Presentation | 2021-07-08 [Short Paper] Performance comparison of multiple deep CNN methods for multiple organ detection in CT images Daiki Kanoh, Xiangrong Zhou, Takeshi Hara, Hiroshi Fujita, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | The scheme of automatically recognizing multiple organs and detecting their localizations in 3D CT images is required for computer-aided diagnosis systems to support doctors' diagnosis. In this study, we compared the performance of our proposed method with three conventional object detection methods for recognition and detection of multiple organs in 3D CT images based on 2D deep CNNs. The proposed method is an improved version of Single Shot MultiBox Detector (SSD). We compared the performance of the proposed method to three conventional object detection methods: SSD and YOLOv3, which are widely used in the field of natural images based on 2D CNNs, and Detection Transformer (DETR), which uses a transformer and has attracted much attention recently. We applied those detection methods to the automatic recognition and detection of 17 organ types in 240 CT cases from a shared database of the research projector “computational anatomy”, in which contrast and non-contrast CT images are mixed. The experimental results demonstrate the effectiveness and challenges of the proposed method for organ detection in 3D CT images. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | 3D CT Image / Detection / SSD / YOLOv3 / Transformer |
Paper # | MI2021-10 |
Date of Issue | 2021-07-01 (MI) |
Conference Information | |
Committee | MI |
---|---|
Conference Date | 2021/7/8(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Online |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | Medical imaging, physics, and recognition |
Chair | Hidekata Hontani(Nagoya Inst. of Tech.) |
Vice Chair | Hideaki Haneishi(Chiba Univ.) / Takayuki Kitasaka(Aichi Inst. of Tech.) |
Secretary | Hideaki Haneishi(Yamaguchi Univ.) / Takayuki Kitasaka(Univ. of Hyogo) |
Assistant | Hotaka Takizawa(Tsukuba Univ.) / Yoshito Otake(NAIST) |
Paper Information | |
Registration To | Technical Committee on Medical Imaging |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | [Short Paper] Performance comparison of multiple deep CNN methods for multiple organ detection in CT images |
Sub Title (in English) | |
Keyword(1) | 3D CT Image |
Keyword(2) | Detection |
Keyword(3) | SSD |
Keyword(4) | YOLOv3 |
Keyword(5) | Transformer |
1st Author's Name | Daiki Kanoh |
1st Author's Affiliation | Gifu University(Gifu Univ.) |
2nd Author's Name | Xiangrong Zhou |
2nd Author's Affiliation | Gifu University(Gifu Univ.) |
3rd Author's Name | Takeshi Hara |
3rd Author's Affiliation | Gifu University(Gifu Univ.) |
4th Author's Name | Hiroshi Fujita |
4th Author's Affiliation | Gifu University(Gifu Univ.) |
Date | 2021-07-08 |
Paper # | MI2021-10 |
Volume (vol) | vol.121 |
Number (no) | MI-98 |
Page | pp.pp.7-10(MI), |
#Pages | 4 |
Date of Issue | 2021-07-01 (MI) |