Stanford team trains computer to evaluate breast cancer - End of pathologists?

[Dock1234]

http://med.stanford.edu/news/all-ne...rains-computer-to-evaluate-breast-cancer.html

So will these technology put pathologists out of job in next 10 years?

---

Members don't see this ad.

 

[coroner]

Ummm....no. Read the article

Matt van de Rijn, MD, PhD, a professor of pathology and co-author of the study. “The computers are pointing us to what is significant, not the other way around.”
Van de Rijn does not see computers replacing pathologists. “We’re looking at a future where computers and humans collaborate to improve results for patients across the world,” he said.

This is used for a grading system after the diagnosis has been made to assess prognosis more accurately than traditional human interpreted SBR/Nottingham score. Similar to using computer image analysis to quantify ER,PR, and Her-2. The tissue still has to get stained with H&E and passed under the scope of a pathologist just like it has been for the past century...

 

[Over9000]

It won't replace pathologists but it will change the field's dynamics. The overall diagnosis will probably still be done by paths, but the grading/counting mitoses scut might be done by computers. I welcome this change.

We've relied on H&E for too long, as if its the only possible way to diagnose, without more research into other complementary or better modalities. Most of the academic research concerns IHC patterns, but very little of it is higher-end stuff like this.

My biggest concern is that other fields will adopt these new technologies as their own, like what is happening with in-vivo microscopy, further reducing our market share in medicine. Pathology isn't doing much in regards to research in these new, exciting fields.

 

[WEBB PINKERTON]

The pipe dream of interventional pathology will likely never happen or if it does it will be on small scale. It makes sense that those on the front lines are going to own the technology. Pathology keeps pushing itself further and further away from the patients, especially as hospital chains/corporate labs are buying up practices and consolidating the AP work to massive core labs. Who do you expect to own the new tests? I've been working with our pulmonologists and the manufacturer of Cell Visio. It is amazing technology but I know it is going to chip away at our field even more. Disruptive technology may not completely eliminate the field but it is going to reduce the number of biopsies significantly.

Dogs are also being trained to sniff out cancer. Maybe we should own that technology first!

 

[Granular]

We perceive the takeover by computerized algorithms as a threat only because we also believe that the software will be cheaper than a pathologist. But since $10/case seems to be the lowest available rate for a pathologist, perhaps the software will be more expensive; I know we already spend tons of money on interfaces, maintenance contracts, renewing software subscriptions, etc. And programmers aren't cheap, either. My hospital employer cannot pay enough to retain any of them.
Although we have many reasons for a bleak outlook in pathology, being replaced by machines in the next 25 years is not one of them.

 

[xy313]

Radiology went digital much faster than pathology, because the image sizes are smaller.

And although CAD in mammography is way ahead of anything in digital pathology, it has not replaced radiologists.

 

[pathstudent]

I could see it going that way one day. Much of surgical pathology is classification based on low power pattern recognition. Computers could probably get pretty good at that.

One day the patient will enter the radiology suite and get biopsy. Half the core will be split and go to histology with automated classification and the other half will go to gene sequencing and voila the oncologist will know all he needs to know.