Formulation and validation of a quantitative method to estimate mammographic breast density [Abstract]

Weerakoon, Bimali Sanjeevani ORCID: https://orcid.org/0000-0003-0843-6389, Lakmini, Chandrasiri, Hewavithana, Padma, Jayasinghe, Ayesh, Gunathilake, Prabhath and Abeysundara, Sachith 2021. Formulation and validation of a quantitative method to estimate mammographic breast density [Abstract]. Presented at: iPURSE 2021 : The Signature Conference of the University of Peradeniya, Peradeniya, Sri Lanka, 11-12 November 2021. Proceedings Peradeniya University International Research Sessions (iPURSE) 2021. , vol.23 Sri Lanka: University of Peradeniya Peradeniya, p. 308.

Full text not available from this repository.

Abstract

Breast density estimation is important due to its strong association with breast cancers. This study aimed to formulate and validate a computer-based programme for the quantitative assessment of mammographic breast density (MBD). This semi-automated computer-based programme was developed to estimate the MBD using the ratio of the area of fibroglandular tissues to the total area of the breast in a digital mammogram where a manual threshold was applied to differentiate fibroglandular tissues from fatty tissues. As the first part of the validation, three experts applied the threshold individually on hundred mammograms and the agreement between the experts was assessed. For the second part of the validation, 400 digital mammograms categorized according to the Breast Imaging-Reporting and Data System (BI-RADS)were obtained from an online database. The mammograms consisted of left craniocaudal (LCC), left mediolateral oblique (LMLO), right craniocaudal (RCC) and right mediolateral oblique (RMLO) projections. The agreement between assigned BI-RADS categories and the measured MBD values were assessed for the above 400 mammograms. Results of the analysis, intra-class correlation coefficients (ICC), showed good agreements among three experts for LCC (ICC=0.858), LMLO (ICC=0.826) and RCC (ICC=0.752) projections while RMLO (ICC=0.272) projection showed a poor agreement. Significant positive correlations (polyserial correlation coefficients) were identified between the BI-RADS categories and the estimated MBD of LCC (r=0.8164, p<0.05), LMLO (r=0.8668, p<0.05), RCC (r=0.8512, p<0.05) and RMLO (r=0.8409, p<0.05). In conclusion, significant positive correlations were obtained for the measured MBD values and qualitative BI-RADS categories. Further, a good agreement among the experts indicates, less subjectivity and optimal reproducibility. Hence, the computer-based programme formulated in this study may be utilized to accurately determine the MBD in digital mammograms.

Item Type:	Conference or Workshop Item (Paper)
Date Type:	Publication
Status:	Published
Schools:	Healthcare Sciences
Publisher:	University of Peradeniya Peradeniya
ISBN:	978-624-5709-07-6
Last Modified:	31 Jan 2025 15:17
URI:	https://orca.cardiff.ac.uk/id/eprint/175617

Actions (repository staff only)

Edit Item