Automated recognition of ROIs for breast thermograms of lateral view-a pilot study. Issue 2 (3rd July 2018)

• Record Type: Journal Article
• Authors: Josephine Selle, J.; Shenbagavalli, A.; Sriraam, N.; Venkatraman, B.; Jayashree, M.; Menaka, M.
• Is Part Of: Quantitative infrared thermography. Volume 15:Issue 2(2018)
• Page Start: 194
• Page End: 213
• Publication Date: 2018-07-03
• Subjects: Breast thermal images -- lateral view -- anisotropic filter -- level set algorithm; Infrared technology -- Periodicals; Thermography -- Periodicals; Infrared technology; Thermography; Periodicals; 621.362
• DOI: 10.1080/17686733.2018.1426137
• Languages: English
• ISSNs: 1768-6733

Abstract

Abstract: This research study proposes an automated region of interest (ROI) for breast thermograms by considering lateral view as well as the frontal view of the breasts. The lateral view helps in contralateral comparison between the different quadrants of the breast and thus ensuring the analysis of the whole breast region including the lymph nodes and armpit areas. The proposed research involves in analysis of normal, lactating and fibro adenoma breast thermograms from a specific ethnic group located in the southern part of India. An optimized anisotropic diffusion filter is applied to preserve the vascular patterns inside the breast regions which appear to be significant edges in the thermal images. Then level set algorithm without re-initialization selects the contour boundary covering maximum area of the breast regions and segments along the canny detected edges which is enhanced through anisotropic diffusion filter. The resultant ROI helps the clinical specialist to conclude the nature of the biomarkers whether normal or abnormal at the preliminary screening level. Experimental simulation result shows maximum similarity ( ρ > 0.05) within the 95% confidence interval of correlation between the segmented and reference images. It can be concluded that the proposed automated ROIs confirm the suitability for clinical validation.