Proposal: Matching of Mammographic Lesions in Different Breast Projections

Figure 1. Lesion matching in different views of the same breast. Matching requires lesions' positions to be known.


Introduction:

Worldwide, breast cancer is the most frequently diagnosed and most lethal form of cancer in women. One way to address the disease is to diagnose patients early as it frequently translates into a better prognosis. As such, many countries implement programs that screen asymptomatic women who are over a certain age. These programs are often based on screening mammography, an exam in which two images (views) are obtained from each breast (see Figure 1). Current state-of-the-art algorithms use these images to detect lesions that are indicative of breast cancer. However, the way they fuse information between the two views is naive. While radiologists often analyze the same lesion in the two views, algorithms fuse the decision at a much higher level, often averaging the two images' diagnosis. Addressing this can decrease the errors made by computer-aided diagnosis tools, ultimately helping patients get a more accurate diagnosis.



Tasks (1st semester):
  • Review the State-of-the-art: multi-view in mammography breast cancer screening;
  • Get acquainted with the mammographic data;
  • Evaluate the baseline model provided;
  • Propose different options for either the improvement of the baseline model or the developing a new one;
  • Write the report.
Tasks (2nd semester):
  • Implement the options proposed;
  • Evaluate the implemented algorithms and compare;
  • Elect a final implementation and evaluate its potential value in a broader setting (e.g. lesion detection).
  • Write the dissertation.


Advantages:
  • Data is available from day 1 - already collected;
  • Strong Support - this is an active area of research from the group;
  • There is already some preliminary work done;
  • Encoragement to publish - important if you are planning on pursuing a PhD;
  • Eligibility to CTM's Best Master Thesis 2021 Award;


Important skills (need to be learned by the 1st semester):
  • Programming (python)
  • Computer Vision
  • Deep Learning
  • Medical Image Analysis
  • Computer-Aided Diagnosis
References:

    [1] - Ribli, D., Horváth, A., Unger, Z. et al. Detecting and classifying lesions in mammograms with Deep Learning. Sci Rep 8, 4165 (2018). https://doi.org/10.1038/s41598-018-22437-z
    [2] - Schaffter T, Buist DSM, Lee CI, et al. Evaluation of Combined Artificial Intelligence and Radiologist Assessment to Interpret Screening Mammograms. JAMA Netw Open. 2020;3(3):e200265. https://doi.org/10.1001/jamanetworkopen.2020.0265