Are we closer to automated pathological assessment of prostate cancer biopsy slides?


According to a recent article in Modern Pathology, a team of researchers at Yale University and at Memorial Sloan-Kettering Cancer Center (MSKCC) have been able to show that an artificial intelligence (AI) system designed and validated at MSKCC could be used to diagnose prostate cancer as either “suspicious” or “not suspicious” based on data from nearly 2,000 slides of prostate tissue acquired at Yale Medicine.

Perincheri et al. used to AI system to analyze 1876 prostate core biopsy slides from 118 consecutive patients procured at Yale Medicine and processed at Yale Pathology in June and July 2019. They found that, compared to expert pathological evaluation by specialists in the pathology of prostate cancer at Yale Pathology, the AI system was able to identify core biopsies with cancer with:

  • A sensitivity of 97. percent
  • A positive predictive value of 97.9 percent
  • A specificity of 99.3 percent
  • A negative predictive value of 99.2 percent

Now we need to be very clear that this AI system is not (well, not yet anyway) capable of identifying specific Gleason scores of Grade Groups with this level of accuracy. It is only capable of telling whether a specific slide shows signs of cancer or the absence of cancer, but for many labs this may on its own be extremely useful because it can potentially be used either to validate the findings of pathologists with less experience in the identification of prostate cancer or to “pre-screen” slides into those that show signs of cancer and those that don’t.

There will undoubtedly need to be further study of this type of AI system to evaluate pathology slides objectively, when it comes to the diagnosis of prostate cancer. What appears to be true, however, is that just as AI has started to be used to evaluate things like MRI scans and CT scans objectively, we are beginning to see the initial use of AI in the evaluation of pathological materials too.

4 Responses

  1. Any idea what is meant by specificity and sensitivity?

  2. Does anyone know of transperineal biopsies being performed in Canada, preferably in Ontario? It sounds superior to transrectal, but the latter seems to be the only biopsy procedure on offer here!!

  3. I do not have specific information, but if they are being performed, it is likely to be at one of the specialized prostate cancer centers like the ones associated with the University of Toronto (e.g., at Sunnybrook and at the Princess Margaret Hospital).

  4. Click here, and scroll down just a bit, and you will find excellent explanations of specificity and sensitivity (and other factors too).

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: