RPS 1801 - Pancreas multi-modality imaging innovation

Author Block: C. Booz, A-I. Nica, I. Yel, V. Koch, L. D. Grünewald, P. Reschke, J. Gotta, K. Eichler, T. J. Vogl; Frankfurt/DE
Purpose: Non-invasive imaging biomarkers derived from routine CT may assist in the early identification of individuals at risk for diabetes mellitus (DM). While both decreased pancreatic attenuation and a reduced pancreas-to-spleen attenuation ratio (PSR) have been associated with dysglycemia, it remains unclear whether a combination of both parameters yields additional diagnostic benefit. This study aimed to evaluate the added value of combining pancreatic attenuation and PSR for risk stratification of DM.
Methods or Background: We retrospectively analyzed non-contrast abdominal CT scans from 5624 patients (51% male; mean age 52 ± 10 years). Patients were categorized into normoglycemic (n=2900), prediabetic (n=1569), and diabetic (n=1155) groups based on clinical and laboratory data. Mean pancreatic attenuation (HU) was measured in head, body, and tail regions and compared to splenic attenuation to calculate PSR. ROC curve analyses and multivariate logistic regression were used to evaluate diagnostic performance for identifying diabetic status using individual and combined CT parameters.
Results or Findings: Pancreatic attenuation decreased progressively from normoglycemic (43 HU) to prediabetic (36 HU) to diabetic patients (32 HU) (all p < 0.001). PSR showed stronger correlation with HbA1c (r = –0.90) than pancreatic HU alone (r = –0.76). The AUC for PSR alone was 0.88, outperforming pancreatic HU (AUC: 0.79). However, combining both parameters slightly improved diagnostic accuracy (AUC: 0.92; p = 0.01 vs. PSR alone). In multivariate models, the combined approach remained an independent predictor of diabetes, offering both improved sensitivity and specificity.
Conclusion: Although PSR is a superior single CT-derived biomarker for assessing diabetes risk, combination of PSR and pancreatic attenuation provides a small but statistically significant improvement in diagnostic performance. This supports usage of combined attenuation-based metrics for more precise opportunistic screening in routine CT.
Limitations: Single-center study; retrospective study design
Funding for this study: No funding
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Ethics committee approval number 2023-1216

Author Block: N. Missima, H. Hill, C-S. Aioanei, P. Liss, D. Espes; Uppsala/SE
Purpose: In type 1 diabetes (T1D), the loss of insulin-producing beta-cells is the hallmark pathophysiological alteration.
However, volumetric and functional abnormalities of the exocrine pancreas are also observed. These changes may result
from the loss of insulin’s anabolic effects or reflect an underexplored aspect of T1D. Imaging techniques have enabled a
better characterization of pancreatic morphology throughout T1D progression. This study examines exocrine pancreatic
alterations at various stages of T1D using CT scans, including assessments conducted prior to diagnosis.
Methods or Background: The study utilized retrospective abdominal CT scans and clinical data collected from Uppsala University Hospital,
including 150 T1D subjects, with 15 examined before diagnosis, and 61 age- and gender-matched non-diabetic controls.
Volume segmentation and 3D reconstruction assessed the exocrine pancreas, and pancreas volume index (PVI) calculations
were standardized using body weight, BMI, and body surface area (BSA). Descriptive and laboratory data were obtained
from electronic medical records.
Results or Findings: Pancreas volume was significantly reduced in T1D patients. The reduction was more pronounced in patients
diagnosed before the age of 20. No significant volume difference was noted in patients before their T1D diagnosis compared
to controls, however, a reduction was observed post-diagnosis. Pancreas volume correlated negatively with disease duration
and HbA1c levels and correlated positively with body surface area and plasma amylase levels.
Conclusion: Pancreas volume reduction is a consistent feature in T1D, correlating with both disease duration and markers of
metabolic control. These findings support the potential of using imaging techniques as a non-invasive method for monitoring
T1D progression.
Limitations: Lack of standardization, absence of relevant laboratory measurements (e.g. amylase, C-peptide) in the original clinical evaluations. Variability in imaging protocols. Many T1D patients and control subjects with native CT scans were excluded due to segmentation challenges.
Funding for this study: Research grants from SciLife Lab, Barndiabetesfonden, Diabetesfonden, Familjen Ernfors fond, EXODIAB and Göran Gustafssons Stiftelse.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The presented study is based on retrospective data collected at Uppsala University Hospital, Sweden. The study has been
approved by the Swedish Ethical Review Authority (Dnr 2018/391).

Author Block: S. Kaya, H. Atasoy, M. A. Arıkan, A. H. H. Yardımcı, M. Karagülle, C. C. Arıkan; Istanbul/TR
Purpose: To evaluate the relationship between abdominal subcutaneous adipose tissue thickness and pancreatic apparent diffusion coefficient (ADC) values in healthy individuals, using diffusion-weighted imaging (DWI).
Methods or Background: A total of 100 abdominal MRI scans (69 females, 31 males; mean age 42.8 ± 12.0 years) performed on a 1.5T scanner were retrospectively reviewed. ADC values were measured in the head-neck, body, and tail of the pancreas with free-hand regions of interest encompassing the entire visible segment. Subcutaneous adipose tissue thickness was measured at the umbilical level. Two radiologists independently performed measurements, with interobserver agreement assessed by intraclass correlation coefficients (ICC). Correlations between ADC values, adipose thickness, age, and gender were analyzed using Pearson correlation and Mann–Whitney U tests.
Results or Findings: No significant differences in ADC values were observed among pancreatic segments or between genders, and no correlation with age was found (p > 0.05 for all). Subcutaneous adipose tissue thickness showed a significant inverse correlation with ADC values across all pancreatic segments (head-neck r = –0.296, p = 0.003; body r = –0.302, p = 0.002; tail r = –0.310, p = 0.002). Interobserver agreement was excellent (ICC range: 0.810–0.862).
Conclusion: Pancreatic ADC values appear stable across age and gender but decrease significantly with increasing subcutaneous adipose tissue thickness. This novel finding identifies adipose thickness as a potential confounding factor in pancreatic DWI interpretation and may improve the reliability of pancreatic imaging in clinical and research settings.
Limitations: This study has no limitations.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: C. Booz, A-I. Nica, I. Yel, V. Koch, L. D. Grünewald, P. Reschke, J. Gotta, K. Eichler, T. J. Vogl; Frankfurt/DE
Purpose: Non-invasive imaging biomarkers may aid in early detection of individuals at risk for diabetes mellitus (DM). While decreased pancreatic attenuation on non-contrast CT has been associated with metabolic disorders, recent evidence suggests that the pancreas-to-spleen attenuation ratio (PSR) may offer improved diagnostic performance by minimizing inter-individual and technical variability. This study aimed to compare the diagnostic value of pancreatic attenuation versus PSR in assessing diabetes risk.
Methods or Background: In this retrospective study, non-contrast abdominal CT scans from 5634 patients (52% male, mean age 53 ± 8 years) were analyzed. Patients were categorized into normoglycemic (n=2902), prediabetic (n=1572), and diabetic groups (n=1160) based on clinical and laboratory data. Mean pancreatic attenuation (HU) was measured at three anatomical regions and compared with splenic attenuation to derive the PSR. Diagnostic performance was evaluated using ROC curve analysis and multivariate logistic regression.
Results or Findings: Mean pancreatic attenuation was significantly lower in diabetic patients (HU:31) compared to prediabetic (HU:37) and normoglycemic controls (HU:44) (all comparisons, p < 0.001). PSR showed a stronger inverse correlation with HbA1c (r = -0.89 vs -0.75) and yielded a higher diagnostic accuracy for identifying diabetes (AUC: 0.88) compared to pancreatic attenuation alone (AUC: 0.79) (p<.001). Optimal cut-off values for pancreatic attenuation and PSR were 35 HU and 0.85, respectively. In multivariate analysis, the PSR remained an independent predictor of diabetic status, outperforming absolute pancreatic HU values.
Conclusion: PSR is a more robust and reliable CT-based biomarker for assessing diabetes risk than pancreatic attenuation alone. Its relative nature reduces scanner- and patient-related variability, making it a preferable parameter for opportunistic screening in routine CT.
Limitations: Single-center study; retrospective study design
Funding for this study: No funding
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Ethics committee approval number 2023-1216

Author Block: B. De Franciscis, D. Palumbo, A. Campisi, D. Tamburrino, S. Partelli, S. Crippa, M. Falconi, F. De Cobelli; Milano/IT
Purpose: Postoperative pancreatic fistula (POPF) remains the major cause of morbidity/mortality after pancreaticoduodenectomy (PD). The 2016 revision of the International Study Group for Pancreatic Surgery defines only grade B and grade C POPF as clinically relevant: grade B requires non surgical interventions, whereas grade C involves organ failure, reoperation or death. However, such classification is burdened by significant limitations, the main one being its retrospective nature. Radiological tools able to early stratify POPF severity and, more generally, morbidity burden after PD remain limited.
Methods or Background: We retrospectively analyzed 809 patients who underwent PD (01/2015–12/2023) and subsequent postoperative CT. 309 patients suffered from clinically relevant (CR)‑POPF. Radiological variables (anastomotic defect/dehiscence, fluid collections, perianastomotic air bubbles, HU values of pancreatic remnant) were considered. A scoring system was derived from significant radiologic features. Model performance was assessed using area under the ROC curve (AUC).
Results or Findings: Among 309 CR‑POPF patients, 259 were grade B and 50 grade C. Mortality was 44 % in grade C versus 0.4 % in grade B. Univariate analysis revealed that the presence of i) posterior anastomotic defect [2 points], ii) intra-abdominal fluid collections [1 point1], iii) perianastomotic air [2 points] and iv) lower pancreatic attenuation on CT [2 points] was associated with grade C POPF (p < 0.05). A postoperative CT score achieved AUC 0.74; a cut‑off ≥ 4 provided 84 % sensitivity and 73 % specificity. Also, the score strongly predicted morbidity (in terms of Clavien-Dindo grading) and mortality (AUC: 0.68).
Conclusion: Several early CT findings differentiate grade C from grade B POPF. A simple CT‑based score offers good discrimination. Prospective validation is necessary to establish these tools for early risk stratification and timely intervention.
Limitations: Retrospective Single Center
Funding for this study: None
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The number of Ethics committee's approvment is 06/INT/2022

Author Block: M-J. Sähn, S. Waltermann, J. Ottemöller, R. I. Danebrock, J. H. Niehoff, B. Gerdes, J. Borggrefe, N. Begum, A. S. Surov; Minden/DE
Purpose: Iodine concentration (IC) derived from dual energy computed tomography was recently explored as a potential biomarker for pancreatic NEN grade and Ki-67. With methodologies exhibiting significant variability, outcomes in recent publicatiobns were ambigous - ranging from, at best, weak correlation to strong predictive performance in complex multivariate analyses.
This study revisits the issue and aims to provide evidence for tumor characterisation using PCCT derived iodine concentration in pancreatic NEN.
Methods or Background: IC in neuroendocrine pancreatic primaries were analyzed in venous phase pre treatment regarding correlation with histopathological tumor grade and Ki-67. Pancreatic NEN IC was normalized vs. aortic IC (NIC), as well as relative to unaffected pancreatic tissue (RIC).
Correlations were analyzed using Spearman correlation, mean concentrations were analyzed using Mann-Whitney-U-test.
Results or Findings: Iodine concentration in P-NEN compared to normal pancreatic tissue (RIC) exhibited a strong correlation with tumor grade (ρ = 0.54, p = 0,021) and Ki-67 (ρ = 0.54, p = 0.021). Mean RIC were higher in high grade tumors (p = 0.028). NIC exhibited weak, non statistically significant correlation with tumor grade (ρ = 0.328, p = 0.183) and Ki-67 (ρ = 0.302, p = 0.224). Mean NIC did not diverge significantly in low grade vs. high grade pancreatic NET.
Conclusion: PCCT derived IC and especially RIC is a promising biomarker for tumor grade and Ki-67 prediction with strong, statistically significant correlations in untreated pancreatic NEN. The method is non-invasive, requires little to no additional resources and may support early, valid therapeutic decisions.
Limitations: Retrospective analysis
Funding for this study: None
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This retrospective study was approved by the Ethics Committee of the Faculty of Medicine, Ruhr-University Bochum (2021–827).

Author Block: L. Endrös¹, J. H. W. Bodden², F. Jungmann², R. L. Walter², M. Eiber², J. S. Kirschke², R. Braren¹, L. Steinhelfer²; ¹Hamburg/DE, ²Munich/DE
Purpose: To evaluate whether automated, opportunistic analysis of volumetric bone mineral density (BMD) from routine staging CT predicts progression-free survival (PFS) in resected pancreatic ductal adenocarcinoma (PDAC) under adjuvant chemotherapy, and to compare its prognostic value with CA 19-9, as established biomarkers such as CA 19-9 have known limitations.
Methods or Background: We retrospectively analyzed 112 patients with resected PDAC who received gemcitabine-based or FOLFIRINOX chemotherapy. Vertebral BMD was quantified preoperatively and after completion of adjuvant therapy using SpineQ (BoneScreen), a CE-certified AI tool providing absolute volumetric BMD values (mg/cm³). Associations with PFS were tested using univariate and multivariate Cox regression models.
Results or Findings: During adjuvant chemotherapy, mean BMD declined from 97.6 ± 42.5 to 73.7 ± 34.1 mg/cm³. In univariate Cox regression, relative BMD change, follow-up BMD, and baseline CA 19-9 were significantly associated with PFS, whereas CA 19-9 change was not. In multivariate analysis, only relative BMD change remained predictive (HR 0.947, p < 0.0001). Each 1% decline in BMD corresponded to an approximately 5% higher risk of progression. Follow-up BMD and baseline CA 19-9 lost significance in the adjusted model.
Conclusion: Automated CT-based volumetric BMD analysis provides robust, independent prognostic information in resected PDAC and outperforms CA 19-9 for predicting PFS. Opportunistic skeletal assessment from routine staging CT enables longitudinal monitoring and individualized risk stratification without additional imaging or patient burden.
Limitations: This was a retrospective single-center study with a moderate sample size. Validation in larger, prospective, multicenter cohorts is required before clinical implementation.
Funding for this study: No funding was provided for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This study was approved by the TUM Ethics Committee (reference number 2024-531-CD).

Author Block: A. I. Vigh, S. Nettesheim, T. Ojejinmi, M. Benke, Á. Szücs, A. Szijártó, P. Maurovich-Horvat, B. K. Budai, I. Dudás; Budapest/HU
Purpose: To compare radiation dose and image quality between photon-counting detector CT (PCD-CT) and energy-integrating detector CT (EID-CT) in patients with pancreatic disease who underwent abdominopelvic venous phase imaging.
Methods or Background: We retrospectively identified patients scanned with both PCD-CT and EID-CT between February 2022 and August 2023. Only patients without significant body-size change (L3 axial slice area difference <15%) were included, yielding 58 matched pairs. Radiation dose was compared using CTDIvol. Agreement in body size between scans was assessed using concordance correlation coefficients (CCC) and Bland–Altman analysis. Linear regression models tested the interaction between modality and body size. In a subset of 32 patients with identical venous phase protocols (3.0 mm slice thickness), signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR, muscle–fat) were measured at the L3 level.
Results or Findings: The mean time interval between scans was 116 ± 99 days, with a body-size difference of 4.3 ± 3.3 cm². Body size agreement was excellent (CCC = 0.968, LoA: –10.0 to 11.3 cm²). CTDIvol was significantly lower for PCD-CT than EID-CT (paired p < 0.0001), independent of body size (interaction p = 0.291). In protocol-matched cases, muscle SNR was significantly higher for PCD-CT (p < 0.0001), while muscle–fat CNR was not significantly different (p = 0.875).
Conclusion: PCD-CT provides substantial radiation dose reduction compared with EID-CT without compromising body-size comparability. In addition, PCD-CT improves muscle SNR, although CNR between muscle and fat remains unchanged. These results support the potential of PCD-CT to enhance dose efficiency and image quality in abdominopelvic imaging of pancreatic disease.
Limitations: This was a single-center study with a retrospective study design. Only venous phase 3.0 slice thickness scans were investigated.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by the institutional ethics committee.

Author Block: C. Lyu, W. Qu, J. Li, Z. Li; Wuhan/CN
Purpose: To develop a non-invasive predictive model by integrating auto-segmented abdominal CT-based volumetric body composition with machine learning to preoperatively distinguish low-grade from high-grade Gastroenteropancreatic Neuroendocrine Neoplasms(GEP-NENs).
Methods or Background: This multicenter analysis enrolled 633 GEP-NENs patients from three institutions. Patients were divided into: Training set (n=403) and internal validation (n=174) (7:3 ratio from Hospital 1); test set (n=56 from 2 other hospitals). An nnUNetv2-based automatic segmentation algorithm of abdominal fat tissue and skeletal muscle on arterial-phase CT was applied. Visceral fat index, subcutaneous fat index, intermuscular fat index and skeletal muscle index were calculated. Features with a p-value < 0.05 were selected using univariate logistic regression and included in the prediction model built with extreme gradient boosting algorithm. Receiver operating characteristic curves and decision curve analysis were performed to evaluate the utility of the model. SHapley Additive explanation was conducted to enhance the model interpretability and visualization.
Results or Findings: The automatic segmentation achieved a Dice coefficient of 0.90. For pathological grading, a model built with body composition parameters achieved an AUC of 0.863 in the training set, 0.750 in the validation set, and 0.717 in the test set. SHAP analysis revealed that the relative intermuscular adipose tissue(rIMAT) contributed most among the body composition parameters to the model decision-making. And rIMAT levels were higher in P53-mutant and CK19-positive cases compared to negative cases.
Conclusion: Auto-segmented abdominal body composition combined with machine learning-based model could provide an assisted, non-invasive tool for prediction of pathological grade in GEP-NENs.
Limitations: First,changes in CT scanning protocols are inevitable. Second, retrospective and observational designs may be prone to have selection bias.
Funding for this study: National Natural Science Foundation of China (NSFC) [grants number 82202127, 82071889 and 62131009].
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study received approval from the institutional review board of our institution (TJ-IRB202411010) on 4 November 2024. We have registered it on the https://www.clinicaltrials.gov/ and the Unique Identifying Number (UIN) is NCT06983106.

Author Block: B. Kataria, I. Khudair, W. C. Bartholomä, P. Sandström, B. Björnsson, M. Woisetschlager; Linköping/SE
Purpose: For accurate diagnosis, optimal contrast timing is crucial in pancreatic CT imaging. To standardize scan timing, fixed protocols or bolus-tracking techniques are used. However, these methods often fail to account for individual physiological differences, potentially compromising image quality. The purpose of this study was to identify patient-specific parameters in order to individualize scan timing.
Methods or Background: Data from 16 patients who underwent a CT perfusion and a 4-phase abdominal examination at the same time point were used to evaluate the relationship between patient-specific parameters and contrast enhancement timing across three specific time intervals:
t1: Time to aortic threshold >100 HU
t2: Time from aortic threshold to maximum pancreatic enhancement
t3: Total time from injection start to peak pancreatic enhancement
Correlation analyses were performed between the three time intervals and variables including height, cardiac volume, lean body mass (LBM), and fat-related metrics.
Results or Findings: At time t1, significant correlations were observed for height (r = 0.550, p = 0.027) and cardiac volume (r = 0.595, p = 0.015); and at time t3 for LBM (r = 0.501, p = 0.048) and cardiac volume (r = 0.528, p = 0.036). No significant associations were observed at time t2. Strong inter-parameter relationships were noted between height and cardiac volume (r = 0.758,p=0.0007), and between cardiac volume and LBM (r = 0.812, p=0.0001).
Conclusion: Cardiac volume and LBM are easily accessible promising predictors of contrast timing in abdominal CT and may enable individualized timing, thus improving diagnostic accuracy in patients with diverse body compositions. Further research is warranted to validate these findings and to support clinical implementation.
Limitations: Small sample size
Funding for this study: LFoU from Region Östergötland, Sweden
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Ref. no. 2016/43-31