Ventricular function

Feature summary

Badge legend: Phenotype-family colors group feature types only. Status colors are reserved for evidence tables: muted green means verified, muted amber means method-dependent or schema debt, gray outline means needs verification, and red is reserved for blocked items. Disease badges use neutral navigation styling and are literature-context links only, not diagnoses, classifiers, or validated phenotype-to-ICD associations.

Ventricular function phenotypes summarize LV and RV pump performance from the same short-axis volume curves documented on Ventricular structure.

Modality
Cine short-axis CMR
UKB source
Data Field 20209
Pipeline step
Short-axis ventricular volume curve extraction
Outputs
ventricular_volume.csv, timeseries/ventricle.npz, ventricular volume and derivative QC plots
Maturity
Source-audited phenotype page

Clinical question

Stroke volume, ejection fraction, cardiac output, and cardiac index are the core pump-performance summaries derived from ventricular volume curves. They are widely used in CMR reference-range work and clinical interpretation, but CardiacNexus values remain research outputs that require segmentation and cohort context [1] [2].

Anatomical and physiological definition

Stroke volume is the amount of blood ejected during one contraction. Ejection fraction divides stroke volume by end-diastolic volume. Cardiac output multiplies stroke volume by heart rate; cardiac index normalizes cardiac output by body surface area.

Source acquisition and UKB field

The source acquisition is UK Biobank cine short-axis CMR, Data Field 20209. CardiacNexus uses the same ED/ES frame definitions as the ventricular-structure page: ED frame 0 and ES as the LV minimum-volume frame.

What exactly CardiacNexus measures

The functional rows use the structural volumes from the same script. T_ED is frame 0; T_ES is selected from the LV minimum-volume curve and then reused for RV systolic rows.

Pressure-volume loop schematic from the CardiacNexus manuscript image cache citing Silverthorn et al. 2013
Pressure-volume loop schematic from the manuscript image cache, cited there to Silverthorn et al. 2013, shown only as broad pump-physiology context. CardiacNexus computes SV, EF, CO, and CI from cine short-axis volume curves, not from invasive pressure measurements.Source: manuscript image cache citing Silverthorn et al. 2013; permission and exact printed-source wording pending for draft use.
Ventricular volume-time and dV/dt curve used to orient peak filling rate concepts
Volume-time and dV/dt curve from the manuscript image cache, cited there to Aquaro et al. 2019, shown to orient PFR-E and PFR-A timing. Current CardiacNexus promoted ventricular-function CSV rows are SV, EF, CO, and CI; filling-rate context is retained as adjacent physiology unless current emitted columns are promoted.Source: manuscript image cache citing Aquaro et al. 2019; permission and exact source-panel review pending for draft use.

Stroke volume

Copyable formulaSV = V_ED - V_ES

Ejection fraction

Copyable formulaEF = SV / V_ED * 100

Cardiac output and index

CardiacNexus computes heart rate from the number of cine frames and temporal resolution, then multiplies stroke volume by heart rate:

Copyable formulaCO_L_per_min = SV_mL * HR_bpm * 1e-3
Copyable formulaCI = CO / BSA
VariableDefinitionUnitOutput dependency
LV: V_ED [mL], RV: V_ED [mL]ED blood-pool volumes from frame 0mLSV and EF denominators
LV: V_ES [mL], RV: V_ES [mL]ES blood-pool volumes at LV-selected T_ESmLSV numerators
heart_ratederived from cine frame count and temporal resolutionbeats/minCO rows
BSA_subjectreturned by query_BSA(subject)m²CI rows

Output columns and units

Display familyExact output columnUnitStatusSchema note
LV stroke volumeLV: SV [mL]mLcurrentEDV minus ESV
RV stroke volumeRV: SV [mL]mLcurrentEDV minus ESV using LV-selected ES frame
LV ejection fractionLV: EF [%]%currentSV divided by LV EDV
RV ejection fractionRV: EF [%]%currentSV divided by RV EDV
LV cardiac outputLV: CO [L/min]L/mincurrentHR derived from cine timing
RV cardiac outputRV: CO [L/min]L/mincurrentHR derived from cine timing
LV cardiac indexLV: CI [L/min/m^2]L/min/m²current when BSA existsBSA-dependent
RV cardiac indexRV: CI [L/min/m^2]L/min/m²current when BSA existsBSA-dependent

Output reconciliation

Evidence layerResult
Implementation source8 ventricular-function outputs documented here: LV/RV SV, EF, CO, and CI
Output inventorydocs/data/output_column_inventory.yml includes the same 8 outputs under ventricular_volume.functional_columns
Phenotype dictionarydocs/data/phenotype_dictionary.yml links the same 8 outputs to this page
Page output tableall 8 output labels are listed above

Required upstream inputs

  • sa.nii.gz and seg_sa.nii.gz.
  • Short-axis segmentation QC passing sa_pass_quality_control.
  • BSA lookup for cardiac index.

Reference ranges with cohort and method context

FeatureSourceCohortReference valueStatusNote
LV/RV SV and EFPetersen et al. UK Biobank CMR reference ranges [1]UK Biobank CMR reference cohortsex- and age-stratified source rowsVerified context sourceUse body-size and segmentation-method context; not copied as CardiacNexus-specific thresholds
SV, EF, CO, CIKawel-Boehm et al. 2020 CMR reference update [2]multi-study CMR reference contextstudy-specificVerified context sourceMethod conventions vary across studies

Source-located registry status: reference_range_sources.yml maps SV/EF rows to Petersen 2017 LV/RV stroke-volume and ejection-fraction reference tables and Kawel-Boehm 2020 CMR reference-update context. Cardiac output and cardiac index rows are retained as method-dependent because heart rate, BSA, and automated volume-curve conventions must match before a numeric normal range is promoted.

Disease interpretation

Reduced LVEF is central to many heart-failure and post-infarction discussions, and RV EF/SV are important in pulmonary vascular and right-heart disease contexts [1] [2]. These statements are interpretation context only; CardiacNexus does not assign HFrEF, infarction, or cardiomyopathy diagnoses.

QC caveats and maturity boundary

All ventricular function outputs inherit segmentation and ED/ES frame-selection errors from the ventricular-structure route. Cardiac output and cardiac index also depend on correct temporal metadata and BSA. Preserved EF does not imply normal strain, torsion, wall thickening, or tissue phenotype.

Implementation provenance

Current outputs are implemented in src/feature_extraction/Short_Axis_20209/eval_ventricular_volume.py.

Feature familyFormula or computational routeExact output columnsSource code file and functionUpstream dependenciesConditional behaviorQC artifactsSchema debt
Stroke volumeEDV minus ESV for LV and RVLV: SV [mL], RV: SV [mL]eval_ventricular_volume.py; stroke-volume blockLV/RV EDV and ESV from seg_sa.nii.gzsubject skipped if short-axis QC fails before feature calculationraw and smoothed ventricle_volume*.png; ventricle.npzRV SV uses LV-selected ES frame
Ejection fractionSV divided by EDV and multiplied by 100LV: EF [%], RV: EF [%]eval_ventricular_volume.py; EF blockSV and EDV rowsinherits volume QC and division dependenciesvolume time-series QCRV EF uses LV-selected ES frame
Cardiac outputSV multiplied by derived heart rate and converted to L/minLV: CO [L/min], RV: CO [L/min]eval_ventricular_volume.py; CO blockcine frame count, temporal resolution, SVsensitive to temporal metadatavolume time-series QCHR is derived from image metadata, not an external ECG value
Cardiac indexCO divided by BSALV: CI [L/min/m^2], RV: CI [L/min/m^2]eval_ventricular_volume.py; indexed feature block; query_BSABSA lookup tablesubject skipped if BSA is unavailableoutput CSV only; inherits CO QCBSA dependency is external to image files

Source audit

  • SV, EF, CO, and CI formula claims were checked against eval_ventricular_volume.py.
  • Full ventricular-function output coverage was reconciled across the implementation contract, page text, docs/data/output_column_inventory.yml, and docs/data/phenotype_dictionary.yml.
  • docs/data/reference_sources.yml is present and used as the curated reference-source registry for this page.
  • One pressure-volume loop schematic is displayed from a page-local public path and registered in docs/data/figure_provenance.yml; it is broad physiology context only, while current CardiacNexus outputs remain cine-volume-derived.

References

  1. Petersen SE, Aung N, Sanghvi MM, Zemrak F, Fung K, Paiva JM, Francis JM, Khanji MY, Lukaschuk E, Lee AM, Carapella V, Kim YJ, Leeson P, Piechnik SK, Neubauer S. Reference ranges for cardiac structure and function using cardiovascular magnetic resonance in Caucasians from the UK Biobank population cohort. Journal of Cardiovascular Magnetic Resonance. 2017;19(1):18. doi:10.1186/s12968-017-0327-9. PMID:28178995; PMCID:PMC5304550.
  2. Kawel-Boehm N, Hetzel SJ, Ambale-Venkatesh B, Captur G, Francois CJ, Jerosch-Herold M, Salerno M, Teague SD, Valsangiacomo-Buechel ER, Van Der Geest RJ, Bluemke DA. Reference ranges for cardiovascular magnetic resonance in adults and children: 2020 update. Journal of Cardiovascular Magnetic Resonance. 2020;22(1):87.
  3. Silverthorn DU, Ober WC, Garrison CW, Silverthorn AC, Johnson BR. Human Physiology: An Integrated Approach. Pearson Education. 2013.