1/4 (4-tweet thread)
A recent study in @ModernPathology
by Xie et el examined the ability of a deep convolutional neural network to diagnose which tumor in pleural effusions? #pathpolls
3/4
The authors claim the sensitivity and specificity of the deep convolutional neural network was somewhere between senior and junior cytopathologists. What were the numbers (sensitivity, specificity)?
cc: @DrNetto
4/4 Answer: the sensitivity of the neural network for a malignant call in this narrow setting was 87.5% and the specificity was 94.4%
Interesting paper! Open access! Give it a read if you can 🙏🏾
1/ Announcing a contest to win a free pathology textbook!
Thanks to a brilliant idea by @DrMissWV and a generous offer by @science_press and @LizMontgomeryMD we are giving away a few free textbooks in an educational tweet contest called #pathbracket . See thread 👇🏾
2/
If you want to enter the contest, create a NEW educational pathology tweet between today and March 15 and tag it with #pathbracket
You must tag your own tweet. The tweet must have educational value. Tweetorials are allowed too. Contest is open to all.
3/ On March 16, #PathTweetAward judges led by @DrGeeONE will donate their time to pick 32 of the best tweets to enter in a bracket similar to #MarchMadness . Credit with the idea of a bracket goes to @DrMissWV
Progression through the bracket will be on the basis of polls.
1. How do we tell primary lung cancer from metastatic cancers from other sites?
It’s correct to say we should use all available clinical, radiologic and pathologic information, but that’s too vague to be helpful in a practical sense.
2. This #pathtweetorial is applicable mainly to #pulmpath . The principles in each organ are different. A carcinoma in a lymph node is a metastasis by definition. Not so in lung or colon.
IMHO it’s a mistake to approach a carcinoma in the lung as carcinoma of unknown primary
3. Most metastatic carcinomas to the lung occur after the primary carcinoma has already been diagnosed in another site.
💥Clinicians: give your pathologists this information! If you don’t, this is substandard, dangerous patient care.
I’m starting a thread on #COVID19 just to keep everything I learn about it in one place.
🌹 New information
🌹 well summarized information
🌹 Things that changed my mind
🌹 Things that I did not expect
2. Great update with Dr. Fauci
Most interesting tidbit: why is well controlled hypertension a risk factor? Could ACE inhibitors be predisposing to coronavirus infection by increasing ACE receptors?
Link to video 👇🏾
3. Podcast episode 3/17 by @nytimes . Very moving interview with a pulmonologist who has treated dozens of patients in Bergamo, Italy. Warning: it will make you cry.
“Nothing is as before”
They see 50-70 patients with severe pneumonia every single day
1/ How about a #Tweetorial on immunotherapy in quiz and answer form? Let’s try it! I’ll post a quiz in one tweet and then the answer in the next, and so on. Scroll down to follow the thread!
2/ I did a quick search to find the first time “talc granulomatosis” was used as a term for lung lesions caused by excipients in pills, and came across this 1970 paper. Were Hopkins and Taylor the first to coin the term? Correct me if I’m wrong @yro854