Cancer care in Africa faces a well-documented shortage of clinical oncologists. Nigeria, for example, is estimated to have around 70 oncologists serving a population of 213 million.
In South Africa, particularly in Johannesburg, millions have been allocated to cancer care to assist hospitals overwhelmed with patients needing critical treatment. While a skills shortage is a significant factor, Dr. Duvern Ramiah, head of radiation oncology at the University of the Witwatersrand and Charlotte Maxeke Johannesburg Academic Hospital, identified a critical bottleneck in treatment access:
“The main bottleneck with radiation and oncology treatment is a lack of radiotherapy planning. That’s the first step to patients getting radiation.”
With patients facing wait times of up to five years for prostate cancer and 18 months for breast cancer, Dr. Ramiah realized the existing equipment wasn’t being utilized efficiently. The focus shifted to improving radiotherapy planning processes.
When a cancer patient requires radiation, a radiation oncologist prescribes the treatment. This is followed by scanning and a detailed radiotherapy planning process. “If a patient has prostate cancer, the doctor will say (for example), this person needs external beam radiation. I’ll see him and make that call. After that, he’ll go have a CT scan which is done in the position that he’s going to have radiation in,” explains Ramiah. Once Ramiah receives the CT scan, he manually draws in all the areas to be radiated, such as the prostate and lymph glands, while also identifying key areas to avoid, like the bladder. “It’s a long process, and only once that’s done, I’ll send the scan off to a dosimetrist or planning radiotherapist,” he continues.
Recognizing the time-consuming nature of this planning stage, Ramiah turned to artificial intelligence (AI) to automate part of the process. A convolutional neural network (CNN) now contours the target areas on scans automatically, significantly reducing the time doctors spend on this mundane task. “It reduces the time gap massively,” he stated.
Ramiah also reduced bottlenecks by utilizing remote radiotherapists in America to create detailed radiotherapy plans for implementation at Charlotte Maxeke. Furthermore, he identified the repetitive nature of the planning process. A breast cancer plan, for instance, can take up to two hours, with approximately 70% of the work being consistent from one patient to the next. As a result, a script has been developed to automate portions of the task.
Through these AI-driven initiatives, Ramiah has cut the waiting time for radiotherapy plans at Charlotte Maxeke from two months to between 48 and 72 hours.
One of the most critical aspects of Ramiah’s AI implementation is its vendor-agnostic nature. Because the technology isn’t tied to a specific radiotherapy machine, it could potentially be implemented in any healthcare facility worldwide.
“You can go to any unit in Africa with this solution, regardless of their technology infrastructure,” he says. “What’s interesting about AI tech is that it increases access. It’s an enabler, which is why I push so hard for it in my department.”
