Revolutionising cancer care: Role of AI in radiation therapy
We are all familiar with radiation therapy for cancer treatment. Utilized in over half of all cancer cases either as a standalone therapy or in conjunction with surgery and chemotherapy, radiation therapy has long been a cornerstone of cancer treatment.
Despite its effectiveness in cancer treatment, there are multiple challenges in treatment – contouring (identifying the precise place that has to receive the radiation), optimal treatment planning (correct beam configuration, energy, dosage, protection of healthy tissues, etc), and more.
While the physics behind radiation has remained fundamentally constant, the planning and delivery of radiation therapy is undergoing a transformative change through the integration of Artificial Intelligence (AI).
AI-powered oncology, through high precision and personalisation, offers the potential to improve clinical outcomes, enhance efficiency, reduce treatment planning time, and tailor interventions to individual patients in ways that have not been feasible so far.
Here is how AI is redefining the planning landscape in radiation therapy, enabling a shift from labour-intensive and variable processes to data-driven, adaptive, and standardized care approach.
Current Radiation Treatment Planning
- Consultation and evaluation – this step requires an oncologist looking at history, images, pathology and other clinical data and defining the goals of radiation treatment – curative, palliative or adjuvant.
- Treatment mapping – this part of the planning uses a CT scan or MRI and results in the creation of markings on the body to guide the daily treatment.
- Target volume and dosimetric planning – once the daily treatment areas have been identified, the next step for the oncologist is to define how much area needs to receive the radiation, how much and how – beam angle, intensity and duration. Clearly defining how much radiation is to be delivered to the tumour and how much can be allowed to normal tissues.
- Plan review – once the plan has been reviewed and approved by the radiation oncologist, the physicist will perform machine-specific QA to ensure that plan can be accurately delivered by available machines.
- Delivering and monitoring – the patient comes for daily treatment for several weeks and depending upon the treatment plan, gets the treatment done. Any change in the tumor or patient anatomy requires replanning.
As can be seen, the therapy requires heavy manual planning and constant intervention. Using AI, many of the manual components can be automated. Some of the advantages of using AI are listed below:
- Auto contouring of tumors and organs at risk – using MRI and CT Scan images, AI can quickly separate the tumor and organs at risk. This will reduce the planning time and bring a higher level of consistency, thereby allowing clinicians to focus on other critical tasks.
- Personalised dose planning and toxicity forecasting – using the historical patient data, the AI model can predict the patient response to different radiation dosages as well as the risk to normal tissues during treatment. Since not all patients react to same treatment in same manner, this can help the clinician simulate different dosage delivery and potential impact, thereby creating the best plan for the specific patient.
- Automatic treatment plan – continuing with the simulation, AI can generate the treatment plan also (beam angle, dose distribution, etc). Clinicians need to only validate the data being provided by the AI module and determine the best course of action.
- Adaptive treatment - during treatment, depending upon the patient reaction to treatment, AI can recommend the adjustments required. AI can also use the patient genomic data, prior treatment data and other data points to make continuous recommendations to the attending oncologist and clinicians, thereby significantly improving the recovery possibility of the patient.
- Operationally, AI can forecast the load on treatment facilities and do treatment scheduling accordingly, leading to maximum operational efficiency for the limited resources.
AI has strong potential to improve radiation therapy for cancer. Having said this, there are multiple challenges as well in AI adoption. Issues like transparency in AI models, the issue of bias in patient physiology can hinder the adoption of AI. Human oversight of any decision-making remains non-negotiable. Medical professionals need to have confidence that the information being delivered by AI is trustworthy.
As AI continues to improve and mature, we can expect more capable, transparent and trustworthy solutions coming to the market and taking their place alongside the radiologist/physicist/clinicians towards improving the patient delivery services.
Sudhanshu Mittal is the Head & Director, Technical Solutions of Meity Nasscom CoE
The opinions expressed in this article are those of the author and do not purport to reflect the opinions or views of THE WEEK.
Health