Coronary disease & ACS

Motion-Informed Neural Network Towards Diastolic Mechanics in Takotsubo Syndrome

Circ Cardiovasc ImagingPublished deep convolutional neural network (2D+t) echocardiographic prediction model

Authors' conclusion

MiNN-Echo in echocardiograms elucidates unique myocardial relaxation kinematics to characterize TTS-associated diastolic mechanics.

Study context

Population
Patients with Takotsubo syndrome (TTS) or anterior myocardial infarctionpatients whose echocardiograms were used, 68 expert echocardiographic readers performed human interpretation · 596 (TTS) · 546 (anterior myocardial infarction)
Participants
1,142 patients
Index test or model
MiNN-Echo vs human readers in differentiating TTS from acute myocardial infarction
  • Human readersn = 68
Target condition
TTS (differentiated from acute myocardial infarction)
Methodological detailsmulti-vendor echocardiograms, multi-institutional cohort, trained and validated
  • Multi-vendor echocardiograms
  • Multi-institutional cohort
  • Trained and validated

Performance

As reported in the abstract

TTS (differentiated from acute myocardial infarction)
Test or modelAUC
MiNN-Echo0.77
Human readers0.69

AUC

MiNN-Echo0.77
Human readers0.69

Further findings

Isovolumic relaxation time

% of diastolic time TTS 28% · Acute myocardial infarction 19%

Passive ventricular filling (before atrial contraction) time

% of diastolic time TTS 44% · Acute myocardial infarction 54%

Median

OutcomeTTSAcute myocardial infarction
Pixel-wise myocardial displacement, isovolumic relaxation phase0.35 (IQR 0.26–0.47)0.38 (IQR 0.28–0.51)
Pixel-wise myocardial displacement, passive ventricular filling phase0.45 (IQR 0.33–0.58)0.50 (IQR 0.38–0.68)
Pixel-wise myocardial displacement, active ventricular filling phase0.43 (IQR 0.32–0.56)0.48 (IQR 0.35–0.63)
Rotation angles (away from the horizontal A-V plane), passive ventricular filling phase−70.19 (IQR −89.62 to −41.33)−60.34 (IQR −87.07 to −30.84)
  • Isovolumic relaxation time: increased in TTS; P <0.001 (ms); P <0.001 (% of diastolic time)
  • Passive ventricular filling (before atrial contraction) time: decreased in TTS; P <0.001 (ms); P <0.001 (% of diastolic time)
  • Pixel-wise myocardial displacement, isovolumic relaxation phase: lower in TTS; P =0.006
  • Pixel-wise myocardial displacement, passive ventricular filling phase: lower in TTS; P <0.001
  • Pixel-wise myocardial displacement, active ventricular filling phase: lower in TTS; P =0.001
  • Rotation angles (away from the horizontal A-V plane), passive ventricular filling phase: altered in TTS; P =0.020

Contribution of diastolic myocardial motion features to the differential diagnosis

Diastolic motion features contributed more than those during systole
≈71%

Original publication

Motion-Informed Neural Network Towards Diastolic Mechanics in Takotsubo Syndrome. Circ Cardiovasc Imaging. Published 11 November 2026. doi:10.1161/circimaging.126.020258

Composed by AI from the published abstract and checked against it. How Briefs are made →

Get them every week

New studies on this topic in My Week and in your Monday e-mail.

We e-mail you a sign-in link and a code. No password, no spam.