ESTRO 2024 - Abstract Book

S3956

Physics - Image acquisition and processing

ESTRO 2024

Figure 1 Example standard (top) and DL (bottom) reconstructed MR images.

Table 1 Median ± half the interquartile range delineator rating, delineation time and inter-delineator variability (DTA mean).

Delineator Rating

Delineation Time / minutes

DTA Mean / mm

Organ

Standard

DL

Standard

DL

Standard

DL

Prostate

2.0 ± 0.5

3.5 ± 0.8

4.8 ± 1.1

4.4 ± 1.4

1.0 ± 0.2

0.9 ± 0.2

Seminal Vesicles

2.0 ± 0.2

3.5 ± 0.2

4.3 ± 0.2

2.6 ± 0.6

1.3 ± 0.1

1.4 ± 0.2

Bladder

2.9 ± 0.1

3.2 ± 0.2

3.2 ± 0.1

2.9 ± 0.2

0.8 ± 0.1

0.7 ± 0.2

Rectum

2.9 ± 0.1

3.0 ± 0.2

3.2 ± 0.3

2.9 ± 0.2

1.0 ± 0.2

0.9 ± 0.1

Penile Bulb

2.8 ± 0.3

3.1 ± 0.3

0.7 ± 0.1

0.6 ± 0.1

2.0 ± 0.5

3.5 ± 1.1

Figure 2 Delineator rating on the standard (blue) and DL (orange) images. The round markers indicate each patient, the square markers the median rating and the error bars the interquartile range.

Median ratings of the automatic contours on the DL image were 3.5 (intermediate/minor) for the prostate and seminal vesicles, 4.5 (minor/accept) for the bladder, 4 (minor) for the rectum and 3 (intermediate) for the penile bulb. The results on the standard image were equal for the prostate, rectum and penile bulb and within 0.5 for the seminal vesicles and bladder. Automatic contours were generated for each image in under 1 minute. Median times to review and edit automatic contours was very similar between the two images (within 0.2 minutes). Reviewing and editing automatic contours was substantially faster than manual delineation, with a median time saving per-patient of 11 ± 1 minutes (standard) and 10 ± 1 minutes (DL). The smaller time saving for the DL image reflects the shorter manual delineation time for the seminal vesicles.

Conclusion: