Breast cancer is the most widespread cancer affecting women globally, with several contributing factors, including breast density. Women with denser breasts, characterised by a higher proportion of glandular and fibrous tissue, face a significantly elevated risk of developing breast cancer, almost four to six times higher compared to women with lower densitybreasts.Earlyidentificationofhighbreastdensityiscrucialforenablingwomento receiveappropriatemonitoringandpotentialpreventativeinterventionsbeforethepossible development of cancer. Owing to its significance in evaluating the risk of breast cancer, breast density has become a critical factor considered in clinical evaluations.
While various imaging modalities, including X-rays (mammograms), CT scans, ultrasound, and MRI, are employed for breast analysis, mammograms remain the primary tool used by radiologists for routine breast cancer screening. Currently, the radiologist manually
Evaluates the breast density and classifies the mammogram into one of the four classes based on breast composition.
Breast composition classification according to BI-RADS estimation is an effective method for identifying women who have dense breasts and are more likely to get breast cancer. The subjective analysis of breast density and its classification leads to inter-rater and intra-rater variability. The classification depends on the radiologist’s interpretation of the mammogram. Different radiologists may classify the same mammogram differently; they will have a unique style of segmentation, which leads to inter-rater and intra-rater variability in the classification. There is no standardisation in how radiologists classify the images, leading to inconsistency in the reporting and communication of results. The existing software, like Quantra, Volpara, and Densitas, are very expensive, so the radiologist chooses subjective measurement, especially in developing countries.
In our case, we employ a three step technique. In the first step, we detect the breast region and segment it from the rest of the image. In the second step, we segment the breast tissues from the fatty tissue in the breast. In the third step, utilising the breast tissue volume and other factors, we classify the breast into the four types, based on breast tissue density.
Proposed Future Work Details
Currently, we have completed the extraction of the breast region from the breast mammograms and segmentation of the breast tissues therein. However, we need to take the next step and classify the breasts into the different types automatically, based on the density.
