2019³â ÄÄÇ»ÅÍÁ¾ÇÕÇмú´ëȸ
ÇѱÛÁ¦¸ñ(Korean Title) |
ÆóÀÇ FDG-PET µö·¯´× ÇнÀÀ» ÅëÇÑ ÀÚµ¿ Á¾¾ç°ËÃâ ¹× ºÐÇÒ |
¿µ¹®Á¦¸ñ(English Title) |
Automatic Tumor Detection and Segmentation of FDG-PET in Lung using Deep Neural Networks |
ÀúÀÚ(Author) |
Duc-Ky Ngo
À̱ͻó
±è¼öÇü
¾çÇüÁ¤
Nguyen Ngoc Hoang
Guee-Sang Lee
Soo-Hyung Kim
Hyung-Jeong Yang
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 46 NO. 01 PP. 0796 ~ 0798 (2019. 06) |
Çѱ۳»¿ë (Korean Abstract) |
|
¿µ¹®³»¿ë (English Abstract) |
Lung cancer is one of the most leading causes of cancer-related death worldwide. Nowadays, preoperative imaging is recommended for evaluating the malignancy grade of lung tumors before treatments. Positron emission tomography (PET) plays a crucially important role in modern cancer diagnosis and therapy because they provide much critical diagnostic information for therapy. In this paper, we build two models for the purpose of automatically detect and segment the tumor in the PET image. The aim of the proposed framework is providing more accurate of tumor detection and measurement of tumors size and the further structures. The FDG-PET tumor detection and segmentation relying on a U-Net Convolutional Neural Network (CNN). We evaluate our model on 11 clinical PET scans from patients with non-small cell lung cancer (NSLS). The accuracy of the segmentation output obtained by our approach was quantified using the metrics of dice similarity coefficient (DSC). |
Å°¿öµå(Keyword) |
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|