Peak AI performance without calibration or training

When Paige Prostate is used as a prescreening tool, pathologists could reduce review volume and greatly enhance efficiency and productivity. 

An independent assessment of an artificial intelligence system for prostate cancer detection shows strong diagnostic accuracy

Published in Modern Pathology
Sudhir Perincheri, Angelique Wolf Levi, Romulo Celli, Peter Gershkovich, David Rimm, Jon Stanley Morrow, Brandon Rothrock, Patricia Raciti, David Klimstra, John Sinard

Modern Pathology, Volume 34, Issue 8, 1588 – 1595

ABOUT THIS STUDY

An independent assessment by Yale Medicine, a highly sub-specialized academic medical center with a high volume of prostate biopsies, conducted a study to determine Paige Prostate’s utility as a prescreening tool.

To evaluate the performance of Paige on prostate biopsies processed and independently diagnosed at an institution unrelated to where Paige was developed, 1,876 prostate core biopsies from 118 consecutive patients were analyzed without any site-specific calibration of the algorithm. Paige Prostate's categorizations were compared to the original pathology diagnoses rendered on glass slides for each core biopsy. The algorithm classified a core biopsy as “suspicious” if it detected adenocarcinoma or glandular atypia (including FGA, PIN-ATYP, and ASAP), and as “not suspicious” if none of these lesions were detected. 

AI Used: Paige Prostate 

BY THE NUMBERS

The Results

This study was developed to determine Paige Prostate’s utility as (1) a prescreening tool to identify cases without carcinoma and (2) as a second read tool to detect manually missed foci of carcinoma. The first use case is a potential productivity tool to allow the pathologist to focus only on cases suspicious for malignancy, allowing greater sign out volume per day. For this scenario, a high negative perineural adenocarcinoma is shown and the predictive value is desirable. The second use case is a potential patient safety tool by increasing the accuracy of rendered diagnosis. In this use case, a high positive predictive value is desirable. 

In this study if Paige Prostate was used as prescreening tool such that only those cores categorized as suspicious or as out of distribution were manually reviewed, a pathologist would have to review only 589 of 1876 core biopsies (31.4%), substantially increasing productivity.

99.3% specificity

97.7% sensitivity 

99.2% negative predictive value (NPV)

68.6% of biopsies could be excluded from manual review with AI support

The study data shows that Paige Prostate is finding even very small foci of atypical glands and proves to be a versatile tool with varied use-case applications in anatomic pathology practice settings. 

Read the published study

Previous
Previous

Independent real-world application of a clinical-grade automated prostate cancer detection system

Next
Next

Artificial Intelligence Helps Pathologists Increase Diagnostic Accuracy