MSK Pathology

In 2016, MSK Pathology reviewed approximately 80,000 cases and generated approximately 800,000 glass slides. Inherent in the use of glass slides are logistical issues of slide transport, archiving, and retrieval as well as limitations in remote review, conferencing, and consultation capabilities. Digital Archiving has been operational since 2015. Telepathology consultation including for frozen sections and intradepartmental digital consultation was started in 2016. We have over 250,000 digital slides scanned to date for education and archiving with the prospective scanning of 40,000 selected slides per month including all consults, biopsies, frozen sections, and surgical resection specimens from 10 sub-specialties. These slides were obtained from outside consults and in-house biopsies, and include molecular slides, hematopathology and cytology slides.

MSK Digital Imaging and Computational Pathology

Digital Pathology involves conversion of tumor tissue samples from glass slides to digital images to improve diagnosis while providing an infrastructure for Computational Pathology. Computational Pathology is based on quantitative measurement, mathematical modeling, and the development of algorithms that can inform and augment the interpretation of disease processes on digitized slides while integrating genetic and clinical information with the morphometric analysis of the image to provide higher level understanding of the tumor pathology. The move to a fully digital workflow will allow this enhancement of diagnosis through computer-augmented diagnostic algorithms.

Once pathology slides are digitized, morphometric analysis allows the application of mathematical modeling to analyze the histologic features. Machine learning can then be applied to derive more data from the slides, and computational analysis allows the development of algorithms to improve efficiency of slide review, calculate numerical data, and merge pathology data with molecular, clinical or other large data sets. The criteria for diagnosis can thus be made more objective, based on the use of deep learning (artificial intelligence).

Artificial Intelligence (AI) as a digital assistant will revolutionize diagnostic pathology and research. It will enable pathologists to be faster, more efficient, and more accurate by supplanting subjective with objective criteria. In addition to clinical care, it will facilitate large scale, quantitative correlations between tumor characteristics, protein expression, and genetic panels like MSK-IMPACT.

At MSK, we have the tumor pathology resources to effectively train machine learning algorithms. Our staff of experienced, sub-specialized tumor pathologists can provide a level of pathologic annotation that ensures that we are uniquely positioned to develop these algorithms. Concurrently, the MSK pathology digital imaging program will seek to exploit the potential for new technology to enhance the applicability of digital medicine. We are developing new technologies and testing the newest technologies for enhanced digital microscopy, such as three-dimensional histology, micro-computed tomography imaging, and rapid ex vivo whole tissue microscopy. Collaborations with clinical departments (e.g., surgery), radiology, medical physics, and informatics groups will enhance the assessment as we seek multidisciplinary applications.