ARTIFICIAL INTELLIGENCE
n Arm B. Digital + AI Read: Cases were read digitally using Concentriq Dx including embedded results of Ibex Prostate. Information provided by the AI was pre-filled into Concentriq Dx (see Fig 1b).
Cases and associated WSIs were read in both arms of the study by all three board- certified pathologists, all of whom had different levels of professional experience and backgrounds. None had previous experience with DP or AI in clinical practice.
Overview of results n Deeply integrated AI-powered
workflows drive laboratory efficiency, allowing pathologists to sign out 2.9 cases per hour instead of 1.2.
n Digital pathology is easy to use and learn, and pathologists are eager to adopt Proscia’s Concentriq with Ibex Prostate.
n Ibex Prostate is highly accurate in detecting cancer and other pathologies, and in alignment with the accuracy of the pathologists’ diagnoses.
Top: Fig 1a. Example of digitised slide of a PCNB tissue sample (WSI) for review in the Concentriq Dx DP workflow (no AI overlay). Bottom: Fig 1b. Example of digitised slide of a PCNB tissue sample (WSI) in the Concentriq Dx DP workflow with Ibex Prostate AI overlay highlighting the high cancer likelihood areas.
which is instrumental in guiding treatment decisions and improving survival rates. However, rising case volumes and complexity can lead to slower turnaround times for biopsies. Reviewing cases using traditional methods – where pathologists view 24 H&E (haematoxylin and eosin) slides per case, on average, using a microscope and manually recorded information – takes time. In addition, subjectivity in Gleason grading, the small size of certain tumours, the increased demand to report features beyond cancer, and the large number of tissue cores per case, complicate case reviews. These realities made PCNBs an ideal area of focus to understand the impacts of DP and AI.
The joint solution used in the study embeds Ibex Prostate, a clinical-grade, AI- powered solution that assists pathologists in improving the detection and grading of prostate cancer, within Concentriq Dx, Proscia’s digital pathology workflow solution. Concentriq Dx streamlines pathologists’ review of tissue samples that have been scanned at high resolution
36
to create digital slides, or whole-slide images (WSIs), which pathologists can then view.
The study quantified the potential value-add of digital pathology workflows in large lab environments, as well as the value-add of embedding AI into these workflows.
The study was designed as a
retrospective observational study that included 180 randomised prostate cases from adult subjects, with a partial crossover design that consisted of two arms: n Arm A. Digital Read: An arm in which WSIs were read digitally using Concentriq Dx (see Fig 1a).
Learning curve The three reading pathologists all indicated to have limited to no prior experience with reading cases on a digital system. Prior to starting the reading, pathologists received a 90-minute training session along with the possibility to use the system on a set of training cases, independent from the study slides. The trend in reading time seen for the individual pathologists shows a small reduction in reading time based on how many cases have been read (see Fig 2). Feedback from the reading pathologists was that they did not experience a noticeable learning curve using Concentriq Dx, but that the variation in reading time for Concentriq Dx without AI was driven mainly by diagnosis and data entry. This provides evidence for the intuitiveness and ease of use of Concentriq Dx, and the ability for pathologists to efficiently use the system for diagnosis even with limited prior experience. Furthermore, the intuitiveness of
DP and AI technologies have enormous potential to impact the pathologists’ experience and overall job satisfaction, while delivering economic and operational returns
AUGUST 2024
WWW.PATHOLOGYINPRACTICE.COM
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52
