brain (1).jpg

blog

 
 
 
 

HOW Can AI help fighting Breast Cancer?

From: https://www.mayoclinic.org/diseases-conditions/breast-cancer/symptoms-causes/syc-20352470

From: https://www.mayoclinic.org/diseases-conditions/breast-cancer/symptoms-causes/syc-20352470

Our society is investing an enormous effort to fight breast cancer. Yet, every year about 2 million new cases are reported. While according to the National Breast Cancer Fondation 1 in 8 American women will be diagnosed with breast cancer during their lifetime, every year over 500,000 women are killed by this desease worldwide (Global Health Estimates, WHO 2013). We are talking about 1,400 women per day, 58 women per hour, 1 woman every minute. By the time you finish to read this article, more than 3 humans have lost their lives because of breast cancer.

We are constantly running out of time. We need a breakthrough that can simultaneously (1) speed up the retrospective research, (2) allow early diagnosis and (3) enhance the treatments. Artificial Intelligence is the best candidate to achieve these goals. The availability of powerful computational resources together with the huge quantity of data collected in the last decades makes machines unbeatable in terms of speed and costs, helping also smaller players to enter in the stage and contribute to this breaktrouth.

Since a few years, the Regina Barzilay Group at the Computer Science and Artificial Intelligence Lab of MIT has focused its attention on applying AI method to fight breast cancer, following three major lines of research:

  • extracting and structuring information from free-text reports to allow retrospective research (1) on large cohorts. This means that studies can now be based on thousands of patients rather than on tens (or, in the best cases, on hundreds). This not only increases the significance of the findings, but it also allows the identification and correction of biases. [SEE: 1, 2, 3]

  • predicting cancer risk directly from full resolution mammographies, rather than from synthetic radiologist assessments (e.g. breast density, for which experts have very low agreement rates). According to WHO, the early diagnosis (2) is the cornerstone for successful treatments. Adopting AI would allow hospitals and clinics to focus their resources on patients at higher risk, while organizing periodic screenings for the others. [SEE: 4]

  • utilizing information extracted from papers, documents and knowledge bases to model and discover new compounds, which may one day become more effective and cheaper drugs. [IF INTERESTED, FIND A DRUG DISCOVERY WORKSHOP HERE: https://www.linkedin.com/company/mit-aidm-conference/]

Althogth these three directions are not exhaustive of all the possible applications of AI to cancer prevention, diagnosis and treatment, they clearly exemplify how AI can be utilized to enhance our healthcare system and improve life quality. Similar systems to those described above are currently being utilized for other types of cancer, affecting bones or other organs (e.g. prostate, lung and pancreas). [SEE: 5]

Enrico Santus




REFERENCES

  1. Adam Yala, Regina Barzilay, Laura Salama, Molly Griffin, Grace Sollender, Aditya Bardia, Constance Lehman, Julliette M Buckley, Suzanne B Coopey, Fernanda Polubriaginof, Judy E Garber, Barbara L Smith, Michele A Gadd, Michelle C Specht, Thomas M Gudewicz, Anthony J Guidi, Alphonse Taghian, Kevin S Hughes. 2017. Using machine learning to parse breast pathology reports.

  2. Enrico Santus, Clara Li, Adam Yala, Donald Peck, Rufina Soomro, Naveen Faridi, Isra Mamshad, Rong Tang, Conor R Lanahan, Regina Barzilay, Kevin Hughes. 2019. Do Neural Information Extraction Algorithms Generalize Across Institutions?

  3. Francisco Acevedo, Victor Diego Armengol, Zhengyi Deng, Rong Tang, Suzanne Coopey, Emanuele Mazzola, Conor Lanahan, Danielle Braun, Adam Yala, Regina Barzilay, Clara Li, Enrico Santus, Amy Colwell, Anthony Guidi, Curt Cetrulo, Judy Ellen Garber, Barbara L Smith, Tari A King, Kevin S Hughes. 2019. Incidental atypical hyperplasia/LCIS in mammoplasty specimens and subsequent risk of breast cancer.

  4. Adam Yala, Constance Lehman, Tal Schuster, Tally Portnoi, Regina Barzilay. 2019. A Deep Learning Mammography-based Model for Improved Breast Cancer Risk Prediction.

  5. Enrico Santus, Susu Yan, Adam Yala, Leonard Wee, Jason Kim, Santiago A. Lozano-Calderon, Joseph H. Schwab, Regina Barzilay, Kevin S. Hughes and Karen De Amorim Bernstein. 2019. Using Deep Learning Algorithm to Extract Tumour Information from Pathology and Surgical Notes of Chondrosarcoma Patients.