Proscia Takes on Subjectivity in Pathology Through AI Innovation

Share Article

Proscia analyzed whole slide images as part of the 2017 ISBI CAMELYON Digital Pathology Challenge to train their algorithm to automatically classify pN-stage in breast cancer patients.

The fact that computer models and humans ‘substantially agree’ is remarkable. Our technology achieved a kappa-score of 0.766, which is quite high compared to inter-pathologist concordance for many other cancers, including breast cancer.

Proscia Inc., a data and analytics solutions provider for digital pathology, today announced their digital pathology technology has the potential to eliminate subjectivity in the detection and classification of breast cancer. Using images provided by five medical centers in The Netherlands as part of the 2017 ISBI CAMELYON Digital Pathology Challenge, organized by the Radboud University Medical Center, Proscia successfully trained their algorithm to automatically classify pN-stage in breast cancer patients with a kappa score exceeding current standards. Among 23 other researchers and innovators, Proscia was one of only a few who developed algorithms using their commercially available technology.

“We knew from our past work with CAMELYON16 that Proscia’s machine learning algorithms could accurately predict breast cancer metastases,” said David West Jr., chief executive officer at Proscia. “It is gratifying that our technology reaches ‘substantial agreement’ with the human read and that an algorithm developed using commercially available technology stands up against technologies that are still being developed in research labs across the world.”

The goal of the CAMELYON17 challenge was to evaluate new and existing algorithms for automated detection and classification of breast cancer metastases in whole-slide images of histological lymph node sections. The presence of metastases in lymph nodes has tremendous implications on the treatment of breast cancer patients. Currently, making this determination requires extensive in-person microscopic assessment by a pathologist. An automated solution could hold great promise in reducing both the workload for pathologists and subjectivity in diagnosis.

Last year’s CAMELYON16 grand challenge was the first challenge ever to use whole-slide images. This year, CAMELYON17 took the challenge one step further by moving from slide-level analysis to patient-level analysis, combining the assessment of multiple lymph node slides into one diagnosis. The organizers of the challenge felt this would better reflect the demands of detecting and classifying images in a clinical setting.

“What made a difference in this year’s competition was not, in-fact, the identification of metastatic tumor, but rather agreement with a human read,” added Mr. West. “The fact that computer models and humans ‘substantially agree’ is remarkable. Our technology achieved a kappa-score of 0.766, which is quite high compared to inter-pathologist concordance for many other cancers, including breast cancer.”

The CAMELYON17 Challenge was held in conjunction with ISBI 2017 in Melbourne, Australia.

About Proscia
Proscia was founded in 2014 by a team out of Johns Hopkins, the Moffitt Cancer Center, and the University of Pittsburgh to improve clinical outcomes and accelerate the discovery of breakthrough advancements in the fight against cancer. Using modern computing technologies that unlock hidden data not visible to the human eye and turning that data into valuable insights in the fight against cancer, the company is dedicated to improving the efficiency, speed and quality of pathology diagnostics and research. To learn more, please visit http://camelyon17.grand-challenge.org/ and http://www.proscia.com.

# # #

Contact
Leigh Minnier
610.228.2108
press(at)proscia.com

Share article on social media or email:

View article via:

Pdf Print

Contact Author

Jessica Stanek
Proscia Inc
+1 215.297.3606
Email >
@ProsciaInc
since: 10/2013
Follow >
Proscia
Like >
Visit website