Kirti Magudia, MD PhD
@KMagudia
Followers
1K
Following
4K
Statuses
777
Assistant Professor @DukeRadiology | @UCSFimaging, @BWHRadEdu, & @TriIMDPhD alum | #AI researcher #RADxx
Joined December 2018
Many thanks to my fantastic mentor @MRosenthalMDPhD and our exceptional collaborating team for this study showing that fully automated and normalized CT-based body composition metrics are associated with MI and stroke @BWHRadEdu @MGHImaging @DukeRadiology
AI-based analysis of body composition on routine abdominal CT exams can predict a patient's future risk of a heart attack or stroke, according to research led by @KMagudia of @MGHImaging in AJR. @ARRS_Radiology #radiology #medicalimaging
3
9
42
RT @Radiology_AI: The @RSNA Abdominal Traumatic Injury CT (RATIC) Dataset @BFlanksteak @KMagudia @AnaVillanuevaCa #…
0
3
0
RT @DukeRadiology: Thinking about where to go for a Radiology Residency? Come to @DukeRadiology's online First Look Event on Wednesday, Aug…
0
4
0
RT @SAR_RFS: Heads up #RadRes, #RadFellows, #FutureRadRes, #RadTrainee! 💥Only 3 DAYS remain to sign up for the SAR-RFS Core Exam Board Rev…
0
1
0
Congrats to @miriam_chisholm, #medstudent at @dukemedschool, whose @RSNA grant funded to investigate the interplay between social determinants of health and body composition was funded! Can't wait to work on this with you and @tdtailor! @DukeRadiology #RandEFoundation
1
3
23
RT @Radiology_AI: Are you in training in #radiology or an #AI-related field? Apply today for the @Radiology_AI Trainee Editorial Board! htt…
0
5
0
RT @DrArleo: Congratulations @DrGMcGinty, ACR Gold Medal winner!!! To learn more about her ongoing outstanding achievement, see: https://t.…
0
11
0
RT @ChrisJRGallo: Are you a 4th year medical student interested in a diagnostic or interventional radiology away rotation at Duke Universit…
0
19
0
RT @MariamAboian: Strongly agree @KMagudia! It is just if we add differences in annotations between difft datasets, we are adding another v…
0
1
0
@ShahriarFaghani @khosravi_bardia @MariamAboian Real world clinical data will throw challenges to any AI tool. Don't we want these tools be able to perform with the variation in protocols that are out in practice? #RadAIchat
2
0
5
RT @ShahriarFaghani: What is the ideal reference standard for the task at hand? If it's human, how many annotators are necessary? What shou…
0
2
0
@Radiology_AI It is very resource intensive to create AI ground truth, thus optimizing the definition of AI ground truth is worth significant thought. Yet another reason why all stake holders should be included before proceeding. #RadAIchat
0
1
9
@khosravi_bardia @Radiology_AI I agree that this gets very confusing when you start talking about internal validation, internal test, external validation datasets. We ran into this issue when writing this paper #RadAIchat
0
1
3
@khosravi_bardia @Radiology_AI Curious as to what others think??? Going with the widespread term used for this dataset versus being specific about what it is being used for... #RadAIchat
2
0
4
@khosravi_bardia @Radiology_AI The issue of bias is key. Using healthcare data for AI groundtruth can encode systemic biases, propagating healthcare disparities. How do we get around this? #RadAIchat
1
0
5
@khosravi_bardia @Radiology_AI Again, it depends! Mostly in my experience, it depends on resource limitations. Ideally, multiple radiologists for image interpretations, but again not always feasible. #RadAIchat
0
0
6