Janan, Faraz and Brady, Michael (2018) Tracking ‘developing’ Focal Densities in Breast Quadrants. In: NCRI 2018, Nov 4 - Nov 6, Glasgow.
Documents |
|
![]() |
PDF
NCRI Abstract 2.pdf - Whole Document 724kB |
Item Type: | Conference or Workshop contribution (Presentation) |
---|---|
Item Status: | Live Archive |
Abstract
Abstract
Background
Focal density (FD) is a dense mammographic region that cannot be accurately identified as a mass without further examination. If a particular breast quadrant is significantly dense than others or has an increase in density over time, this could be associated with neoplasm especially in the presence of a tangible mass. We have developed a method to study and track quadrant-wise increase in FD over time.
Method
A set of 10 temporal patient cases collected over a period of up to 6 years were used. Each quarter of the breast is assigned a FD score, where quadrants are defined by first differentiating a border between the breast region and skin line. Then a nipple detection method is used to correctly identify nipples, including those ‘not in profile’. Afterward, the nipple location is used as a reference point to divide the breast into four quarters (see Figure 2). Further on, FDs are quantified [1] and a score assigned to each quadrant of the breast, and to the breast as a whole.
Results
Results show that our method can detect increase in FD over time in some quarters of breast; a finding that can be verified by Volpara density grade [2]. It can be seen (Figure-2) that Q1-left (upper-interior-UI) has a significantly higher FD score as compared others. Clinical evaluation for this BIRADS-C mammogram (Figure-1, left craniocaudal) confirms the presence of 6mm grade-4 screen detected invasive lobular carcinoma in the left UI quadrant of the breast. Figure-3 shows a FD comparison of all quadrats of the bilateral pair over the course of 6 years. Q1-left remained the densest throughout.
Conclusion
The study suggests that tracking FD (both ‘developing’ and ‘stable’) over time could potentially help in better understanding of the risk of breast cancer development in any particular quadrant of the breast.
Keywords: | breast cancer, early diagnosis, breast density, image processing, computer vision, mammograms |
---|---|
Subjects: | B Subjects allied to Medicine > B800 Medical Technology G Mathematical and Computer Sciences > G400 Computer Science |
Divisions: | College of Science > School of Computer Science |
ID Code: | 34104 |
Deposited On: | 14 Nov 2018 09:20 |
Repository Staff Only: item control page