RT @RadiologyEditor: Radiology: In Training (RIT) is a year-long RSNA initiative offering radiology, nuclear medicine, and imaging sciences….

In this review, we explore a rare and under-recognized entity called Unilateral CNS Vasculitis! @TheAJNR . #neurotwitter #radiology #MedTwitter #radres.

What do you want to know?.

9/9 🎯 Key takeaway:.Sample size isn’t just stats—it’s about responsible research, meaningful outcomes, and ethical design. Consult a statistician early & always report transparently!. 📖 Full details: @RITEditor @VChernyakMD @RadiologyEditor @DrLindaMoy.

8/9 🤖 Sample size for AI studies:.AI needs sufficient data to avoid overfitting & ensure robustness. Currently, there are no standard guidelines; Use “learning curves” to determine optimal size. 👇

7/9 📉 What if you didn’t estimate sample size beforehand?.• Comparative studies: Report 95% CI; clarify smallest detectable difference post-hoc.• Descriptive studies: Ensure your 95% CI isn’t too wide to be clinically useful (e.g., prevalence CI of 20–80% is unhelpful!).

6/9 🧐 Equivalence studies → why do you need bigger samples?.Demonstrating equivalence (no meaningful difference) between tests requires larger samples than typical comparative studies. The entire 95% CI must fit within the equivalence range (±Δ). 👇

5/9 ⚠️ Sample size too large? Don’t panic:.• Use continuous variables (more precise). • Opt for paired/repeated measures. • Improve measurement precision (e.g., high-res imaging). • Accept wider 95% CI if patient numbers aren’t logistically feasible.

📝 Sample Size Quiz! What’s the usual α-level for statistical significance in research?. Key parameters for sample size calculation: Minimal Clinically Important Difference (MCID)—smallest meaningful effect, α-level, Power (1-β), Attrition/Dropout rate.

3/9 🧪 When to estimate sample size?.Two main study types:.• Comparative (e.g., RCTs comparing modalities, agents, AI models).• Descriptive (e.g., prevalence studies, average radiation dose calculations).If you report a P-value or 95% CI → a sample size estimate is implied!.

2/9 🧮 Why estimate sample size?.• Ensures meaningful results, whether positive or negative.• Avoids Type II errors (missing a true effect).• Ethical: Minimizes unnecessary radiation or risks to participants.Remember: Too small = inconclusive; Too big = wasteful & unethical.

1/9 🔎 Ever had your paper questioned over sample size?.Determining the right sample size is critical—not only for statistical rigor but also ethical integrity. Whatever project you are planning, here’s what researchers often overlook! #RadRes #RadInTraining @radiology_rsna.

Just another day in academia….

RT @AMCancerImaging: 🧠✨ New in Discover Oncology:.“Radiomics in pediatric brain tumors: from images to insights”.Explore how advanced imagi….

RT @MohitTCCNeuro: Are you ready for the first ever mentoring session of its kind at #ASFNR25? @theASFNR and @thecortexclub collaboration b….

RT @TheAJNR: "Neuroimaging Spectrum of Erdheim-Chester Disease: An Image-Based Review". @pranj_rad; @tmhradiology;….

The purpose of this write-up was just a nudge to the younger academics out there, if you have worked on it, make sure it’s seen! All the best! 🧠 . Here’s the link for those interested: #MedTwitter.

I worked on this article when I was in med school. Never got to publishing due to some or the other delay. Now it’s out after 4 years, and while it’s not even remotely related to radiology, it feels good finally getting it out there for my OCD brain!

RT @AlessiaGuarnera: May #acupuncture improve the management of post-concussive symptoms (PCS) in mild traumatic brain injury (mTBI)? Check….

RT @DocNavarrow: ✅ Erdheim-Chester disease. Consider ECD in the differential diagnosis of atypical enhancing brain lesions when lesions are….

Step 2 ✅