What it does

Obstetric Ultrasound reads pregnancy scans from the first trimester to the labour ward: standard-plane stills with their calipers and measurement tables, cine loops, and sweeps, including the blind sweeps taken by novice operators with handheld probes. The read is trimester-aware: it sets the gestational-age basis from the crown-rump length, from biometry or from the LMP field, and runs the sections that fit the scan.

It returns the number of fetuses and viability, each standard plane with its quality, biometry with gestational age and estimated fetal weight, growth centiles, amniotic fluid, placenta, cervix and the ISUOG checklist. First-trimester scans add early-pregnancy classification and nuchal translucency; anomaly scans add the fetal heart screen; late-pregnancy scans add presentation and Doppler; labour-ward scans add labour progress.

Intended use

Decision support for obstetric scans that general ob-gyns, family doctors and midwives perform and read, often far from a fetal-medicine unit: dating, growth, the mid-trimester checklist, placenta and fluid, and the labour ward. Twins, caliper overlays and the choice of dating basis are common pitfalls, so every gestational age states its basis and every fetus its identifier.

Who it is for

Obstetricians and gynaecologists, maternal-fetal medicine specialists, sonographers, midwives, family physicians and radiologists. Developers call the run from antenatal-care software, pregnancy registries and handheld-ultrasound programmes.

Accepted input

• Standard-plane stills: DICOM Ultrasound Image objects; burned-in calipers and on-screen measurement tables are read and reconciled with the computed values.
• Cine and sweeps: Ultrasound Multi-frame objects or MP4, including blind sweeps for presentation and fluid.
• Vendor measurements: OB-GYN SR objects (TID 5000) are read as a second source.
• Earlier scans: prior studies in prior_files feed the growth section.

Requirements

Requirement	Why
SequenceOfUltrasoundRegions calibration	Biometry, CRL, NT and cervical length are millimetre measurements; uncalibrated clips give qualitative sections only
Correct standard plane (TT, TV and TC brain, abdominal circumference, femur, mid-sagittal)	Each measurement is taken only on an adequate plane
Cine for cardiac activity and the fetal heart sweep	Heart motion cannot be judged on stills
PW Doppler strips with a calibrated velocity axis	Pulsatility index, PSV and end-diastolic flow are traced from the spectrum
Transperineal images in labour	Angle of progression and head–symphysis distance

Optional context

The LMP or a prior dating scan, maternal age and the clinical question go in clinical_context.

The result

One JSON result in sections with findings, measurements, scores and a pregnancy block (gestational-age basis, number of fetuses, trimester). In twins, every value is tied to fetus A or B.

Section	Content
Standard planes	Plane per frame (brain planes, 4CH, abdominal circumference plane, femur, lips, spine, kidneys, cervix and others) with quality
Early pregnancy	Gestational sac, yolk sac, embryo, cardiac activity, MSD and CRL; intrauterine pregnancy or pregnancy of unknown location; viability category; ectopic signs
Nuchal translucency	NT, CRL, nasal bone and plane adequacy at 11–14 weeks
Fetal biometry	BPD, OFD, HC, TAD and APAD, AC and FL; gestational age by Hadlock or INTERGROWTH-21st; EFW by Hadlock
Growth centiles	Centiles and z-scores from biometry, gestational age and earlier scans; SGA, LGA and FGR flags
Fetal heart screen	Adequacy of the 4CH, LVOT, RVOT and 3VV/3VT views, cardiothoracic ratio and screening flags
Anomaly screen	Mid-trimester checklist completion and anomaly flags by body region
Amniotic fluid	Deepest vertical pocket and AFI; oligohydramnios and polyhydramnios
Placenta	Location, distance to the internal os (praevia or low-lying) and accreta signs
Cervical length	Transvaginal cervical length and funnelling
Fetal Doppler	Umbilical artery PI, RI and end-diastolic flow; MCA PI; MCA PSV in MoM; CPR; uterine artery PI
Fetal presentation	Presentation and lie, number of fetuses and cardiac activity, from stills or blind sweeps
Labour progress	Pubic symphysis and fetal head outlines; angle of progression, head–symphysis distance and head direction
Draft report	Opt-in (options.report): an English narrative that interprets the result for the reader (findings, impression, limitations); every number is checked against the findings

Standards

• ISUOG: biometry and growth 2019, mid-trimester routine scan 2022, fetal cardiac screening 2023, Doppler 2021, intrapartum ultrasound 2018.
• Charts and formulas: INTERGROWTH-21st, Hadlock EFW, Delphi FGR consensus 2016, FMF nuchal translucency criteria, SRU 2013 early-pregnancy nonviability criteria.
• DICOM: SR OB-GYN Ultrasound Procedure Report TID 5000, with TID 5220 for the fetal heart; SEG for the fetal head and other outlines; GSPS calipers; Encapsulated PDF.

Scoring notes

Gestational age comes from the basis stated in the pregnancy block, and centiles are computed by deterministic code from the chart named in the result. SGA and LGA flags use the 10th and 90th centiles. The FGR flag follows the Delphi 2016 consensus, where an AC or EFW below the 3rd centile is sufficient on its own and smaller deviations need Doppler or a fall across centiles. Nonviability follows the SRU 2013 criteria, for example a CRL of 7 mm or more without cardiac activity.

Related models

• us/gynecology reads adnexa in early-pregnancy pain or bleeding.
• us/fast checks for free fluid when an ectopic pregnancy is suspected.

Limitations and failure modes

• First trimester: CRL, early-pregnancy and NT sections are not in version 1.0.0; on a first-trimester study the run returns planes, and biometry is marked outside the formulas' range.
• Handheld blind sweeps: frames from untargeted sweeps rarely pass the landmark-confidence gate; biometry is then not_assessable with the reason, never a guessed number.
• Plane adequacy is judged from the plane confidence and the landmark confidence, not from the individual ISUOG criteria (midline, cavum septi pellucidi, stomach bubble…).
• Twins: fetuses are not separated; a twin study is read as one set of planes.
• Uncalibrated exports (PNG / JPEG / MP4 without options.mm_per_px): diameters in pixels, the cephalic index, and no GA, EFW or centiles.
• Burned-in calipers and measurement tables on the image are not read.
• 3D/4D volumes are not read.