Segmentation of the Surrounding Arteries and Pancreatic Ductal Adenocarcinoma (PDAC) Utilizing A Deep Learning-Based Cascade Method for Pancreatic Tumour Segmentation
Keywords:
Artificial intelligence, Endoscopic ultrasound, microbubble, contrast-enhanced and pancreatic cancer.Abstract
This study proposes a convolutional neural network (CNN)-based approach to segment PDAC mass and nearby arteries in CT images by combining powerful conventional characteristics. TAU-Net is demonstrated by incorporating dense Scale-Invariant Feature Transform (SIFT) & LBP descriptors into the attention U-Net. A 3D-CNN & an ensemble model is then employed to combine the advantages of both networks. Since the sample size was insufficient for vascular segmentation, the previously stated pre trained networks were used and improved.
References
Bipat S., Phoa S.S.K.S., van Delden O.M., Bossuyt P.M.M., Gouma D.J., Lameris J.S., Stoker J. Ultrasonography, computed tomography and magnetic resonance imaging for diagnosis and determining resectability of pancreatic adenocarcinoma: A meta-analysis. J. Comput. Assist. Tomogr. 2005;29:438–445. Abraham S.C., Wilentz R.E., Yeo C.J., Sohn T.A., Cameron J.L., Boitnott J.K., Hruban R.H.
Pancreaticoduodenectomy (Whipple resections) in patients without malignancy: Are they all ‘chronic pancreatitis’? Am. J. Surg. Pathol. 2003;27:110–120. doi: 10.1097/00000478-200301000-00012.
