Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Fig 1.

Fetal positions.

Illustration of cephalic and non-cephalic fetal positions. Non-cephalic fetal presentations are important to identify prior to delivery.

More »

Fig 1 Expand

Fig 2.

Placental positions.

Illustration of different placental positions. The fundal position straddles anterior and posterior positions. Fundal, anterior, and posterior placental positions have no significant clinical impact. However, they are vital to distinguish from placenta previa which can be life-threatening.

More »

Fig 2 Expand

Fig 3.

Biometry measurements of the head circumference and biparietal diameter.

Head circumference (ellipse) and biparietal diameter (line) measurements are illustrated.

More »

Fig 3 Expand

Fig 4.

Proposed system for automatic obstetric diagnosis without an experienced sonographer or radiologist.

Schematic diagram shows utilization of VSI combined with U-Net for rapid automatic image interpretation. The blue arrows signify the input of information into the automatic diagnostic framework. The green arrows signify the output of information.

More »

Fig 4 Expand

Fig 5.

Obstetric ultrasound volume sweep imaging protocol.

Poster depicting each step in the VSI protocol.

More »

Fig 5 Expand

Fig 6.

Representative 2D ultrasound images from VSI acquisition.

Red arrows identify the head (A) and placenta (B).

More »

Fig 6 Expand

Fig 7.

Image information contained in the training data set.

The table shows the number of images labeled (blue) and without labels (orange). Bar plots indicate the average number of images for each classification. The error bar depicts the standard deviation between sets in leave-one-out cross-validation.

More »

Fig 7 Expand

Table 1.

Set formation using leave-one-out cross-validation.

More »

Table 1 Expand

Fig 8.

Overview of the proposed training method.

(a) Automatic segmentation using the U-Net model [21]. (b) Fetal presentation prediction. (c) Placental location prediction.

More »

Fig 8 Expand

Fig 9.

Scheme for the generation of the spatial location likelihood of the fetal head.

Stage 1: The zero-matrix represents the pregnant abdomen, and it is formed considering the largest number of frames for horizontal and vertical sweeps. Sweeps with fewer frames are re-scaled to properly form the matrix. Stage 2: The frames containing the fetal head are colored. Stage 3: A Gaussian filter is applied only for representative purposes to finally produce the spatial location likelihood. Based on this map, the algorithm produces a diagnosis.

More »

Fig 9 Expand

Fig 10.

Ultrasound images.

2D ultrasound images from two representative patients (patient 1, upper row; patient 2, lower row) with overlaid color masks showing actual (yellow) versus predicted locations of the fetal head (blue) and placenta (green).

More »

Fig 10 Expand

Fig 11.

Heatmaps.

Heatmaps of two representative patients showing the spatial locations of the fetal head and placenta. (A) Cephalic fetal presentation. (B) Non-cephalic fetal presentation. (C) Placenta anterior. (D) Placenta posterior.

More »

Fig 11 Expand

Table 2.

Average detection and segmentation metrics for leave-one-out cross-validation.

More »

Table 2 Expand

Table 3.

Comparison of qualitative diagnostic assessment of fetal presentation and placenta location assigned by an obstetrician from VSI versus automatic diagnosis.

More »

Table 3 Expand

Table 4.

Comparison of quantitative diagnostic assessment of fetal head circumference and biparietal diameter assigned by an obstetrician from VSI versus automatic diagnosis.

More »

Table 4 Expand

Table 5.

Comparison of qualitative diagnostic assessment of fetal presentation and placenta location assigned by a radiologist from standard of care imaging versus automatic diagnosis.

More »

Table 5 Expand

Table 6.

Comparison of quantitative diagnostic assessment of fetal head circumference and biparietal diameter assigned by a radiologist from standard of care imaging versus automatic diagnosis.

More »

Table 6 Expand

Table 7.

Analysis of gestational age accuracy across U-Net, VSI, and standard of care imaging.

More »

Table 7 Expand