← Return to Dr. Bert Vorstman skeptical of any PC treatment. What do you think?

Discussion
Comment receiving replies
Profile picture for brianjarvis @brianjarvis

@jeff1963 It’s important to remember what a Gleason score is —> It’s one pathologist’s educated, expert opinion of what is seen under a microscope in the biopsied tissues.

But, being more an art than it is a science, what one pathologist sees as a 3+3, another might see as a 3+4. (Which is why it’s so important to get 2nd opinions on any test, biopsy, or scan that requires a doctor’s opinion or interpretation.)

With a 3+3, all of the cell structures that the pathologist can see under a microscope are all “3” cell structures. What happens with a 6(3+3) Gleason score, is that there may also be some sub-microscopic “4” cell structures that are too small to see, but in time grew large enough to be a visible cell structure in the biopsy samples taken. So, what was a 3+3 eventually becomes a 3+4 as those cells multiply. But, this is not a metastases (i.e., growth distant from the prostate). The “4” cell structures were always there; just too small to see.

In this PCRI presentation, Dr. Scholz mentions that if there is a metastasis of a Gleason 6, that in clinical trials those were shown to have been initially misdiagnosed:
> https://youtu.be/NV8QHzbgamI

Jump to this post


Replies to "@jeff1963 It’s important to remember what a Gleason score is —> It’s one pathologist’s educated, expert..."

@brianjarvis you give such amazing answers I'm wearing out the like and helpful buttons, I hope that's OK with you! I have a trendy question maybe you've run across answers for. When you talk about Gleason scores being expert opinion of images in a microscope - what might you (or others) have heard about AI system evaluations of these biopsy images?

Medical image analysis is something they've been working AI systems with since the 1980s (maybe longer), it seems to be everywhere, so I imagine there are people out there working on it? Thanks.

OK, I should also Google it myself, right? And here's the AI conclusion on this AI question:

Current Status & Future

AI algorithms are reaching pathologist-level performance and are becoming mature for clinical use.

Integration into clinical workflows is ongoing, with AI acting as a "digital colleague" to support diagnosis and prognosis.

Research continues to compare commercial vs. research-based AI and validate performance in diverse real-world settings.