract

The identification of the carcinogenic risk of chemicals is currently mainly based on animal studies. The in vitro Cell Transformation Assays (CTAs) are a promising alternative to be considered in an integrated approach. CTAs measure the induction of foci of transformed cells. CTAs model key stages of the in vivo neoplastic process and are able to detect both genotoxic and some non-genotoxic compounds, being the only in vitro method able to deal with the latter. Despite their favorable features, CTAs can be further improved, especially reducing the possible subjectivity arising from the last phase of the protocol, namely visual scoring of foci using coded morphological features. By taking advantage of digital image analysis, the aim of our work is to translate morphological features into statistical descriptors of foci images, and to use them to mimic the classification performances of the visual scorer to discriminate between transformed and non-transformed foci. Here we present a classifier based on five descriptors trained on a dataset of 1364 foci, obtained with different compounds and concentrations. Our classifier showed accuracy, sensitivity and specificity equal to 0.77 and an area under the curve (AUC) of 0.84. The presented classifier outperforms a previously published model.
Bayesian Trees; Statistical image descriptors; Foci classification; In vitro carcinogenicity testing; BALB/c 3T3 Cell Transformation Assay; Toxicology
Settore SECS-S/01 - Statistica
Article (author)
File in questo prodotto:
File Dimensione Formato  
2017-A comprehensive statistical classifier_foci.pdf

accesso riservato

Tipologia: Publisher's version/PDF
Dimensione 720.93 kB
Formato Adobe PDF
720.93 kB Adobe PDF   Visualizza/Apri   Richiedi una copia
Pubblicazioni consigliate

Caricamento pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2434/834835
Citazioni
  • ???jsp.display-item.citation.pmc??? 0
  • Scopus 4
  • ???jsp.display-item.citation.isi??? 3
social impact