Computational pathology is a widely used term nowadays, but its exact meaning is often subject to interpretation. For us at Ibex Medical Analytics, it’s all about the smart use of algorithms and computing power to better diagnose diseases, their severity and progression. It’s also about making the pathology practice more accurate, efficient, and accessible, ultimately leading to better patient outcomes.
We see early signs of computational pathology applications in precision medicine, drug and biomarkers discovery, more efficient workflows, decision support tools and more. With the increasing adoption of digital pathology and advancements in image analysis, artificial intelligence and enabling hardware, we expect to see a whole new set of computational pathology tools and capabilities emerge. In this article we cover a few key trends and future directions in the field.
Standardization is key for driving adoption of any new technology. In order to reap the full rewards of computational pathology, industry stakeholders need to cooperate and create standardized environments for imaging, data collection, storage, and exchange formats. Much like PACS and DICOM standards and their contribution to the advancement of medical imaging, similar discussions around digital pathology have started to emerge. This will enable smoother processing of whole slide images (WSI) and better interoperability between scanners, workflow solutions, AI-based decision support tools and laboratory information management.
Computer-aided cancer diagnosis
In pathology, the transition from glass slides to digital images allow for the use of computer vision and machine learning algorithms for image analysis, providing access to increased amounts of morphologic data that can now be obtained from each slide. Deep learning algorithms allow for a more accurate classification of morphological features in the tissue, leading to more consistent interpretations and reduced chances of overlooking suspicious areas, since all the pixels in a slide image are being analyzed.
One useful example of an AI-based application that can make a real difference in the pathology lab is Ibex Medical Analytics’ Second Read™ system. The system analyzes slide images, for example from a prostate biopsy, identifies discrepancies between its findings and the pathologist’s report, and sends an alert if there is a need for a second review of the case, to ensure diagnostic accuracy.
Artificial intelligence and computational pathology will have a big role in making precision and personalized medicine an integral part of routine care. The combination of customized algorithms and quantification techniques, used in analyzing pathology specimens, together with follow-up clinical data on patients’ response to treatment and long term survival rates are a source of new insights. Such insights lead to higher success rates in drug development programs, as well as more personalized treatments and better outcomes - for example through improving patient stratification during clinical trials or by enabling discovery of novel AI-based bio-markers.
Lack of access to quality medical services is a real challenge in remote areas as well as in developing countries. Adoption of digital pathology, together with communication networks and cloud computing provide a platform for tele-pathology by facilitating convenient and reliable data sharing for quality cancer diagnosis and screening by pathologists in remote or otherwise under-served areas.
The growing capabilities of artificial intelligence and big data processing power have the potential to significantly change the practice of pathology. While we are constantly on the lookout for new applications for artificial intelligence and big data in pathology, we know that this is just the tip of the iceberg; no current vision will match reality 10-15 years from now. We can expect more accurate, objective and rapid diagnosis, leading to better clinical outcomes across multiple tissue types and medical conditions.