computer trained to read Lung cancer cases

[nr-xxx]

Advertisement - Members don't see this ad

please go to www.medscape.com and read article by A. Castellano about Compute trained to read Lung cancer cases. it had higher accuracy than a trained pathologist it can read H & E slides. My guess is that other cancers shall be next. I would not be surprised if the classic pathologist was phased out in the next 10-15 years. With HPV testing the PAP test shall be obsolete in about 5 years. now with this computer program it is a matter of time for surgical pathology to be curtailed. When I was in private practice about 10 years ago I was at a well respected group in Florida. A solid , well thought of,practicing surgical pathologist once remarked to me that "if they ever get a number or a value that tells the clinician something is either benign or malignant and can plug that into an oncology protocol for treatment, then we shall all be out of a job" it looks that such a situation is in the near future. . let a word to the wise be sufficient. pathology may not be a viable specialty in the not too distant future

 

[HmanPath]

Here is a link to the actual article, not just medscape.

http://www.nature.com/articles/ncomms12474#f3

 

[sc1085]

http://www.archivesofpathology.org/doi/pdf/10.5858/arpa.2016-0046-ED

 

[tiredguy]

Will the computer also need an R-0-1?

 

[Mad Jack]

People always forget that computers cannot absorb liability, and thus a pathologist we'll be required for any final read just so there is someone to sue.

 

[Sulfinator]

I say, "Meh. Interesting, but meh."


Sent from my iPhone using SDN mobile

 

[icpshootyz]

Ah yes, the demise of pathology is yet again portended. Not the first time, won't be the last.

 

[dr4n6]

NR-XXX: Thanks for your sage advice, I shall file your nugget of wisdom right up there with my concerns about the dogs trained to dx lung ca's and the pigeon that can diagnose breast ca with also a reportedly high specificity.

 

[WEBB PINKERTON]

When pathology is "replaced" it won't be by machines reading H+E stained slides. That is a bit archaic.

 

[yaah]

"One limitation of this study is that cases submitted for TCGA and TMA databases might be biased in terms of having mostly images in which the morphological patterns of disease are definitive, which could be different from what pathologists encounter at their day-to-day practice."

You think?


Meh.

 

[Torsed]

How does the computer read atelectasis?

 

[12751]

Dr. Kun-Hsing Yu is first author on this paper and a friend of mine from some classes at Stanford. I wont try to speak for him because that would be inaccurate and inappropriate but I can report this much: Kun seems to be a genuine, honest, and supportive guy and his conscience is apparently untouched by knowing that I'm about to start a Pathology residency. He tends to focus on the decision support function that this work may eventually lead to. I'm excited that this body of work may eventually be helping doctors and patients understand prognoses better, and who knows that might clue us into treatment options that are more precise.

But... Hey if you want to freak out about robots taking over the world I can help you out with that too! Check out this video: -