Automatic detection of punctate white matter lesions in infants using deep learning of composite images from two cases
Abstract Punctate white matter lesions (PWMLs) in infants may be related to neurodevelopmental outcomes based on the location or number of lesions. This study aimed to assess the automatic detectability of PWMLs in infants on deep learning using composite images created from several cases. To create...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2023-03-01
|
| Series: | Scientific Reports |
| Online Access: | https://doi.org/10.1038/s41598-023-31403-3 |
| _version_ | 1827984537900023808 |
|---|---|
| author | Xuyang Sun Tetsu Niwa Takashi Okazaki Sadanori Kameda Shuhei Shibukawa Tomohiko Horie Toshiki Kazama Atsushi Uchiyama Jun Hashimoto |
| author_facet | Xuyang Sun Tetsu Niwa Takashi Okazaki Sadanori Kameda Shuhei Shibukawa Tomohiko Horie Toshiki Kazama Atsushi Uchiyama Jun Hashimoto |
| author_sort | Xuyang Sun |
| collection | DOAJ |
| description | Abstract Punctate white matter lesions (PWMLs) in infants may be related to neurodevelopmental outcomes based on the location or number of lesions. This study aimed to assess the automatic detectability of PWMLs in infants on deep learning using composite images created from several cases. To create the initial composite images, magnetic resonance (MR) images of two infants with the most PWMLs were used; their PWMLs were extracted and pasted onto MR images of infants without abnormality, creating many composite PWML images. Deep learning models based on a convolutional neural network, You Only Look Once v3 (YOLOv3), were constructed using the training set of 600, 1200, 2400, and 3600 composite images. As a result, a threshold of detection probability of 20% and 30% for all deep learning model sets yielded a relatively high sensitivity for automatic PWML detection (0.908–0.957). Although relatively high false-positive detections occurred with the lower threshold of detection probability, primarily, in the partial volume of the cerebral cortex (≥ 85.8%), those can be easily distinguished from the white matter lesions. Relatively highly sensitive automatic detection of PWMLs was achieved by creating composite images from two cases using deep learning. |
| first_indexed | 2024-04-09T23:01:22Z |
| format | Article |
| id | doaj.art-140d5b328cbe49f5b9cb665cf1193abd |
| institution | Directory Open Access Journal |
| issn | 2045-2322 |
| language | English |
| last_indexed | 2024-04-09T23:01:22Z |
| publishDate | 2023-03-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj.art-140d5b328cbe49f5b9cb665cf1193abd2023-03-22T10:56:48ZengNature PortfolioScientific Reports2045-23222023-03-011311810.1038/s41598-023-31403-3Automatic detection of punctate white matter lesions in infants using deep learning of composite images from two casesXuyang Sun0Tetsu Niwa1Takashi Okazaki2Sadanori Kameda3Shuhei Shibukawa4Tomohiko Horie5Toshiki Kazama6Atsushi Uchiyama7Jun Hashimoto8Department of Radiology, Tokai University School of MedicineDepartment of Radiology, Tokai University School of MedicineDepartment of Radiology, Tokai University School of MedicineDepartment of Radiology, Tokai University School of MedicineDepartment of Radiology, Tokai University School of MedicineDepartment of Radiology, Tokai University HospitalDepartment of Radiology, Tokai University School of MedicineDepartment of Pediatrics, Tokai University School of MedicineDepartment of Radiology, Tokai University School of MedicineAbstract Punctate white matter lesions (PWMLs) in infants may be related to neurodevelopmental outcomes based on the location or number of lesions. This study aimed to assess the automatic detectability of PWMLs in infants on deep learning using composite images created from several cases. To create the initial composite images, magnetic resonance (MR) images of two infants with the most PWMLs were used; their PWMLs were extracted and pasted onto MR images of infants without abnormality, creating many composite PWML images. Deep learning models based on a convolutional neural network, You Only Look Once v3 (YOLOv3), were constructed using the training set of 600, 1200, 2400, and 3600 composite images. As a result, a threshold of detection probability of 20% and 30% for all deep learning model sets yielded a relatively high sensitivity for automatic PWML detection (0.908–0.957). Although relatively high false-positive detections occurred with the lower threshold of detection probability, primarily, in the partial volume of the cerebral cortex (≥ 85.8%), those can be easily distinguished from the white matter lesions. Relatively highly sensitive automatic detection of PWMLs was achieved by creating composite images from two cases using deep learning.https://doi.org/10.1038/s41598-023-31403-3 |
| spellingShingle | Xuyang Sun Tetsu Niwa Takashi Okazaki Sadanori Kameda Shuhei Shibukawa Tomohiko Horie Toshiki Kazama Atsushi Uchiyama Jun Hashimoto Automatic detection of punctate white matter lesions in infants using deep learning of composite images from two cases Scientific Reports |
| title | Automatic detection of punctate white matter lesions in infants using deep learning of composite images from two cases |
| title_full | Automatic detection of punctate white matter lesions in infants using deep learning of composite images from two cases |
| title_fullStr | Automatic detection of punctate white matter lesions in infants using deep learning of composite images from two cases |
| title_full_unstemmed | Automatic detection of punctate white matter lesions in infants using deep learning of composite images from two cases |
| title_short | Automatic detection of punctate white matter lesions in infants using deep learning of composite images from two cases |
| title_sort | automatic detection of punctate white matter lesions in infants using deep learning of composite images from two cases |
| url | https://doi.org/10.1038/s41598-023-31403-3 |
| work_keys_str_mv | AT xuyangsun automaticdetectionofpunctatewhitematterlesionsininfantsusingdeeplearningofcompositeimagesfromtwocases AT tetsuniwa automaticdetectionofpunctatewhitematterlesionsininfantsusingdeeplearningofcompositeimagesfromtwocases AT takashiokazaki automaticdetectionofpunctatewhitematterlesionsininfantsusingdeeplearningofcompositeimagesfromtwocases AT sadanorikameda automaticdetectionofpunctatewhitematterlesionsininfantsusingdeeplearningofcompositeimagesfromtwocases AT shuheishibukawa automaticdetectionofpunctatewhitematterlesionsininfantsusingdeeplearningofcompositeimagesfromtwocases AT tomohikohorie automaticdetectionofpunctatewhitematterlesionsininfantsusingdeeplearningofcompositeimagesfromtwocases AT toshikikazama automaticdetectionofpunctatewhitematterlesionsininfantsusingdeeplearningofcompositeimagesfromtwocases AT atsushiuchiyama automaticdetectionofpunctatewhitematterlesionsininfantsusingdeeplearningofcompositeimagesfromtwocases AT junhashimoto automaticdetectionofpunctatewhitematterlesionsininfantsusingdeeplearningofcompositeimagesfromtwocases |