Automated Follicular Assessment Using a Novel Two-dimensional Ultrasound-based Solution.

Area of Science:

• Reproductive medicine
• Medical imaging
• Software development

Background:

• High variability in manual follicular assessment using 2D ultrasound (US) poses a clinical challenge.
• Developed a novel software for automated follicle count and diameter measurement from 2D US images.
• The software aims to improve objectivity and reproducibility in ovarian assessments.

Purpose of the Study:

• To compare the accuracy of an automated software solution against manual 2D US assessment for follicular count and diameter.
• To evaluate the software's performance in reducing intra- and interobserver variability.

Main Methods:

• Software developed using follicular US sweeps from 54 subjects.
• Validation phase involved 10 subjects with manual 2D US recordings.
• Follicle count and diameters (≥5 mm) were recorded by sonologists and the software.

Main Results:

• High correlation (0.787) for total follicle count between automated and manual methods.
• Excellent correlation (0.817) for mean follicular diameter measurement.
• The software showed a tendency to slightly underestimate mean diameter but performed better than manual observer variability.

Conclusions:

• The automated solution is feasible for follicle count and diameter assessment with accuracy comparable to manual 2D US.
• The software's performance surpasses known intra- and interobserver variability.
• Further validation is recommended to confirm results and assess potential time savings.