caret icon Back to all discussions

Is the PI-RADS score determined by computer analysis or by humans?

Hi! I've been away from the site for a while, as I've been having surgery for what appeared to be bladder cancer (but the CT scan was in error, and what seemed to be a urothelial carcinoma turned out to merely be debris!) Because my bladder surgeon said that the prostate MRI had to be performed before the bladder surgery (because it would cause tumult in the prostate that would distort the images of the MRI), I had the MRI on May 20th. I'll soon be having a discussion of the results of my prostate MRI, and I'll be given my PI-RADS score (I assume). I realized I didn't know if the PI-RADS score was determined by computer software or artificial intelligence-- or a human being examining the images. So that's my question: how is the PI-RADS category arrived at?

  1. PI-RADS scores are determined by specialized radiologists who should presumably be able to determine a uniform assessment. Of course, like any fils study, it’s subjective, but two competent radiologists at leading hospitals should be able to concur on scores, particularly under the relatively new PI-RADS 2 protocol

    1. Hi ajeffrey10, I apologize for my delay in responding to you-- I assure you it doesn't mean I'm not very appreciative, just ridiculously busy!


      So first, I wanted to ask you why you said TWO competent radiologists-- is there some protocol that requires (or at least strongly recommends) multiple brains be involved in the assessment? I see that at the hospital where I had my prostate MRI, evidently only a single radiologist did the analysis. Should I regard that result with a degree of skepticism?


      Second, more importantly, I want your opinion (since you seem uncommonly knowledgeable about these matters) on a subject of great concern to me. I went to my current hospital precisely because they promoted the fact that they do fusion biopsies of the prostate, electronically fusing the information gleaned from the mpMRI with the normally 'dumb' ultrasound, to yield an MRI-guided TRUS that is utterly brilliant and far more capable of finding cancer of the prostate when it exists, and with fewer samples taken, than the old standard, the TRUS.


      But now, as it turns out, I have a semi-celebrity urologist, and I think what he does is a 'cognitive' MRI-guided TRUS, which means that there's no actual fusion of the MRI images with the ultra-sound, but rather, the urologist studies the MRI, gets a 3-dimensional picture of the patient's prostate in his head-- carefully noting the suspicious areas--and uses THAT mental image to guide his ultrasound biopsy. I read an article on these 'cognitive' MRI-guided biopsies and the urologist who wrote it was quite equivocal on whether they are to be preferred to fusion biopsies. Evidently, it's entirely a matter of the experience and skill of the urologist doing the cognitive MRI-guided biopsy, which can outweigh the value of a fusion biopsy if the fusion is in the hands of just an AVERAGE urologist. (Fusion biopsies have their own problems, it seems, making them harder to conduct successfully than you might imagine.)


      So, ajeffriey10, what do you think? If I learn that my urologist is definitely intending to do a cognitive MRI-guided biopsy instead of a fusion, should I take a leap of faith and accept that his decades of experience (and his ability to hold in his head an accurate 3-D image of my prostate's suspicious areas) will actually make it preferable to a fusion biopsy?

      1. Hi . I understand your question and concerns about the human element in reading scans and tests. explained the current method. I'm actually going to change your question a little and offer you some information on the accuracy of PI-RADS in distinguishing between benign and cancerous prostate lesions under the traditional reading method. This study found it to be quite accurate: https://www.sciencedirect.com/science/article/pii/S0378603X16300596. Next, I want to offer you some recent information on introducing AI into the process to support PI-RADS scoring: https://www.quantib.com/blog/the-basics-of-pi-rads-scoring-and-ai. Hopefully any assist from AI will only improve that further. You doctor can hopefully provide additional information. Best, Richard (ProstateCancer.net Team)

        1. Hey . To be clear, I said “two radiologists” only to emphasize the fact that PIRADS2 Is a standard such that any two radiologists reading an mpMRI would reach the same scoring conclusions. Of course, there’s always subjectivity, but PIRADS2 is designed to reduce confusion and generate more consistent scores.


          I am a layperson, not an MD. I’ve just done a ton of research around my own situation and I think I understand conceptually how PIRADS2 works. As to Fusion MRI or in-bore MRI or cognitive MRI, I really can’t opine. My oncologist is at UCSF, which I consider a top-5 hospital anywhere in the world, and he told me mpMRI and fusion biopsy are standard of care. I believe him, but you should absolutely do your own research.


          Hope that helps

          Please read our rules before posting.