Ventricular mechanics and strain

Ventricular mechanics and strain phenotypes describe myocardial deformation beyond ventricular volume and ejection fraction. This page is source-audited for current CardiacNexus output names and processing boundaries, but it deliberately treats strain, strain rate, torsion, recoil, and tagged-MRI outputs as registration-sensitive research phenotypes rather than diagnostic classifiers.

Modality

Cine short-axis CMR, cine long-axis CMR, and tagged MRI where available

UKB source

Data Fields 20209, 20208, and 20211

Pipeline step

MIRTK-backed feature tracking, contour propagation, strain extraction, smoothing, and torsion analysis

Outputs

strain_sax.csv, strain_lax.csv, strain_tagged.csv, strain_sax.npz, strain_lax.npz, strain_tagged.npz, strain and torsion QC plots

Maturity

Source-audited high-risk phenotype page

Clinical question

Ventricular strain can detect impaired deformation even when chamber volumes or ejection fraction look preserved. The current CardiacNexus page is useful for cohort phenotyping and method-aware interpretation, but strain values must be read alongside acquisition, tracking, smoothing, and QC context [1] [2].

Anatomical and physiological definition

Lagrangian strain measures change in myocardial length relative to a reference length, usually end diastole. Circumferential strain describes tangential shortening around the LV cavity, radial strain describes wall thickening toward the LV center, and longitudinal strain describes base-to-apex shortening. In the current output convention, raw circumferential and longitudinal strain values are negative for shortening, while page-level absolute-value outputs are explicitly named as absolute values.

Strain rate is the time derivative of strain and is reported in 1/s. Torsion describes apical-basal twist normalized by ventricular length, and recoil rate describes diastolic untwisting speed from the torsion curve [2].

Source acquisition and UKB field

Short-axis cine strain uses UK Biobank Data Field 20209 and the short-axis segmentation/contour route. Long-axis longitudinal strain uses Data Field 20208 and long-axis myocardial contour tracking. Tagged MRI outputs use Data Field 20211 and a separate tagged-MRI backend; those outputs are kept separate from cine feature-tracking outputs.

What exactly CardiacNexus measures

Short-axis circumferential and radial strain

eval_strain_sax.py calls the short-axis motion/strain route, evaluates tracked myocardial contours, and records AHA-segment and global extrema. Circumferential strain uses the minimum strain value, while radial strain uses the maximum strain value.

Long-axis longitudinal strain

eval_strain_lax.py calls the long-axis motion/strain route and records six segmental longitudinal strain rows plus a global row. Peak and post-systolic features use smoothed global curves, and timing indices require either ventricular timing or ECG-derived RR interval where available.

Strain rate and timing

SAX and LAX strain rates are derived from smoothed global strain curves. The implementation reports peak systolic, early diastolic, and late diastolic extrema when the time-series segmentation and peak checks succeed. SI-DI uses the difference between end-systolic strain and strain at one-third diastolic duration, normalized by end-systolic strain.

Torsion and recoil

eval_strain_sax.py calls the torsion helper after strain extraction. The current outputs include endocardial, epicardial, and global torsion, recoil rate, and time-to-peak torsion. Basal/apical slice selection and tracking consistency are material caveats.

Tagged MRI boundary

eval_strain_tagged.py uses tagged-MRI strain arrays and writes a separate strain_tagged.csv. Its output names should not be merged with cine-derived SAX/LAX strain without method labels, because the backend, deformation model, and QC expectations differ.

Output columns and units

CardiacNexus writes these phenotype families into strain_sax.csv, strain_lax.csv, and strain_tagged.csv. Segmental ranges are exact schema families; the current dictionary records selected global and method-critical rows, while the inventory records the broader artifact contract.

SAX global strain, timing, SI-DI, and torsion

Segmental SAX strain

The current short-axis CSV also emits AHA-segment strain rows:

LAX and tagged outputs

Output reconciliation

Required upstream inputs

• short-axis cine images, short-axis segmentations, and ventricular time-series metadata for SAX strain and torsion;
• long-axis cine images and segmentations for LAX strain;
• tagged MRI files and backend outputs for tagged strain;
• MIRTK and feature-tracking configuration paths from runtime config;
• ECG timing only for RR-normalized timing-index rows.

Reference ranges with cohort and method context

Source-located registry status: reference_range_sources.yml records source locators for strain terminology, CMR torsion/recoil context, automated CMR strain/QC context, and tagged-strain method context. No universal CardiacNexus strain normal range is promoted here because feature-tracking backend, smoothing, sign convention, tagged versus cine route, and registration QC materially change numeric values.

Disease interpretation

Reduced or abnormal strain has been reported in ischemic heart disease, myocardial infarction, cardiomyopathy, hypertrophic cardiomyopathy, heart failure, and infiltrative disease contexts [1] [3]. Torsion and recoil have pressure-load and cardiomyopathy associations in CMR feature-tracking literature [2]. These disease badges are navigation aids only; CardiacNexus does not diagnose CAD, HCM, HF, or valve disease from a strain row.

QC caveats and maturity boundary

Strain is one of the highest-risk CardiacNexus phenotype families. Values depend on contour tracking, MIRTK/feature-tracking backend behavior, temporal smoothing, ED/ES timing, basal/apical slice selection, and segmentation quality. Tagged MRI outputs are backend-specific and should be compared with cine-derived outputs only after method harmonization.

Future backend changes must keep the classical MIRTK route as a fallback until regression checks cover global strain, segmental strain, ED/ES timing, contour or warped-segmentation consistency, temporal smoothness, and deformation plausibility.

Implementation provenance

Source audit

• Current output labels and conditional behavior were checked against eval_strain_sax.py, eval_strain_lax.py, and eval_strain_tagged.py.
• One strain-type schematic is displayed from a page-local public path and registered in docs/data/figure_provenance.yml; permission and exact source-panel review remain pending for draft use.
• docs/data/reference_sources.yml is present and used as the curated reference-source registry for this page.
• Textbook context boundary: page-specific strain definition, torsion, CMR automation, and tagged-MRI context sources are sufficient for draft rollout; Braunwald/Hurst background was not needed for the phenotype-specific output contract.

Related pages

• Registration and feature tracking
• Cine short-axis imaging
• Cine long-axis imaging
• Tagged MRI
• Myocardial mass and wall thickness

References

1. Voigt JU, Pedrizzetti G, Lysyansky P, Marwick TH, Houle H, Baumann R, Pedri S, Ito Y, Abe Y, Metz S, Song JH, Hamilton J, Sengupta PP, Kolias TJ, d'Hooge J, Aurigemma GP, Thomas JD, Badano LP. Definitions for a common standard for 2D speckle tracking echocardiography. Journal of the American Society of Echocardiography. 2015;28(2):183-193.
2. Kowallick JT, Lamata P, Hussain ST, Kutty S, Steinmetz M, Sohns JM, Fasshauer M, Staab W, Unterberg-Buchwald C, Bigalke B, Lotz J, Hasenfuss G, Schuster A. Quantification of left ventricular torsion and diastolic recoil using cardiovascular magnetic resonance myocardial feature tracking. PLoS ONE. 2014;9(10):e109164.
3. Ruijsink B, Puyol-Anton E, Oksuz I, Sinclair M, Bai W, Schnabel JA, Razavi R, King AP. Fully automated, quality-controlled cardiac analysis from CMR: validation and large-scale application to characterize cardiac function. JACC: Cardiovascular Imaging. 2020;13(3):684-695.
4. Ferdian E, Suinesiaputra A, Fung K, Aung N, Lukaschuk E, Barutcu A, Maclean E, Paiva J, Piechnik SK, Neubauer S, Petersen SE, Young AA. Fully automated myocardial strain estimation from cardiovascular MRI-tagged images using a deep learning framework in the UK Biobank. Radiology: Cardiothoracic Imaging. 2020;2(1):e190032. doi:10.1148/ryct.2020190032.
5. Jeung MY, Germain P, Croisille P, Ghannudi SE, Roy C, Gangi A. Myocardial tagging with MR imaging: overview of normal and pathologic findings. RadioGraphics. 2012;32(5):1381-1398.