Generating controllable ultrasound images of the fetal head
Synthesis of anatomically realistic ultrasound images could be potentially valuable in sonographer training and to provide training images for algorithms, but is a challenging technical problem. Generating examples where different image attributes can be controlled may also be useful for tasks such...
| Main Authors: | , |
|---|---|
| Format: | Conference item |
| Language: | English |
| Published: |
IEEE
2020
|
| _version_ | 1826270751760580608 |
|---|---|
| author | Lee, LH Noble, JA |
| author_facet | Lee, LH Noble, JA |
| author_sort | Lee, LH |
| collection | OXFORD |
| description | Synthesis of anatomically realistic ultrasound images could be potentially valuable in sonographer training and to provide training images for algorithms, but is a challenging technical problem. Generating examples where different image attributes can be controlled may also be useful for tasks such as semi-supervised classification and regression to augment costly human annotation. In this paper, we propose using an information maximizing generative adversarial network with a least-squares loss function to generate new examples of fetal brain ultrasound images from clinically acquired healthy subject twenty-week anatomy scans. The unsupervised network succeeds in disentangling natural clinical variations in anatomical visibility and image acquisition parameters, which allows for user-control in image generation. To evaluate our method, we also introduce an additional synthetic fetal ultrasound specific image quality metric called the Fréchet SonoNet Distance (FSD) to quantitatively evaluate synthesis quality. To the best of our knowledge, this is the first work that generates ultrasound images with a generator network trained on clinical acquisitions where governing parameters can be controlled in a visually interpretable manner. |
| first_indexed | 2024-03-06T21:45:46Z |
| format | Conference item |
| id | oxford-uuid:49859279-9cca-4105-b9f3-1bd9bbfa92df |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-03-06T21:45:46Z |
| publishDate | 2020 |
| publisher | IEEE |
| record_format | dspace |
| spelling | oxford-uuid:49859279-9cca-4105-b9f3-1bd9bbfa92df2022-03-26T15:32:06ZGenerating controllable ultrasound images of the fetal headConference itemhttp://purl.org/coar/resource_type/c_5794uuid:49859279-9cca-4105-b9f3-1bd9bbfa92dfEnglishSymplectic ElementsIEEE2020Lee, LHNoble, JASynthesis of anatomically realistic ultrasound images could be potentially valuable in sonographer training and to provide training images for algorithms, but is a challenging technical problem. Generating examples where different image attributes can be controlled may also be useful for tasks such as semi-supervised classification and regression to augment costly human annotation. In this paper, we propose using an information maximizing generative adversarial network with a least-squares loss function to generate new examples of fetal brain ultrasound images from clinically acquired healthy subject twenty-week anatomy scans. The unsupervised network succeeds in disentangling natural clinical variations in anatomical visibility and image acquisition parameters, which allows for user-control in image generation. To evaluate our method, we also introduce an additional synthetic fetal ultrasound specific image quality metric called the Fréchet SonoNet Distance (FSD) to quantitatively evaluate synthesis quality. To the best of our knowledge, this is the first work that generates ultrasound images with a generator network trained on clinical acquisitions where governing parameters can be controlled in a visually interpretable manner. |
| spellingShingle | Lee, LH Noble, JA Generating controllable ultrasound images of the fetal head |
| title | Generating controllable ultrasound images of the fetal head |
| title_full | Generating controllable ultrasound images of the fetal head |
| title_fullStr | Generating controllable ultrasound images of the fetal head |
| title_full_unstemmed | Generating controllable ultrasound images of the fetal head |
| title_short | Generating controllable ultrasound images of the fetal head |
| title_sort | generating controllable ultrasound images of the fetal head |
| work_keys_str_mv | AT leelh generatingcontrollableultrasoundimagesofthefetalhead AT nobleja generatingcontrollableultrasoundimagesofthefetalhead |