Standing Position Ultrasonography: Imaging Efficacy in Detecting Liver Diaphragmatic Dome Blind Area Lesions

Patient Collection:

Enroll patients hospitalized in our Hepatobiliary-Pancreatic Department between February 2024 and February 2025 with subphrenic liver lesions confirmed by CT or MRI. These patients underwent ultrasound examinations. Collect and record demographic data (sex, age, height, weight, BMI) and lesion characteristics (size, location, type, liver background).

Inclusion & Exclusion Criteria

Inclusion Criteria:

1. Liver dome lesions (≤3 cm) confirmed by CT/MRI (plain + enhanced) within 6 months (for multiple lesions, include the one closest to the diaphragm);
2. Superior border of lesion ≤1 cm from the diaphragm;
3. Lesion size ≤3 cm.

Exclusion Criteria:

1. Inability to cooperate with three-position ultrasound examination;
2. Subcutaneous emphysema affecting ultrasound;
3. Chilaiditi syndrome (interposition of colon).

Equipment & Methods

Equipment:

GE LOGIQ-E9 ultrasound system with C1-6-D low-frequency convex probe.

Procedure:

1. Patients fasted for 2-3 hours and rested for 10-20 minutes before examination.

2. Three scanning positions (Fig 2.1):

   • Supine: Arms raised above head, fully exposing intercostal/subcostal areas; back parallel to bed.

     • Left Lateral Decubitus: Arms raised, fully exposed intercostal/subcostal areas; side-facing away from examiner; back perpendicular to bed.

       • Standing: Arms raised, fully exposed intercostal/subcostal areas; upright with body axis vertical to floor.

Ultrasound Image Acquisition:

Three sonographers (>3 years' experience) randomly selected one position per patient. Scanning included intercostal/subcostal approaches. Patients cooperated with breathing/positioning. If a lesion was detected, the clearest image was saved; if not, no image was captured. Sonographers were blinded to lesion presence and prior CT/MRI results. Each sonographer left the room after scanning their assigned position.

Image Quality Evaluation:

Evaluation:

Two blinded radiologists (>3 years' experience) independently scored lesion visibility. Discrepancies were resolved by re-evaluation and consensus. Evaluators were blinded to patient position.

Scoring System:

1. 2 = Fully visible: Clear lesion without gas/bone shadowing; complete margins; measurable size.
2. 1 = Partially visible: Suboptimal visualization; obscured by gas/bone; incomplete margins; unmeasurable size.
3. 0 = Not visible: Undetectable.

Sonographer Consistency Test:

20 randomly selected patients were re-scanned by all three sonographers across all positions. Each sonographer recorded lesion detectability per position.

Bias Control:

Selection Bias: Consecutive enrollment of eligible patients. Sonographer Bias: Blinded scanning; no prior knowledge of CT/MRI/history; Fleiss' kappa for inter-operator consistency.

Reader Bias: Independent blinded scoring with consensus for discrepancies.

Quality Control:

Strict adherence to protocol; trained/certified personnel; standardized data collection/entry/analysis; complete records; predefined statistical methods; unbiased reporting.

Statistical Analysis:

1. SPSS 26.0 software. Fleiss' kappa assessed inter-sonographer agreement. Normally distributed data: mean±SD; non-normal: median (IQR). Friedman test compared scores across positions; Wilcoxon signed-rank test for pairwise comparisons. Significance: P<0.05.
2. Subgroup Analysis: Stratified by BMI (normal/underweight, overweight/obese), liver background (normal, fatty liver, liver damage [diffuse disease/cirrhosis]), lesion location (S4, S7, S8), size (0-1 cm, 1.1-2.0 cm, 2.1-3.0 cm), and type (HCC, non-HCC). Non-parametric tests repeated per subgroup.
3. Generalized Estimating Equations (GEE): Binary logistic model (patient ID as subject variable; outcome: complete visibility) with exchangeable working correlation matrix. Covariates: position (supine as reference), location (S8 as reference), liver background, lesion type, size, and BMI. Results: OR (95% CI). Sensitivity analysis compared exchangeable, AR1, and unstructured correlation matrices.