Image Credit: mi_viri/Shutterstock.com
Artificial intelligence (AI) applications in healthcare is a growing area as the technology is becoming increasingly sophisticated and more efficient. AI is now assisting clinicians in the early detection of diseases.
Now, Microsoft and SRL Diagnostics have created an AI network in order to detect cervical cancer which would relieve some of the burden placed on overworked healthcare systems and laboratories.
Based in India, SRL Diagnostics process over 100,000 Pap smear samples of which about 2% require further investigation.
We were looking for ways to ensure our cytopathologists were able to find those 2% abnormal samples faster.
Dr. Arnab Roy, Technical Lead for New Initiatives & Knowledge Management, SRL Diagnostics
India has the highest mortality rate in the world of those suffering from cervical cancer, accounting for more than 25% of the 260,000 deaths worldwide. One of the main challenges faced when trying to combat this preventable disease is the efficient processing of screenings. Due to the high-volume of tests cytology labs must process compared with the number of doctors that can process pap smears, those that do require further attention may not receive it in a time efficient manner.
Therefore, having an AI diagnostic system that can rapidly process pap smear tests and quickly identify the 2% of samples that require further attention would help doctors initiate treatment procedures faster. For SRL Diagnostics to develop such a system a team of cytopathologists examined and marked-up a wide range of scans taken from Whole Slide Imaging (WSI) slides that contained around 300-400 cells per slide. Once the vast amount of data had been studied and processed accordingly, the team could then use this data to train a Cervical Cancer Image Detection API.
Yet, one of the difficulties that SRL faced was the task of overcoming subjectivity as interpretations of the WSI slides can differ from person to person. “Different cytopathologists examine different elements in a smear slide in a unique manner even if the overall diagnosis is the same. This is the subjectivity element in the whole process, which many a time is linked to the experience of the expert,” stated Dr. Roy.
This is where the collaboration with Microsoft was effective when processing the data-sets and developing a system that could overcome the challenge of subjectivity. Manish Gupta, Principle Applied Researcher at Microsoft Azure Global Engineering, said, “The idea was to create an AI algorithm that could identify areas that everybody was looking at and “create a consensus on the areas assessed.”
The collaboration between SRL and Microsoft has brought about encouraging results with the first API for screening cervical cancer set to undergo internal preview at SRL Diagnostics, according to a Microsoft blog post. The system runs on the Microsoft Azure platform and has the ability to rapidly process liquid-based cytology slide images for early-stage detection which can then be relayed to pathologists in the labs.
Cytopathologists now have to review fewer areas, 20 as of now, on a whole slide liquid-based cytology image and validate the positive cases thus bringing in greater efficiency and speeding up the initial screening process.
Thus, as well as saving valuable time for cytopathologists, the Microsoft-SRL system could also save thousands of women’s lives as they could get early access to critical treatment. “The API has the potential of increasing the productivity of a cytopathology section by about four times. In a future scenario of automated slide preparation with assistance from AI, cytopathologists can do a job in two hours what would earlier take about eight hours!” said Dr. Roy.
The long-term goal of both SRL Diagnostics and Microsoft is to further develop their APIs and machine learning capabilities in order to extend the technology to other areas of pathology for diagnostics in detecting kidney related diseases and other cancers such as liver and pancreatic.
The future of AI and pattern recognition for the identification of patients at risk of developing a condition – or the deterioration of such because of lifestyle, genomic, environmental, or other factors – is an area where this technology could soon begin to make significant progress and be a valuable asset to clinicians worldwide.