Artificial Intelligence is increasingly used to detect tumors. You can read the two following articles on this subject.
AI can also improve traditional treatments used to treat cancer : chemotherapy and radiotherapy.
AI also has a role to play in the emergence of a new form of treatment against tumors: immunotherapy.
AI for Chemotherapy
Chemotherapy is a treatment for cancer, which is based on the use of drugs. It aims to eliminate cancer cells wherever they are in the body. In 2017: 324,465 people in France were treated with chemotherapy. 3,000,000 hospitalizations mentioning chemotherapy were carried out in healthcare institutions.
As part of this treatment, Artificial Intelligence is used to measure the effectiveness of drugs in patients. It is used to determine the patient’s tolerance to drugs and to optimize dosages during chemotherapy.
Adrien Coulet, a lecturer at the University of Lorraine and a researcher in a joint Inria and Loria team, in collaboration with researchers at Stanford University, has developed an algorithm that uses electronic health data to predict (even before the start of treatment) whether patients will need a lower dose of medication, thus reducing the suffering caused by the side effects of chemo.
To analyze the data in the patient’s history, the scientific approach used is a method commonly used in machine learning: the “Random Forest” method. The algorithm proved effective in predicting dose reductions for 23 of the 34 drugs studied. However, it was ineffective in predicting dose increases.
AI for Radiotherapy
Radiotherapy is a method of treating cancer, using radiation to destroy cancer cells by blocking their ability to multiply. The goal of irradiation is to locally destroy all tumor cells while sparing healthy peripheral tissues.
In France, in 2017, 196,000 patients underwent just over 4.1 million radiotherapy sessions. Approximately 60% of cancer patients receive radiation treatment at some point in their disease.
During radiotherapy, the radiophysicist and/or the dosimetrist in conjunction with the radiation oncologist sets up a treatment plan that validates the dose and the irradiation technique according to anatomical data, the contours of the area to be irradiated and the organs to be protected.
Artificial Intelligence allows practitioners to automate and accelerate this work prior to the irradiation session. AI via the use of CNN (Convolutional Neural Network) takes between 10 and 20 minutes to perform the work compared to 4 to 5 hours previously.
“The automated treatment planning system works by training the algorithm to treatment plans of former patients. It is able to detect the former patients that most closely resemble a new patient and create a treatment plan for that new patient without any caregiver interaction other than pressing the Play button” Leigh Conroy, PhD, physics resident, at Princess Margaret Cancer Center, part of the University Health Network in Toronto.”
AI for Immunotherapy
Immunotherapy is a new treatment that, instead of directly attacking tumour cells like chemotherapy and radiotherapy, helps the immune system recognize and destroy them. Several strategies are already being used in the treatment of different cancers. The two most promising are immunomodulators and CAR-T cells.
- The immunomodulators remove tumor-induced immune system inhibition mechanisms. For example, pembrolizumab and nivolumab, block the PDL1/PD1 interaction between tumor cells and T lymphocytes that renders tumors invisible to T lymphocytes.
- CAR-T cells : the idea is to take white blood cells from the patient that specialize in recognizing and destroying pathogenic cells, then “improve” them by modifying their genetic makeup (CAR-T), make them multiply, and finally inject them back into the patient so that they destroy the tumor cells present in his or her body.
The effectiveness of immunotherapy remains limited because for the moment this treatment is effective in only 20 to 40% of patients depending on the cancers treated. This treatment also remains very expensive: approximately $200,000 per patient per year in the USA.
Artificial Intelligence is used to visualize whether or not the patient responds positively to immunotherapy. In the near future, the researchers’ objective is to be able to use algorithms to predict the patient’s response to treatment before the start of immunotherapy in order to choose between immunomodulators or CAR-Ts treatment.
Anant Madabhushi , director of the Center of Computational Imaging and Personalized Diagnostics (CCIPD) at Case Western Reserve University in Cleveland, and his team have developed an algorithm that can detect changes in the texture, volume and shape of a given lesion, not just its size.
“We have found that texture change is a better indicator of the effectiveness of immunotherapy. Sometimes, for example, the nodule may appear larger after therapy for another reason, such as a broken vessel inside the tumor, but the therapy actually works. Now we have a way to know it thanks to the algorithm” Khorrami, a graduate student working at the CCIPD.”
