Microscopy as a field has been existent since 700BC and has been constantly evolving due to its widespread ability to see and understand the workings of the human body. Use of brightfield microscopy has been a popular tool for diagnosis of human diseases. However, there has not been much change in the microscopy standard practices since the last 100 years.  

The rise of technology has revolutionized the field of healthcare. It’s impact in the field of microscopy especially in Pathology laboratories is evident from the increase in number of Artificial Intelligence (AI) algorithms being developed and submitted to the US-FDA as a clinical aid for diagnostic decision making. Particularly, the biggest breakthrough came in 2021 when the US-FDA approved the first AI based pathology software for clinical use. Since then, several hospitals have announced partnership with new age AI based software companies to overhaul their conventional pathology workflows by adoption of digital workflows. With a growing shortage of pathologists worldwide coupled with the increasing burden of diseases such as cancer (as shown below in Figure 1) there is a need to embrace AI-based computational pathology  to: 1) Provide accurate and consistent diagnosis leading to better patient care 2) Increase the turnaround time for reports, and 3) Facilitate sharing of data for collaborations and educational trainings.

Figure 1: Need for AI in Pathology
How can AI change the landscape of Pathology practice?
  • Accelerating Workflows in a Diagnostic Pathology Laboratory:

    Once the sample enters the pathology lab, the tissue gets sliced, processed and imaged. The high-resolution images capture all the micro and macro level details of cellular and tissue architecture which are then analyzed by the pathologist. According to a recent study from MSKCC, a pathology laboratory processes up to 200,000 slides a month which will generate data in petabytes. Processing such high volume data is an humongous task and adaptation of AI based workflow in pathology lab will provide following benefits:

  • Accelerating “Molecule to Market” in BioPharma Lab:

    Pathology plays a key role in the area of drug development by assessing the degree and severity of the disease to evaluating the efficacy or toxicity of the drug using biomarkers. Successful applications of pathology-based biomarkers for patient stratification (Example: PD-L1 levels in patients for immunotherapy trials), treatment monitoring and end points in clinical trials has been steadily rising in the recent years. In addition to providing quantitative and accurate analysis, AI based algorithms facilitate biomarker discovery, identify novel features (to predict gene mutations) and predict treatment responses (combining other data modalities). Recently, major pharma companies such as Roche, GlaxoSmithKline have incorporated AI based pathology workflows for better biomarker discovery and patient stratification. With the recent US-FDA approvals for use of AI based software products for clinical diagnostic use especially in Oncology, the adoption of AI for other applications (as shown below) has been steadily on the rise.

Integration of AI based Computational Pathology into clinical practice:

The use of AI for clinical assessment and diagnosis is on the rise and is expected to revolutionize the current standard of patient care. While AI has made the most impact in the field of Oncology, its applications are extensive in several non-oncology applications including:

Figure 2: Non Oncology applications of AI in Pathology

For successful implementation of AI based workflows, the solutions must be readily usable by pathologists, provide seamless integration with other hospital systems for easy data access/management with strong data security and provide real time results. This task requires adoption of technology enabling automation, improving the accuracy and efficiency of diagnostic pathology thus making it possible for pathologists to provide more timely and specialized services to patients.

Key Technology Enablers for Computational Pathology:
Cloud ComputingEdge ComputingFederated Learning
  • Enables remote access to data and tools.
  • Improves the scalability and flexibility of solutions.
  • Easily accommodates additional storage and processing needs without expensive hardware upgrades.
  • Allows for real-time analysis of medical images of tissue samples.
  • Reduces latency, or the delay between when data is collected and when it is analyzed.
  • Improves the security of sensitive medical data by processing and analyzing data at the edge of the network, rather than in a central location.
  • Offers better data privacy since it allows for the data to remain on the devices without having to move to a centralized repository, protecting patient privacy.
  • Improves accuracy as the machine learning model is trained using a diverse set of data across geographies.
  • Reduces computational costs because the training is distributed across multiple devices.

With a strong track record in HealthCare digital transformation, our technology capabilities at Persistent can enable better adoption of Computational Pathology in the customer settings. Please connect with us to explore how we can help build your own AI based computational pathology solutions that will improve the efficiency of your diagnostic workflows thereby saving lives. Reach out to us, to know more about our Healthcare & Life Sciences offerings.