Cancer has a huge effect on society as a whole. Every year, cancer kills nearly 600,000 people in the United States alone. Cancer treatment is very expensive. In 2010, the total cost of cancer care in the U.S. was estimated to be more than $140 billion. Optimizing treatment planning requires putting together and making sense of a lot of complicated data from many different sources. AI can help with this.
AI could help doctors and radiation oncologists plan the best treatments by predicting how sensitive cancer cells will be. AI can increase the accuracy and sensitivity of radiation treatment by utilizing an algorithm. For example, based on a patient's body, AI can find problems with treatment planning. This technology will also make it easier to plan treatments by automatically checking transcriptions at key points in the process of radiation oncology.
AI could also be used to help rebuild images. AI can look at data from thousands of CT scans and figure out what shapes and patterns are signs of cancer. Then, a radiologist uses this algorithm to look over the computer's decisions and add more information. The AI algorithm can also find problems in cancer patients who have been underrepresented in the past. These AI improvements could make patient care much better and save a lot of time.
AI is already changing oncology by letting doctors look at data from more than 500 clinical trials every day. AI tools can find people with cancer who have similar symptoms and figure out the best way to treat them. AI can also be used to change treatments for different disease sites. For instance, the AI can use machine learning to figure out which drugs are best for a patient's condition. As AI is used more, it could change how cancer patients are treated by doctors.
AI applications in oncology have a lot of potential, but there are many problems with how they are used. Some of these problems are biased and inconsistent data, a lack of standards for reporting research, and outdated regulatory frameworks. But these problems can be solved with training and education, standardizing data and workflows, and funding research for the future. AI tools can help solve these problems and make sure that patients are safe.
Most of the software on the market is for processing images and planning treatments. By 2020, these segments are expected to bring in the most money, making up more than half of the market. In oncology, more and more advanced image processing and treatment planning systems are being used. They help radiologists make better decisions about doses and tissue toxicity while also reducing the amount of radiation a patient gets. They also give radiologists a wide range of tools.
AI-powered medical devices can speed up the work of radiologists and make RT specialists' jobs easier. AI can save time and money by automating tasks like planning and shaping. This makes treatments go faster and saves both time and money. Software that automates treatment planning and contouring can make the work of medical physicists even easier, giving them more time to care for patients. Also, it becomes easier to change radiation-oncology equipment online.
Around 2016, when several studies in this area came out, it became popular to use ML in radiotherapy quality assurance (QA). Li and Chan also made an ML model that learned from log files and could predict how well Linac would work over time. In a recent study by Osman et al., an ML model was used to predict how tumors would move. A big part of what they do is look at image data and try to guess how the movement of tumors might affect treatments.
As AI continues to change the way medical imaging is done, more and more radiotherapy equipment will have AI built in. For instance, on CT scans, CNN-based models were 85% accurate at finding extranodal extension. Radiographs can be hard for doctors to use to diagnose this condition, and extranodal extension can be a very important factor in patients with head and neck cancer. As a way to help doctors make decisions, models based on CNN show a lot of promise.