Gastrointestinal - Learning Modules - CTisus.com CT Scanning

Learning Modules ❯ Gastrointestinal

Pancreatic Neuroendocrine Tumors: CT Features to Help Predict Tumor Grade

Satomi Kawamoto, MD
Johns Hopkins Hospital

Disclosure

Dr. Kawamoto receives research support from The Lustgarten Foundation

Teaching Points

Pancreatic neuroendocrine tumors (PanNETs) are increasingly diagnosed in the recent decades
Treatment strategies for patients with a PanNETs have been evolved due to better understanding of tumor biology
Tumor stage and grade are critical for predicting prognosis and therefore for guiding treatment
Although difficult on imaging features alone, there are CT imaging characteristics that can help predict histological tumor grade.
Radiomics has emerged as a valuable and its potential and limitations in the application of predicting PanNET grade are reviewed and discussed.
These CT imaging features are presented and discussed with original pathology proven cases, and potential role of radiomics to predict PanNET grad are reviewed and discussed

Background

Incidence

Pancreatic neuroendocrine neoplasms (PanNENs) are 2-5% of clinically detected pancreatic tumors.
Driven by increasing incidental diagnosis of small (≤2 cm) tumors and early-stage disease.
6.4-fold rise in incidence of neuroendocrine neoplasms of all sites (including lung, GI tract, and pancreas) from 1973 to 2012, particularly rapid in PanNEN (Dasari et al.).

Classification

Heterogenous spectrum of behavior and presentation.
Divided into well-differentiated PanNETs and carcinomas (PanNECs) histologically.
50-85% of PanNENs are nonfunctional.
Functional PanNENs (e.g. insulinoma, gastrinoma, glucagonoma, VIPoma, and somatostatinoma): Up to 10% are syndromic (e.g. MEN1, VHL, NF-1). Diagnosed earlier (often < 2cm owing to symptoms). Surgery and/or medical management guided by symptomatology.

WHO Grading of PanNENs

PanNENs are classified based on their degree of differentiation that is determined by mitotic rate and Ki-67 labeling index
Grade 3 (G3) PanNET in 2010 WHO classification was divided into well-differentiated G3 PanNET and poorly differentiated pancreatic neuroendocrine carcinoma (PanNECs) in 2017 according to tumor morphology and Ki-65 index
Well-differentiated (WD) PanNETs:
- Grade 1 [G1]: Ki-67 labeling index of <3% and <2 mitoses/10 high-power field (HPF)
- Grade 2 [G2]: Ki-67 3-20% or 2-20 mitoses/10 HPF
- Grade 3 [G3]: Ki-67 >20% or >20 mitoses/10 HPF (Often only Ki-67 exceeds this threshold)
- Higher grade regions can occur within a single PanNET
Poorly differentiated PanNECs (classified based on cell morphology):
- Small cell NEC
- Large cell NEC

Challenges

NF-PanNENs pose a greater challenge, and accurate grading is essential for appropriate management
- The risk of tumor progression increases by 2% for each 1% increase in the Ki-67 index (Bian et al.)
- G2 PanNETs show a poorer prognosis, often require more intensive treatment than G1 PanNETs
- Surveillance options for small (≤2 cm in clinical practice), low-grade (G1) PanNETs
- Surgical resection and/or chemotherapy in higher-grade PanNETs
- However, tumors ≤2 cm can be higher (>G2) grade, and to distinguish between G1 and G2 PanNETs is still very difficult on imaging studies
EUS-FNA, the standard for preoperative grading, is invasive and often yields uncertain results due to technical limitations (e.g., requiring ≥500 cells for grading) and focal sampling, possibly missing higher-grade components
- Only 70% of patients obtain a histologic grade by EUS-FNA
- 60-90% discrepancy between EUS-FNA and surgical grading

Pancreatic Protocol CT

Multiphase CT including:
- Noncontrast phase (for calcifications and hemorrhage)
- Arterial (25–30 s after contrast injection) or pancreatic phase (40–45 s) particularly for small functioning PanNETs
- Portal venous phase (70–80 s)
Dual-source 64-multidetector (or more) CT scanners
120mL of nonionic contrast material injected intravenously through a peripheral venous line at an injection rate of 4-5 mL/sec for optimum enhancement
In our institution, typical CT protocol include arterial and venous phase acquisition, with scan parameters in order of 120kVp, reference 290 mAs, pitch 0.6, and collimation of 128×0.6 mm, 3 mm and 0.75 mm section thickness axial reconstructions with multiplanar reconstructions at 2-3 mm intervals

Imaging Features of Higher Grade PanNEN on CT

Although difficult on imaging features alone and there are controversial study results, reported CT imaging features of high-grade PanNENs include

Large size
Ill-defined tumor margin
Hypoenhancing/hypoattenuating tumor
Necrotic changes
Calcifications
Pancreatic ductal dilatation
Biliary ductal dilatation
Vascular involvement
Lymph node metastasis
Liver and other distant metastasis

Tumor Size

Most studies reported that there is statistical correlation between tumor size and degree of tumor aggressiveness.
G1 PanNETs are often <2 cm, and G2 and G3 PanNETs are often >2 cm (Bian et al., Takumi et al., Canellas et al.), however, there is significant overlap and small tumor does not preclude malignant behavior (Kuo et al.)
G3 PanNETs and PanNECs are generally larger (>3 cm) with poorly defined margins and may appear lobulated or irregular (Khanna et al., Kim et al.)
1.5-fold increase in probability of aggressive behavior (G3, nodal/distant metastasis, and/or disease recurrence) for every 1 cm increase in size (Shen et al.)

Well-defined vs. ill-defined tumor margin

qPoorly defined radiologic tumor border is more frequently associated with infiltrating macroscopic and microscopic growth pattern, and higher PanNET grade (Ahn et al.)
Other studies also reported higher grade tumors (G3/PanNECs) more likely demonstrate ill-defined borders compared to lower grade tumors (G1/G2) (Kim et al., Park et al.)

Vascularity & Enhancement

Many studies agreed that hyper-attenuating PanNETs compared to the pancreatic parenchyma tend to be low grade (G1), and iso- or hypoattenuating PanNETs particularly in portal venous phase are higher grade (G2 or higher) (Shen et al, Takumi et al.)
PanNETs show range from homogenous to heterogenous enhancement with enhancement patterns include:
- Hypovascularity in arterial phase
- Heterogeneous or rimlike enhancement in arterial phase
- Persistent enhancement in portal venous phase
- Hyperenhancement in delayed phase

Hypervascular PanNET

Arterial phase enhancement/homogeneity are a surrogate for vascularization, may indicate tumor differentiation.
Low-grade (G1) PanNETs are typically hypervascular with intense homogeneous enhancement on arterial phase and early wash out.

Hypoattenuating PanNET

As PanNETs become more aggressive, they lose angiogenic propensity with decreasing microvascularity.
Higher-grade tumors are iso/hypoattenuating on arterial phases and hypoattenuating on portal venous phases
Hypoenhancement is associated with ~2-fold reduction in post-resection survival even with other factors controlled (Worhunsky et al.)

Enhancement patterns

PanNETs that show early enhancement and rapid wash-out pattern is more commonly G1 (89% of this pattern was G1) (Cappeli et al.)
Persistent even enhancement pattern (early and venous/delayed phase enhancement) can be either (72% was G1, 28% G2/G3).
Late enhancement pattern in venous/delayed phase is more common in G2/3 (17% was G1, 83% G2/G3).
Both even and late enhancement patterns can have uncertain tumor behavior, including those of <2 cm (Cappeli et al.)

Enhancement Ratio (AER & PER)

Arterial enhancement ratio (AER): Tumor/Pancreas HU in arterial phase is lower in PanNEC compared to WD-PanNETs
- AER <0.7 is indicative of G3 and PanNECs (94.1% sensitivity, 78.9% specificity) (Park et al.)
Portal enhancement ratio (PER): Tumor/Pancreas HU in portal venous phase
- PER: ≤1.1 is indicative of G3 and PanNECs (92.3% sensitivity, 80.5% specificity) (Kim et al.)
- PER is an independent predictor for PanNECs vs. PanNETs (G1/G2/G3), with PER <0.8 showing highest sensitivity of 94.1%, specificity 88.5% (Park et al.)

Cystic and Necrotic Changes

Cystic and necrotic changes can appear similar on CT, but cystic PanNETs are often more well-defined, containing homogeneous fluid attenuation.
Pathologically, cystic PanNETs are typically filled with clear to straw-colored fluid (sometimes hemorrhagic), and lined by well-preserved NET cells. Necrosis is uncommonly associated with cystic PanNETs.
The majority of cystic PanNETs are grade 1 (Singhi et al.)
Cystic (>50% of total tumor volume) PanNETs are less commonly associated with lymph node metastases, and with a more favorable recurrence free survival after resection (Makris et al.)

Cystic and Necrotic Changes

Necrosis is typically seen as more ill-defined area of low attenuation, usually < 30 HU (Benedetti et al.)
Necrosis is uncommon in WD-PanNET, and are associated with higher-grade PanNETs and PanNECs (Khanna et al.)
Large PanNET may undergo necrosis with subsequent calcification (Khanna et al.)
However, some studies reported no correlation between cystic or necrotic changes and tumor grade, which may be due to difficulty in differentiating cystic vs. necrotic changes on imaging.

Calcifications

Studies reported calcifications are associated with higher-grade (G2 and G3) PanNETs and PanNECs, and poor long-term survival (Makris et al.)
However, some studies reported no significant correlation between presence of calcification and tumor grades (G1 vs. G2, or G1 vs. G2/G3, or G1/2 vs. G3/PanNEC) (Takumi et al., Canellas et al., Bian et al., Shen et al., Kim et al., Park et al.)

PanNET with Calcifications

Biliary duct obstruction

Dilatation of common bile duct (CBD) is more commonly associated with higher grade PanNETs (Shen et al., Kim et al., Park et al.)
CBD dilatation, vascular invasion, and peripancreatic tissue invasion are significantly more frequent in G3 than vs. G1/G2 PanNETs.

Pancreatic duct dilatation

Dilatation of the pancreatic duct (PD) is more commonly associated with higher grade PanNETs (Shen et al., Park et al.)
However, studies report conflicting results on the significance of ductal dilatation between G1 vs. G2 PanNETs (Takumi et al.) or G1/G2 vs. G3/PanNEC (Kim et al.)
For example, small G1 PanNET which produces serotonin may cause significant PD dilatation due to local fibrosis and stenosis of pancreatic duct.

Vascular and Local Invasion

Vascular invasion (vessel occlusion, stenosis, encasement, and tumor thrombus) is significantly more frequent in G2 than G1 PanNETs (Bien et al.), G1/G2 than G3 PanNETs/PanNECs (Kim et al.), and G1/G2/G3 PanNETs than PanNECs (Park et al.)

Vascular and Local Invasion

Vascular and local invasion are significantly more frequent in G3 than G1/G2 tumors.
However, some studies report conflicting results on the significance of organ invasion between G1 vs. G2 PanNETs (Takumi et al.)

Lymph nodes & Distant Metastases

Regional lymph node metastasis is more commonly associated with aggressive PanNET (Shen et al., Canellas et al.)
Most PanNETs with hepatic metastasis are G2 or higher grade (Takumi et al.)
However, some studies reported no significant difference of presence of lymph node metastasis in G1 vs. G2 tumors (Bian et al.)

Features associated with PanNEC

Features associated with PanNEC include vascular invasion, CBD and PD dilatation, and lymph node and hepatic metastases
Nearly 90% of PanNEC patients present with liver metastases on diagnosis (Basturk et al.)

Combining Imaging Features to Determine Tumor Grade

Using combination of aggressive imaging features increases accuracy to differentiate low grade vs. higher grade PanNETs.
- > 2cm tumor size, liver metastasis, iso- or hypoattenuating on portal venous phase: Accuracy to diagnose G1 from G2 PanNETs: 71%, 61%, 71% for each criteria, 82% with 3 combination (Takumi et al.)
- PD dilatation, CBD dilatation, vascular invasion, portal enhancement ratio <0.8m, and ill-defined borders: Combination of at least 3 of 5 features, sensitivity 88.2% and specificity 88.5% for diagnosing PanNEC from PanNET (G1/G2/G3) (Park et al.)
However, some tumors are difficult to predict tumor grade based on imaging features, in particular, differentiation of small (<2 cm) G1 and G2 PanNETs are challenging on conventional CT imaging features alone.

Pitfall: G1 PanNET with higher grade features

Pitfall: Higher grade PanNETs with low grade features

Can CT Radiomics Fill in the Gaps?

Radiomic texture analysis allows tissue characterization by converting biomedical images for quantitative analysis and extracting high-dimensional mineable features, and has been applied to PanNEN CT imaging to predict histologic grade, treatment response, and overall survival.
Notably, PanNENs demonstrate significant histopathologic heterogeneity, and limited representativeness of a presurgical FNA sample, CT radiomics may provide a more complete picture of tumor phenotype and microenvironment through noninvasive means for preoperative.
The addition of radiomic features to traditional cross-sectional imaging descriptors previously described has been shown to significantly improve model sensitivity, specificity, and interobserver reliability in such studies.

Radiomics: Feature Selection & Model Development

Radiomics encompasses a broad range of 2D and 3D region of interest (ROI) based features, including first-order statistical signal intensity features, shape and texture features derived from gray-level matrices, and wavelet-transformed features
In addition, machine learning-based (ML) features, particularly those from deep radiomics, involve automatically learned weights from convolutional neural networks (CNNs) instead of predefined mathematical equations in traditional radiomics
A high number of features and a low number of cases in a group for a classification task can result in overfitting of the model, thus a subset of features is selected
Machine learning techniques, like maximum relevance minimum redundancy (MRMR), LASSO, and random forest (RF) algorithms are typically employed to select the most correlated features with clinical targets, ensuring stable and robust markers

Typical Radiomics Workflow (Simplified)

Radiomics studies applied to PanNET Grading

Most studies dichotomize the classification and focus on differentiating G1/G2 from G3 or G1 from G2/G3, while some have tried to develop more nuanced models
Single-phase CT radiomic signatures can discriminate between tumor grades (G1 vs. G2/3), but combining signatures from different phases can better capture tumor vascularity, discriminating between shape-derived and contrast-imaging derived features (Benedetti et al. and Gu et al.)
Canellas et al. (2018) reported significant differences between G1 and G2/G3 PanNET in texture parameters (skewness, mean of positive pixels, and entropy), with entropy being an independent predictor
- Entropy showed significant differences in G1 vs. G3, and G2 vs. G3 (D'Onofrio et al.)
In the study by Guo et al. (2018), texture parameters grey-level intensity, entropy, and uniformity achieved specificity up to 100% in G1/G2 vs. G3 PanNET/PanNEC
- Mean grey-level intensity showed up to a 100% sensitivity and 91% specificity for distinguishing G1 vs. G2 PanNETs

Radiomics studies applied to PanNET Grading

Gu et al. (2019) combined tumor margin with a radiomics signature, achieving high sensitivity (86.7%), specificity (89.5%), and accuracy (88.2%) for G1 vs. G2/G3 tumors
Zhao et al. (2020) assessed patients with non-functional PanNETs and developed a signature using nonenhanced and portal venous phase CT features to achieve a sensitivity of 90.9% and specificity of 88.9% in identifying G2 tumors
Ye et al. (2024) studied biologic validation of a range of 2D, 3D and learning-based texture features and found that 3D wavelet-derived high-order texture features had the strongest correlations with clinical stage, pathological grade, and Ki-67 index
Javed et al. (2024) demonstrated that a radiomics model using both arterial and venous phase CT features achieved high sensitivity (87.5%) and specificity (73.3%) for differentiating G1 from G2/G3 tumors, particularly, the model outperformed EUS-FNA, which had significantly lower sensitivity (42.3%) in grading tumors, especially for small PanNETs

Challenges & Future Direction

Predictive models based on combined classic imaging features and radiomics features could guide the management of PanNETs, particularly in determining which patients with small tumors (<2 cm) could be safely monitored without invasive biopsies
Challenges of radiomics analysis include the need for standardized CT image preprocessing and the limited sample size for high-grade (G3) PanNETs in most studies
Moreover, differentiation between G1 vs. G2 PanNETs, and G3 PanNETs vs. PanNECs categories is crucial to determine management strategies due to their distinct clinical behaviors.
Radiomics could be also pivotal in monitoring tumor progression and treatment response in metastatic PanNETs, offering non-invasive assessment across multiple anatomical sites, including lymph node and liver metastases.

References

Ahn, Bokyung, Hyo Jung Park, Hyoung Jung Kim, and Seung-Mo Hong. “Radiologic Tumor Border Can Further Stratify Prognosis in Patients with Pancreatic Neuroendocrine Tumor.” Pancreatology 24, no. 5 (August 1, 2024): 753–63. https://doi.org/10.1016/j.pan.2024.05.524.
Basturk, Olca, Laura Tang, Ralph H. Hruban, Volkan Adsay, Zhaohai Yang, Alyssa M. Krasinskas, Efsevia Vakiani, et al. “Poorly Differentiated Neuroendocrine Carcinomas of the Pancreas: A Clinicopathologic Analysis of 44 Cases.” The American Journal of Surgical Pathology 38, no. 4 (April 2014): 437. https://doi.org/10.1097/PAS.0000000000000169.
Benedetti, Giulia, Martina Mori, Marta Maria Panzeri, Maurizio Barbera, Diego Palumbo, Carla Sini, Francesca Muffatti, et al. “CT-Derived Radiomic Features to Discriminate Histologic Characteristics of Pancreatic Neuroendocrine Tumors.” La Radiologia Medica 126, no. 6 (June 1, 2021): 745–60. https://doi.org/10.1007/s11547-021-01333-z.
Bian, Yun, Hui Jiang, Chao Ma, Li Wang, Jianming Zheng, Gang Jin, and Jianping Lu. “CT-Based Radiomics Score for Distinguishing Between Grade 1 and Grade 2 Nonfunctioning Pancreatic Neuroendocrine Tumors.” American Journal of Roentgenology 215, no. 4 (October 2020): 852–63. https://doi.org/10.2214/AJR.19.22123.
Canellas, Rodrigo, Kristine S. Burk, Anushri Parakh, and Dushyant V. Sahani. “Prediction of Pancreatic Neuroendocrine Tumor Grade Based on CT Features and Texture Analysis.” American Journal of Roentgenology 210, no. 2 (February 2018): 341–46. https://doi.org/10.2214/AJR.17.18417.
Cappelli, Carla, Ugo Boggi, Salvatore Mazzeo, Rosa Cervelli, Daniela Campani, Niccola Funel, Benedetta Pontillo Contillo, and Carlo Bartolozzi. “Contrast Enhancement Pattern on Multidetector CT Predicts Malignancy in Pancreatic Endocrine Tumours.” European Radiology 25, no. 3 (March 1, 2015): 751–59. https://doi.org/10.1007/s00330-014-3485-2.
Das, Satya, and Arvind Dasari. “Epidemiology, Incidence, and Prevalence of Neuroendocrine Neoplasms: Are There Global Differences?” Current Oncology Reports 23, no. 4 (March 14, 2021): 43. https://doi.org/10.1007/s11912-021-01029-7.
Dasari, Arvind, Chan Shen, Daniel Halperin, Bo Zhao, Shouhao Zhou, Ying Xu, Tina Shih, and James C. Yao. “Trends in the Incidence, Prevalence, and Survival Outcomes in Patients With Neuroendocrine Tumors in the United States.” JAMA Oncology 3, no. 10 (October 1, 2017): 1335–42. https://doi.org/10.1001/jamaoncol.2017.0589.
D’Onofrio, Mirko, Valentina Ciaravino, Nicolò Cardobi, Riccardo De Robertis, Sara Cingarlini, Luca Landoni, Paola Capelli, Claudio Bassi, and Aldo Scarpa. “CT Enhancement and 3D Texture Analysis of Pancreatic Neuroendocrine Neoplasms.” Scientific Reports 9 (February 18, 2019): 2176. https://doi.org/10.1038/s41598-018-38459-6.
Dromain, C., D. Déandréis, J. -Y. Scoazec, D. Goere, M. Ducreux, E. Baudin, and L. Tselikas. “Imaging of Neuroendocrine Tumors of the Pancreas.” Diagnostic and Interventional Imaging 97, no. 12 (December 1, 2016): 1241–57. https://doi.org/10.1016/j.diii.2016.07.012.
Gu, Dongsheng, Yabin Hu, Hui Ding, Jingwei Wei, Ke Chen, Hao Liu, Mengsu Zeng, and Jie Tian. “CT Radiomics May Predict the Grade of Pancreatic Neuroendocrine Tumors: A Multicenter Study.” European Radiology 29, no. 12 (December 1, 2019): 6880–90. https://doi.org/10.1007/s00330-019-06176-x.
Guilmette, Julie M., and Vania Nosé. “Neoplasms of the Neuroendocrine Pancreas: An Update in the Classification, Definition, and Molecular Genetic Advances.” Advances in Anatomic Pathology 26, no. 1 (January 2019): 13. https://doi.org/10.1097/PAP.0000000000000201.
Guo, Chuangen, Xiaoling Zhuge, Qidong Wang, Wenbo Xiao, Zhonglan Wang, Zhongqiu Wang, Zhan Feng, and Xiao Chen. “The Differentiation of Pancreatic Neuroendocrine Carcinoma from Pancreatic Ductal Adenocarcinoma: The Values of CT Imaging Features and Texture Analysis.” Cancer Imaging 18, no. 1 (October 17, 2018): 37. https://doi.org/10.1186/s40644-018-0170-8.
Javed, Ammar A., Zhuotun Zhu, Benedict Kinny-Köster, Joseph R. Habib, Satomi Kawamoto, Ralph H. Hruban, Elliot K. Fishman, Christopher L. Wolfgang, Jin He, and Linda C. Chu. “Accurate Non-Invasive Grading of Nonfunctional Pancreatic Neuroendocrine Tumors with a CT Derived Radiomics Signature.” Diagnostic and Interventional Imaging 105, no. 1 (January 1, 2024): 33–39. https://doi.org/10.1016/j.diii.2023.08.002.
Karmazanovsky, Grigory, Ivan Gruzdev, Valeriya Tikhonova, Evgeny Kondratyev, and Amiran Revishvili. “Computed Tomography-Based Radiomics Approach in Pancreatic Tumors Characterization.” La Radiologia Medica 126, no. 11 (November 1, 2021): 1388–95. https://doi.org/10.1007/s11547-021-01405-0.
Khanna, Lokesh, Srinivasa R. Prasad, Abhijit Sunnapwar, Sainath Kondapaneni, Anil Dasyam, Varaha S. Tammisetti, Umber Salman, Alia Nazarullah, and Venkata S. Katabathina. “Pancreatic Neuroendocrine Neoplasms: 2020 Update on Pathologic and Imaging Findings and Classification.” RadioGraphics 40, no. 5 (September 2020): 1240–62. https://doi.org/10.1148/rg.2020200025.
Kim, Dong Wook, Hyoung Jung Kim, Kyung Won Kim, Jae Ho Byun, Ki Byung Song, Ji Hoon Kim, and Seung-Mo Hong. “Neuroendocrine Neoplasms of the Pancreas at Dynamic Enhanced CT: Comparison between Grade 3 Neuroendocrine Carcinoma and Grade 1/2 Neuroendocrine Tumour.” European Radiology 25, no. 5 (May 1, 2015): 1375–83. https://doi.org/10.1007/s00330-014-3532-z.
Kuo, Eric J., and Ronald R. Salem. “Population-Level Analysis of Pancreatic Neuroendocrine Tumors 2 Cm or Less in Size.” Annals of Surgical Oncology 20, no. 9 (September 1, 2013): 2815–21. https://doi.org/10.1245/s10434-013-3005-7.
Liang, Wenjie, Pengfei Yang, Rui Huang, Lei Xu, Jiawei Wang, Weihai Liu, Lele Zhang, et al. “A Combined Nomogram Model to Preoperatively Predict Histologic Grade in Pancreatic Neuroendocrine Tumors.” Clinical Cancer Research 25, no. 2 (January 15, 2019): 584–94. https://doi.org/10.1158/1078-0432.CCR-18-1305.
Makris, Eleftherios A., John G. Douglas Cannon, Jeffrey A. Norton, Alexandra G. Lopez-Aguiar, Mary Dillhoff, Eliza Beal, Kamran Idrees, et al. “Calcifications and Cystic Morphology on Preoperative Imaging Predict Survival After Resection of Pancreatic Neuroendocrine Tumors.” Annals of Surgical Oncology 30, no. 4 (April 1, 2023): 2424–30. https://doi.org/10.1245/s10434-022-12783-8.
Park, Hyo Jung, Hyoung Jung Kim, Kyung Won Kim, So Yeon Kim, Sang Hyun Choi, Myung-Won You, Hee Sang Hwang, and Seung-Mo Hong. “Comparison between Neuroendocrine Carcinomas and Well-Differentiated Neuroendocrine Tumors of the Pancreas Using Dynamic Enhanced CT.” European Radiology 30, no. 9 (September 1, 2020): 4772–82. https://doi.org/10.1007/s00330-020-06867-w.
Pulvirenti, Alessandra, Rikiya Yamashita, Jayasree Chakraborty, Natally Horvat, Kenneth Seier, Caitlin A. McIntyre, Sharon A. Lawrence, et al. “Quantitative Computed Tomography Image Analysis to Predict Pancreatic Neuroendocrine Tumor Grade.” JCO Clinical Cancer Informatics 5 (June 17, 2021): CCI.20.00121. https://doi.org/10.1200/CCI.20.00121.
Segaran, Nicole, Catherine Devine, Mindy Wang, and Dhakshinamoorthy Ganeshan. “Current Update on Imaging for Pancreatic Neuroendocrine Neoplasms.” World Journal of Clinical Oncology 12, no. 10 (October 24, 2021): 897–911. https://doi.org/10.5306/wjco.v12.i10.897.
Shen, Xiaoding, Fan Yang, Taiyan Jiang, Zhenjiang Zheng, Yonghua Chen, Chunlu Tan, Nengwen Ke, et al. “A Nomogram to Preoperatively Predict the Aggressiveness of Non-Functional Pancreatic Neuroendocrine Tumors Based on CT Features.” European Journal of Radiology 171 (February 1, 2024): 111284. https://doi.org/10.1016/j.ejrad.2023.111284.
Singhi, Chu, Tatsas, Shi, Ellison, Fishman, Kawamoto, Schulick, Wolfgang, Hruban, Edil. Cystic pancreatic neuroendocrine tumors: a clinicopathologic study. Am J Surg Path 36 (11), 1666-1673 https://doi.org/10.1097/PAS.0b013e31826a0048.
Takumi, Koji, Yoshihiko Fukukura, Michiyo Higashi, Junnichi Ideue, Tomokazu Umanodan, Hiroto Hakamada, Ichiro Kanetsuki, and Takashi Yoshiura. “Pancreatic Neuroendocrine Tumors: Correlation between the Contrast-Enhanced Computed Tomography Features and the Pathological Tumor Grade.” European Journal of Radiology 84, no. 8 (August 2015): 1436–43. https://doi.org/10.1016/j.ejrad.2015.05.005.
Xu, Lei, Yan Wu, Xiaoyong Shen, Luping Zhou, Yongkai Lu, Ze Teng, Jichen Du, Mingchao Ding, Hongbin Han, and Tianye Niu. “Exploring the Biological Basis of CT Imaging Features in Pancreatic Neuroendocrine Tumors: A Two-Center Study.” Physics in Medicine & Biology 69, no. 12 (June 2024): 125013. https://doi.org/10.1088/1361-6560/ad51c7.
Ye, Jing-Yuan, Peng Fang, Zhen-Peng Peng, Xi-Tai Huang, Jin-Zhao Xie, and Xiao-Yu Yin. “A Radiomics-Based Interpretable Model to Predict the Pathological Grade of Pancreatic Neuroendocrine Tumors.” European Radiology 34, no. 3 (2024): 1994–2005. https://doi.org/10.1007/s00330-023-10186-1.
Zamboni, Giulia A., Maria Chiara Ambrosetti, Caterina Zivelonghi, Fabio Lombardo, Giovanni Butturini, Sara Cingarlini, Paola Capelli, and Roberto Pozzi Mucelli. “Solid Non-Functioning Endocrine Tumors of the Pancreas: Correlating Computed Tomography and Pathology.” HPB: The Official Journal of the International HepatoPancreatoBiliary Association 19, no. 11 (November 2017): 986–91. https://doi.org/10.1016/j.hpb.2017.06.013.
Zhao, Zengrui, Yun Bian, Hui Jiang, Xu Fang, Jing Li, Kai Cao, Chao Ma, et al. “CT-Radiomic Approach to Predict G1/2 Nonfunctional Pancreatic Neuroendocrine Tumor.” Academic Radiology 27, no. 12 (December 2020): e272–81. https://doi.org/10.1016/j.acra.2020.01.002.