Founded in 1958, the American Society for Therapeutic Radiology and Oncology (ASTRO) enjoys the most famous reputation in the industry all over the world. Now there are more than 10,000 members including doctors, nurses, biologists, physicists, radiotherapy technicians and other medical workers.

ASTRO attracts more than 10,000 professionals from all over the world every year, which is highly regarded by the medical community, especially that of radiology and oncology. Results reported in the conference will act as academic references for clinical practice or further researches.

The 61st ASTRO was successfully held in Chicago (local time) during September 15~18, 2019, with the theme of “Innovation, Cooperation, Transformation”, which aims to promote targeted radiotherapy more orderly, and tap its potential in oncotherapy.

The wave of AI is sweeping the globe at present, theme forum named “Artificial Intelligence and Deep Learning in Medicine” was also held in ASTRO 2019. Experts at home and abroad were invited to deliver special speeches about hot issues in recent years as “clinical decisions based on AI and challenges faced by AI-adided radiotherapy”. Suchi Saria of Johns Hopkins University, David Magnus, the director of the Stanford Center for Biomedical Ethics, and Professor Yue Jinbo, the director of Department of Radiotherapy in Shandong Cancer Hospital and Institute all attended the conference.

Radiotherapy is commonly used in cancer treatment, but the lack of professional medical technicians and relative lag of treatment technologies impede the development of oncotherapy. It also indicates that radiology and oncology should be systematized, standardized and processed as soon as possible, and the application of AI technology is expected to alleviate this dilemma.

PVmed's self-developed Intelligent Radiotherapy Contouring Solution (IRCS) made a wonderful appearance in ASTRO 2019.

As an intelligent contouring software, IRCS is easy to install and operate with safe and reliable quality. The software can be deployed in stand-alone mode or deployed within hospital LAN or medical association network. Docking with information systems of medical institutes, it has become a reliable node in the workflow of data management and processing. IRCS supports archived data by label to simplify the management of multi-purpose data at one time, and allows for one-click export of contouring results to the local or DICOM devices.

So, what is the clinical value in IRCS?  It is well known that target contouring asks for high demands in personal experience and knowledge of doctors who need to also delineate OARs when contouring the target volumes. It is not easy for doctors to contour manually since there are many problems in it. They have to spend half of the work time to do this job which requires spatial imagination and prediction of lesion transmission and expansion range. 

PVmed's IRCS helps doctors quickly contour OARs, GTV and CTV and get them free from this repetitive and time-consuming task so as to carry out other researches on oncotherapy.

IRCS employs convolutional neural networks and integrates a variety of image processing techniques to train reliable algorithms of organ segmentation with a small batch of data, which can automatically segment tumor and non-tumor tissues and accurately draw target volumes. The doctor's work substantially reduces to reviewing, fine-tuning and modifying the images. At the same time, medical images reviewed by senior physicians can also provide references for young ones, decreasing the error probability and improving the accuracy of target contouring in real time.

In fact, precise radiotherapy is like shooting, and contouring on images within target area is like aiming at the bullseye. PVmed boasts the most qualified identification image database and algorithms in the peer around the world, and its study samples are all from experienced physicians in the industry. Therefore, the average accuracy of IRCS is as high as 95% which is comparable to that of specialists.

It used to take 3 or 4 hours to manually contour, but it shortens to 10 minutes now since IRCS has greatly improved the overall efficiency in target contouring as an AI-aided tool. In addition, IRCS can delineate automatically 58 OARs throughout the body, which also improves the quality of radiotherapy.

Health care represents people's livelihood. PVmed's IRCS has fully integrated AI technology with clinical medicine in order to allocate limited resources more rationally, improve the diagnosis and treatment of primary hospitals, and solve patient's concerns timely, playing a necessary role in medical system and health insurance in China.

PVmed will dive into in-depth study in medical AI to simplify doctors' work, and provide with more intelligent clinical treatment and solutions with creative AI technologies in the coming years.