講演抄録/キーワード |
講演名 |
2021-02-18 13:50
FDG-PET/CT画像を用いたattention mechanismに基づく乳癌の腋窩リンパ節転移の検出 ○李 宗曜・藤後 廉・平田健司(北大)・北島一宏(兵庫医科大)・竹中淳規(北大)・三好康雄(兵庫医科大)・工藤與亮・小川貴弘・長谷山美紀(北大) |
抄録 |
(和) |
Determination of axillary nodal status is significant to treatment of breast cancer. Typically, the diagnosis of axillary lymph node (LN) metastasis is performed by using invasive methods which impose considerable burden on patients. On the other hand, noninvasive FDG-PET/CT can be also used for diagnosing axillary LN metastasis but has inferior performance. In this paper, we focus on detecting axillary LN metastasis of breast cancer with FDG-PET/CT images by using convolutional neural networks (CNNs). Specifically, we equip a 3D residual CNN with an attention mechanism. The attention mechanism can make the network focus on the most meaningful regions related to the diagnosis. As a result, the diagnostic performance is considerably improved by the attention mechanism compared to a simple CNN. |
(英) |
Determination of axillary nodal status is significant to treatment of breast cancer. Typically, the diagnosis of axillary lymph node (LN) metastasis is performed by using invasive methods which impose considerable burden on patients. On the other hand, noninvasive FDG-PET/CT can be also used for diagnosing axillary LN metastasis but has inferior performance. In this paper, we focus on detecting axillary LN metastasis of breast cancer with FDG-PET/CT images by using convolutional neural networks (CNNs). Specifically, we equip a 3D residual CNN with an attention mechanism. The attention mechanism can make the network focus on the most meaningful regions related to the diagnosis. As a result, the diagnostic performance is considerably improved by the attention mechanism compared to a simple CNN. |
キーワード |
(和) |
breast cancer / axillary lymph node / FDG-PET/CT / convolutional neural network / / / / |
(英) |
breast cancer / axillary lymph node / FDG-PET/CT / convolutional neural network / / / / |
文献情報 |
映情学技報, vol. 45, no. 4, ME2021-7, pp. 33-36, 2021年2月. |
資料番号 |
ME2021-7 |
発行日 |
2021-02-11 (MMS, ME, AIT) |
ISSN |
Print edition: ISSN 1342-6893 Online edition: ISSN 2424-1970 |
PDFダウンロード |
|