WEKO3
-
RootNode
アイテム
Automated Prediction of Thoracic Vertebral Body Diameters from Computed Tomography Scans Using Deep Learning for Personal Identification in Mass Disasters
http://hdl.handle.net/10191/0002001625
http://hdl.handle.net/10191/0002001625cc7ddf73-e0c7-42cf-9143-92e60ecabe05
| 名前 / ファイル | ライセンス | アクション |
|---|---|---|
AA12680484-21-1-10-20.pdf (3.9 MB)
|
|
| Item type | 紀要論文 / Departmental Bulletin Paper(1) | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 公開日 | 2025-04-16 | |||||||||||||||
| タイトル | ||||||||||||||||
| タイトル | Automated Prediction of Thoracic Vertebral Body Diameters from Computed Tomography Scans Using Deep Learning for Personal Identification in Mass Disasters | |||||||||||||||
| 言語 | en | |||||||||||||||
| 言語 | ||||||||||||||||
| 言語 | eng | |||||||||||||||
| キーワード | ||||||||||||||||
| 言語 | en | |||||||||||||||
| 主題Scheme | Other | |||||||||||||||
| 主題 | Deep learning | |||||||||||||||
| キーワード | ||||||||||||||||
| 言語 | en | |||||||||||||||
| 主題Scheme | Other | |||||||||||||||
| 主題 | Thoracic vertebrae | |||||||||||||||
| キーワード | ||||||||||||||||
| 言語 | en | |||||||||||||||
| 主題Scheme | Other | |||||||||||||||
| 主題 | Computed tomography | |||||||||||||||
| キーワード | ||||||||||||||||
| 言語 | en | |||||||||||||||
| 主題Scheme | Other | |||||||||||||||
| 主題 | Personal identification | |||||||||||||||
| 資源タイプ | ||||||||||||||||
| 資源 | http://purl.org/coar/resource_type/c_6501 | |||||||||||||||
| タイプ | departmental bulletin paper | |||||||||||||||
| アクセス権 | ||||||||||||||||
| アクセス権 | open access | |||||||||||||||
| アクセス権URI | http://purl.org/coar/access_right/c_abf2 | |||||||||||||||
| 著者 |
Ichikawa, Shota
× Ichikawa, Shota
× Kondo, Yohan
× Okamoto, Masashi
× Kondo, Tatsuya
× Takahashi, Naoya
|
|||||||||||||||
| 抄録 | ||||||||||||||||
| 内容記述タイプ | Abstract | |||||||||||||||
| 内容記述 | The diameters of the thoracic vertebral bodies from antemortem and postmortem computed tomography (CT) images can be biological fi ngerprints for forensic personal identification. However, measuring these diameters manually is time-consuming. This study proposes a novel approach using deep learning to automatically predict vertebral morphological features from CT scans. Eighty-four CT scans from antemortem and corresponding postmortem patients were analyzed. The shortest diameter (in millimeters) of the depth, width, and height of the 1st through 12th thoracic vertebral bodies served as the ground truth. Our methodology involved two stages: first, U-Net-based models were employed to automatically segment thoracic vertebral bodies from CT images, achieving a Dice similarity coefficient (DSC) of 0.94 ± 0.01. Second, ResNet18-based regression models were employed to predict vertebral body diameters from volume-rendering (VR) images generated from the segmented vertebral bodies. The regression model showed strong correlations (ρ > 0.900, p < 0.001) for depth and width, with over 80% of the CT scans falling within the prediction difference range of ±10%. In contrast, a moderate correlation was observed in the prediction of height (ρ = 0.777, p < 0.001), with 70.4% of the CT scans being within the prediction difference range of ±10%. The proposed method offers a promising step toward automating personal identification based on vertebral features of CT in forensic radiology. | |||||||||||||||
| 言語 | en | |||||||||||||||
| bibliographic_information |
ja : 新潟大学保健学雑誌 en : Journal of Health of NiigataUniversity 巻 21, 号 1, p. 10-20, 発行日 2025-03 |
|||||||||||||||
| 出版者 | ||||||||||||||||
| 出版者 | 新潟大学医学部保健学科 | |||||||||||||||
| 言語 | ja | |||||||||||||||
| ISSN | ||||||||||||||||
| 収録物識別子タイプ | PISSN | |||||||||||||||
| 収録物識別子 | 2188-4617 | |||||||||||||||
| item_7_source_id_11 | ||||||||||||||||
| 収録物識別子タイプ | NCID | |||||||||||||||
| 収録物識別子 | AA12680484 | |||||||||||||||
| 出版タイプ | ||||||||||||||||
| 出版タイプ | VoR | |||||||||||||||
| 出版タイプResource | http://purl.org/coar/version/c_970fb48d4fbd8a85 | |||||||||||||||
