Validating Blind Ultrasound Sweeps Performed by Non-Expert Ultrasound Operators in First Trimester Pregnancies
Poster #: 154
Session/Time: A
Author:
Madeleine Blue Peterson, BA, MS
Mentor:
Andrew Moore, MD
Research Type: Public Health
Abstract
INTRODUCTION:
Ultrasound is an essential imaging modality for providing quality obstetric care. However, the costs of equipment, need for trained sonographers, and costs of on-site radiologists present limitations in rural and under-resourced communities, contributing to a gap in care for pregnant patients. We tested a diagnostic framework involving standardized blind sweep imaging to address this gap in patients in their first trimester of pregnancy. This type of protocolized blind sweep obstetric ultrasound exam was previously investigated in both low-risk and high-risk second and third trimester pregnancies. Our objective was to assess the reproducibility of expert interpretation of blind ultrasound sweeps performed by non-experts using a low-cost, portable ultrasound (US) device to evaluate first trimester pregnancies.
METHODS:
In this multi-center prospective cohort study (2/28/2025-7/15/2025), first-trimester pregnant individuals underwent blind ultrasound sweeps by non-expert operators using a point-of-care probe. Operators received standardized instruction on an eight-step scanning protocol. Images were independently reviewed by six blinded experts: two generalist OBGYNs, two MFMs, and two MFM trainees. The primary outcome was inter-rater agreement on key early pregnancy findings (gestational sac, yolk sac, fetal pole, number of fetuses, fetal cardiac activity) using Fleiss' kappa. Secondary outcomes included examination and image quality rated by experts, with final classifications by majority vote. Analyses were conducted in R.
RESULTS:
Of 1,958 patients screened, 394 were eligible, and 109 were included in the final analysis (see Figure 1). Median BMI was 27.5 (IQR 8.6), and median gestational age at scan was 12.6 weeks (IQR 1.6). Among those scanned, 105 were singleton and 4 were multiple gestations. Inter-rater agreement was almost perfect for identifying a gestational sac (κ = 0.86), substantial for yolk sac (κ = 0.66), multiple gestation (κ = 0.71), and fetal cardiac activity (κ = 0.78), and moderate for identification of a fetal pole (κ = 0.51). Experts found the majority of images had excellent or acceptable image quality, but there were limitations of imaging in early pregnancies among individuals with abdominal adiposity (Table 1).
CONCLUSION:
Blind US sweeps guided by external anatomic landmarks and performed by non expert US operators using a battery-powered handheld device produced images interpretable with moderate to almost perfect agreement by expert reviewers. This technique may provide a scalable, low-resource solution for expanding access to early pregnancy US. Further studies assessing diagnostic accuracy of this approach in first trimester pregnancies are warranted.
Ultrasound is an essential imaging modality for providing quality obstetric care. However, the costs of equipment, need for trained sonographers, and costs of on-site radiologists present limitations in rural and under-resourced communities, contributing to a gap in care for pregnant patients. We tested a diagnostic framework involving standardized blind sweep imaging to address this gap in patients in their first trimester of pregnancy. This type of protocolized blind sweep obstetric ultrasound exam was previously investigated in both low-risk and high-risk second and third trimester pregnancies. Our objective was to assess the reproducibility of expert interpretation of blind ultrasound sweeps performed by non-experts using a low-cost, portable ultrasound (US) device to evaluate first trimester pregnancies.
METHODS:
In this multi-center prospective cohort study (2/28/2025-7/15/2025), first-trimester pregnant individuals underwent blind ultrasound sweeps by non-expert operators using a point-of-care probe. Operators received standardized instruction on an eight-step scanning protocol. Images were independently reviewed by six blinded experts: two generalist OBGYNs, two MFMs, and two MFM trainees. The primary outcome was inter-rater agreement on key early pregnancy findings (gestational sac, yolk sac, fetal pole, number of fetuses, fetal cardiac activity) using Fleiss' kappa. Secondary outcomes included examination and image quality rated by experts, with final classifications by majority vote. Analyses were conducted in R.
RESULTS:
Of 1,958 patients screened, 394 were eligible, and 109 were included in the final analysis (see Figure 1). Median BMI was 27.5 (IQR 8.6), and median gestational age at scan was 12.6 weeks (IQR 1.6). Among those scanned, 105 were singleton and 4 were multiple gestations. Inter-rater agreement was almost perfect for identifying a gestational sac (κ = 0.86), substantial for yolk sac (κ = 0.66), multiple gestation (κ = 0.71), and fetal cardiac activity (κ = 0.78), and moderate for identification of a fetal pole (κ = 0.51). Experts found the majority of images had excellent or acceptable image quality, but there were limitations of imaging in early pregnancies among individuals with abdominal adiposity (Table 1).
CONCLUSION:
Blind US sweeps guided by external anatomic landmarks and performed by non expert US operators using a battery-powered handheld device produced images interpretable with moderate to almost perfect agreement by expert reviewers. This technique may provide a scalable, low-resource solution for expanding access to early pregnancy US. Further studies assessing diagnostic accuracy of this approach in first trimester pregnancies are warranted.