“…The major limitation of the model is the assumption about the presence of the whole pancreas. The model might get deviated when the pancreas has a history of surgical interventions, such as a Pancreatectomy [ 27 ] or Whipple procedure [ 28 ] (partial removal of the pancreas). This would also include cases when the size of the subregions varies due to underlying disorders, such as pancreatic inflammation.…”
The accurate segmentation of pancreatic subregions (head, body, and tail) in CT images provides an opportunity to examine the local morphological and textural changes in the pancreas. Quantifying such changes aids in understanding the spatial heterogeneity of the pancreas and assists in the diagnosis and treatment planning of pancreatic cancer. Manual outlining of pancreatic subregions is tedious, time-consuming, and prone to subjective inconsistency. This paper presents a multistage anatomy-guided framework for accurate and automatic 3D segmentation of pancreatic subregions in CT images. Using the delineated pancreas, two soft-label maps were estimated for subregional segmentation—one by training a fully supervised naïve Bayes model that considers the length and volumetric proportions of each subregional structure based on their anatomical arrangement, and the other by using the conventional deep learning U-Net architecture for 3D segmentation. The U-Net model then estimates the joint probability of the two maps and performs optimal segmentation of subregions. Model performance was assessed using three datasets of contrast-enhanced abdominal CT scans: one public NIH dataset of the healthy pancreas, and two datasets D1 and D2 (one for each of pre-cancerous and cancerous pancreas). The model demonstrated excellent performance during the multifold cross-validation using the NIH dataset, and external validation using D1 and D2. To the best of our knowledge, this is the first automated model for the segmentation of pancreatic subregions in CT images. A dataset consisting of reference anatomical labels for subregions in all images of the NIH dataset is also established.
“…The major limitation of the model is the assumption about the presence of the whole pancreas. The model might get deviated when the pancreas has a history of surgical interventions, such as a Pancreatectomy [ 27 ] or Whipple procedure [ 28 ] (partial removal of the pancreas). This would also include cases when the size of the subregions varies due to underlying disorders, such as pancreatic inflammation.…”
The accurate segmentation of pancreatic subregions (head, body, and tail) in CT images provides an opportunity to examine the local morphological and textural changes in the pancreas. Quantifying such changes aids in understanding the spatial heterogeneity of the pancreas and assists in the diagnosis and treatment planning of pancreatic cancer. Manual outlining of pancreatic subregions is tedious, time-consuming, and prone to subjective inconsistency. This paper presents a multistage anatomy-guided framework for accurate and automatic 3D segmentation of pancreatic subregions in CT images. Using the delineated pancreas, two soft-label maps were estimated for subregional segmentation—one by training a fully supervised naïve Bayes model that considers the length and volumetric proportions of each subregional structure based on their anatomical arrangement, and the other by using the conventional deep learning U-Net architecture for 3D segmentation. The U-Net model then estimates the joint probability of the two maps and performs optimal segmentation of subregions. Model performance was assessed using three datasets of contrast-enhanced abdominal CT scans: one public NIH dataset of the healthy pancreas, and two datasets D1 and D2 (one for each of pre-cancerous and cancerous pancreas). The model demonstrated excellent performance during the multifold cross-validation using the NIH dataset, and external validation using D1 and D2. To the best of our knowledge, this is the first automated model for the segmentation of pancreatic subregions in CT images. A dataset consisting of reference anatomical labels for subregions in all images of the NIH dataset is also established.
“…Pancreaticoduodenectomy is a complex procedure that involves resection of the gallbladder, bile duct, first part of the duodenum, and head of the pancreas [1][2][3][4][5]. It demands a meticulous surgical technique due to complex anastomosis (hepaticojejunostomy, gastrojejunostomy, and pancreaticojejunostomy).…”
Section: Discussionmentioning
confidence: 99%
“…It demands a meticulous surgical technique due to complex anastomosis (hepaticojejunostomy, gastrojejunostomy, and pancreaticojejunostomy). This procedure is associated with high postoperative morbidity and mortality, especially in developing countries, and therefore demands invasive postoperative management [3]. Nonetheless, the procedure remains the treatment of choice for curative management of non-complicated and non-metastatic ampullary or periampullary neoplasms [7,10].…”
Section: Discussionmentioning
confidence: 99%
“…It is commonly indicated for a myriad of biliary, ampullary, pancreatic, and periampullary neoplasms [2]. Periampullary carcinomas have a dismal prognosis and are ranked as the fourth leading cause of cancer-related mortalities worldwide [3]. The risk factors associated with the incidence of pancreatic head cancers, ampullary and periampullary carcinomas increased due to age, ethnicity, use of tobacco, alcoholism, chronic pancreatitis family history, familial adenomatous polyposis syndrome (FAP), and genetic susceptibility [2].…”
“…Pancreaticoduodenectomy (PD), also known as the Whipple procedure, proposed by Whipple et al in 1935, is the reference treatment modality for resectable pancreatic cancer [ 3 ]. The resection extent of PD covers the duodenum, the proximal 15 cm of the jejunum, the common bile duct, gall bladder, head of the pancreas, and a distal gastrectomy [ 4 ]. According to the literature, the incidence of postoperative complications of PD is 27.1% or even higher [ 5 ].…”
To assess the role of protein-energy malnutrition on perioperative outcomes in patients with pancreatic cancer undergoing open pancreaticoduodenectomy. We conducted a retrospective observational cohort study and investigated patients ≥ 18 years old with pancreatic cancer undergoing open pancreaticoduodenectomy within the National inpatient sample database during 2012–2014. The study population was divided into two groups based on the presence of protein-energy malnutrition. In-hospital mortality, length of stay, cost of hospitalization, and in-hospital complications were compared between the two groups. Logistic and linear regression analyses were used to adjust for potential confounders. A trend analysis was further conducted on the in-hospital outcomes. Of the 12,785 patients aged ≥ 18 years undergoing open pancreaticoduodenectomy during years 2012–2014, 9865 (77.0%) had no protein-energy malnutrition and 2920 (23.0%) had protein-energy malnutrition. Patients with protein-energy malnutrition were found to have significantly higher mortality rate, longer length of hospital stay, and higher total hospital cost compared to those without protein-energy malnutrition. The risks of gastroparesis, small bowel obstruction, intraoperative and postoperative hemorrhage, infectious complications, and several systemic complications were found to be significantly higher in the protein-energy malnutrition group in a multivariate regression model. A study of trends from 2009 to 2012 revealed an increasing prevalence of protein-energy malnutrition, a declining trend in mortality and length of stay and a stable total hospital cost in the protein-energy malnutrition group. Protein-energy malnutrition was found to be associated with higher mortality, longer length of hospital stay and greater hospital cost in pancreatic cancer patients undergoing open pancreaticoduodenectomy, as well as increased occurrence of various systemic complications. Attention should be paid to patients’ nutritional status, which can be corrected before surgery as an effective means to optimize postoperative results.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.