SSSylvia Saalfeld

Sylvia Saalfeld

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Sylvia Saalfeld has a total of 25 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Machine Learning-Based Prediction of Delayed Cerebral Ischemia and Functional Outcome in Aneurysmal Subarachnoid Hemorrhage: The Role of CNS Infection-Related Parameters

    by Vanessa Magdalena Swiatek, Tom Tobias Kummer, Conrad-Jakob Swiatek, Lea Ehrhardt, Klaus-Peter Stein, Ali Rashidi, Robert Werdehausen, I. Erol Sandalcioglu, Sylvia Saalfeld, Belal Neyazi

    Background/Objectives: Aneurysmal subarachnoid hemorrhage (aSAH) carries high morbidity and mortality, with delayed cerebral ischemia (DCI) as a major secondary complication. Although infection has increasingly been implicated in DCI pathogenesis, the role of central nervous system (CNS) infection remains unclear. We evaluated CNS infection-related parameters for predicting DCI and functional outcome after aSAH using machine learning. Methods: This retrospective single-center study included 191 patients with confirmed aSAH treated in the intensive care unit between 2019 and 2024. Demographic, clinical, radiographic, treatment-related, microbiological, cerebrospinal fluid (CSF), and longitudinal laboratory data were extracted from electronic records. DCI was analyzed as a binary outcome, and discharge functional outcome was dichotomized using the modified Rankin Scale (0–2 vs. 3–6). Twelve feature selection methods and twelve classifiers were evaluated using five-fold cross-validation. Results: For DCI prediction, the models achieved a mean accuracy of 0.96, F1-score of 0.96, and AUC of 0.98. Key predictors included antibiotic therapy, aneurysm treatment modality, history of thrombosis, and EVD revision, along with mean albumin, red cell distribution width, INR, minimum aspartate aminotransferase, and CSF glucose, lactate, nucleated cell count, and cellular debris. For functional outcome prediction, accuracy was 0.948, F1-score 0.95, and AUC was 0.97. Predictors comprised cerebral vasospasm, number of spasmolysis procedures, pre-admission anticoagulation, CSF pathogen detection and type, aPTT, fibrinogen, aspartate aminotransferase, lactate dehydrogenase, and CSF mononuclear cell proportion, leukocyte count, and glucose concentration. Conclusions: Machine learning models integrating clinical, systemic, and CSF-derived parameters demonstrated excellent performance in predicting both DCI and functional outcome after aSAH. The identified predictors highlight the importance of the neurocritical care course, including infection-related and thromboinflammatory processes, and extend beyond traditional vasospasm-centered paradigms. These findings may inform risk stratification, although the temporal relationship between several predictors and outcome onset means the models should be regarded as hypothesis-generating rather than tools for early clinical prediction.

  • Open Access

    Couinaud segment-aware deep learning on point clouds for major liver resection planning

    by Joy Rakshit, Janine Rothert, Georg Hille, Tobias Huber, Hauke Lang, Rabea Margies, Florentine Huettl, Sylvia Saalfeld

    Purpose: In this study, we address the problem of automatic liver resection planning for major surgical procedures, including hemi-hepatectomy and extended hemi-hepatectomy, using deep learning. Motivated by clinical practice, where Couinaud liver segments are routinely used to describe tumor location and guide surgical decision-making, we investigate whether incorporating this anatomical information can improve model performance and clinical relevance. Methods: We propose a point cloud-based geometric deep learning approach based on a modified RandLA-Net architecture to predict liver resection zones. The model was trained and evaluated on 70 hemi-hepatectomy cases from Johannes Gutenberg University, Mainz, Germany (internal dataset). Two composite loss functions were evaluated: cross-entropy (CE) combined with intersection over union (IoU) and CE combined with Dice loss. For each loss function, models were trained with and without Couinaud segment information. Generalizability was assessed on an external dataset of 30 hemi-hepatectomy cases from the colorectal liver metastases (CRLM) cohort. Results: Both loss functions achieved comparable performance across the evaluated datasets, with CE $$+$$ + IoU consistently outperforming CE $$+$$ + Dice. On the internal test set, incorporating Couinaud segment information increased the IoU mean from 0.787 to 0.804 and the F1-score from 0.864 to 0.870. A Wilcoxon signed-rank test on 15 paired cases confirmed a statistically significant improvement in IoU mean ( p = 0.030), with 80% of cases showing improvement. On the external dataset, IoU mean improved from 0.666 to 0.702 and F1-score from 0.786 to 0.799 when Couinaud information was included. Excluding five anatomically complex cases, a Wilcoxon signed-rank test on the remaining 25 paired cases showed a significant improvement in IoU mean ( p = 0.019), with 68% of cases demonstrating improved performance. Conclusion: Explicit integration of Couinaud segment information improves both quantitative performance and clinical relevance in automatic major liver resection planning, particularly by better preserving critical vascular structures.

  • Open Access

    Data-Driven Analysis of Systemic Indicators Linking Stroke-Associated Pneumonia, Delayed Cerebral Ischemia, and Outcome After Aneurysmal Subarachnoid Hemorrhage

    by Vanessa Magdalena Swiatek, Conrad-Jakob Schiffner, Tom Tobias Kummer, Lea Ehrhardt, Klaus-Peter Stein, Ali Rashidi, Sylvia Saalfeld, Robert Werdehausen, I. Erol Sandalcioglu, Belal Neyazi

    Background/Objectives: Delayed cerebral ischemia (DCI) is a major cause of poor outcome after aneurysmal subarachnoid hemorrhage (aSAH). Beyond large-vessel vasospasm, DCI reflects a systemic, multifactorial process involving inflammation, hematologic dysregulation, and organ dysfunction. Stroke-associated pneumonia (SAP), a frequent aSAH complication linked to stroke-induced immunodepression, may aggravate secondary ischemic injury. Unlike prior studies focusing on classical predictors alone, we included pneumonia and longitudinal respiratory parameters alongside inflammatory, hematologic, and renal markers. Using machine learning, this study aimed to identify predictors of DCI and functional outcome from routinely collected intensive care data. Methods: In this retrospective single-center study, 182 aSAH patients treated in a neurosurgical intensive care unit were included. Clinical data, SAP status, and longitudinal inflammatory, hematologic, renal, and respiratory parameters were extracted. DCI and functional outcome were assessed. Continuous variables were summarized as minimum, maximum, and mean values. Supervised machine learning models combining 12 feature selection methods and 12 classifiers were trained using five-fold cross-validation and evaluated by accuracy, F1-score, and AUC. Results: DCI occurred in 22% of patients, and SAP in 27%. The machine learning models achieved a mean accuracy of 59.7% (F1-score 58.8%, AUC 59.7%) for DCI prediction. No single dominant feature emerged; predictive patterns included leukocyte counts, CRP, erythrocyte indices, platelet variability, renal function, and oxygenation metrics. Functional outcome prediction performed moderately better (mean AUC 65.7%) and shared overlapping predictors. Conclusions: DCI reflects systemic instability in aSAH, with longitudinal inflammatory and respiratory variability outperforming static thresholds. Dynamic risk stratification may enable earlier detection of deterioration, supporting future time-series modeling and external validation.


  • Open Access

    Prognostic potential of radiomics evaluation of lung artery thrombus for pulmonary embolism patients

    by Lea Ehrhardt, Patrique Fiedler, Alexey Surov, Sylvia Saalfeld

    Purpose: This study evaluates radiomics correlation with mortality and suitability as prognostic indicator for troponin for pulmonary embolism to enhance prognostic accuracy and guide personalized treatment strategies with the help of machine learning. We conducted an initial study focusing on texture information of the arterial thrombus. Method: Computed tomography (CT) of the lung from 86 patients with pulmonary embolism was used. As target variables, we used patients 30-day mortality and troponin results. Each arterial thrombus was manually segmented. After the extraction of their radiomics features and the reduction via correlation analysis and 12 feature selection methods, these and the target variables were given to 12 different classification methods to record the accuracies (Acc.), F1-scores (F1) and ROC curve areas under the curve (AUC) for comparison and evaluation. Results: The resulting accuracy achieved was 0.967, the F1-score 0.973 for class 0 and 0.967 for class 1 and the AUC around 0.9686. The feature selection methods which resulted in the highest results were ReliefF (RF), Logistic Regression (LOR) and CART Classification (CARTC). For the classification methods, Support Vector Machines (SVM), eXtreme Gradiant Boosting (XGB) and Ensemble Bagged Trees (EBT) lead to the highest results. Firstorder, Shape and gray-level co-occurrence matrix (GLCM) were the most selected radiomics feature classes. Conclusion: Within this study, we conducted radiomics feature extraction within a medical image data analysis pipeline with subsequent correlation analysis and training of classifiers for patients with pulmonary lung embolism. We could show that the radiomics features correlated with patient’s morphology as well as troponin range with an accuracy of 0.967 and 0.9302, respectively, yield high potential for prognosis and treatment strategy of pulmonary embolism patients in the future.


  • Open Access

    Deep learning-based segmentation of acute pulmonary embolism in cardiac CT images

    by Ehsan Amini, Georg Hille, Janine Hürtgen, Alexey Surov, Sylvia Saalfeld

    Purpose: Acute pulmonary embolism (APE) is a common pulmonary condition that, in severe cases, can progress to right ventricular hypertrophy and failure, making it a critical health concern surpassed in severity only by myocardial infarction and sudden death. CT pulmonary angiogram (CTPA) is a standard diagnostic tool for detecting APE. However, for treatment planning and prognosis of patient outcome, an accurate assessment of individual APEs is required. Methods: Within this study, we compiled and prepared a dataset of 200 CTPA image volumes of patients with APE. We then adapted two state-of-the-art neural networks; the nnU-Net and the transformer-based VT-UNet in order to provide fully automatic APE segmentations. Results: The nnU-Net demonstrated robust performance, achieving an average Dice similarity coefficient (DSC) of 88.25 ± 10.19% and an average 95th percentile Hausdorff distance (HD95) of 10.57 ± 34.56 mm across the validation sets in a five-fold cross-validation framework. In comparison, the VT-UNet was achieving on par accuracies with an average DSC of 87.90 ± 10.94% and a mean HD95 of 10.77 ± 34.19 mm. Conclusions: We applied two state-of-the-art networks for automatic APE segmentation to our compiled CTPA dataset and achieved superior experimental results compared to the current state of the art. In clinical routine, accurate APE segmentations can be used for enhanced patient prognosis and treatment planning.


  • Open Access

    Semi-automatic segmentation of elongated interventional instruments for online calibration of C-arm imaging system

    by Negar Chabi, Alfredo Illanes, Oliver Beuing, Daniel Behme, Bernhard Preim, Sylvia Saalfeld

    Purpose: The C-arm biplane imaging system, designed for cerebral angiography, detects pathologies like aneurysms using dual rotating detectors for high-precision, real-time vascular imaging. However, accuracy can be affected by source-detector trajectory deviations caused by gravitational artifacts and mechanical instabilities. This study addresses calibration challenges and suggests leveraging interventional devices with radio-opaque markers to optimize C-arm geometry. Methods: We propose an online calibration method using image-specific features derived from interventional devices like guidewires and catheters (In the remainder of this paper, the term”catheter” will refer to both catheter and guidewire). The process begins with gantry-recorded data, refined through iterative nonlinear optimization. A machine learning approach detects and segments elongated devices by identifying candidates via thresholding on a weighted sum of curvature, derivative, and high-frequency indicators. An ensemble classifier segments these regions, followed by post-processing to remove false positives, integrating vessel maps, manual correction and identification markers. An interpolation step filling gaps along the catheter. Results: Among the optimized ensemble classifiers, the one trained on the first frames achieved the best performance, with a specificity of 99.43% and precision of 86.41%. The calibration method was evaluated on three clinical datasets and four phantom angiogram pairs, reducing the mean backprojection error from 4.11 ± 2.61 to 0.15 ± 0.01 mm. Additionally, 3D accuracy analysis showed an average root mean square error of 3.47% relative to the true marker distance. Conclusions: This study explores using interventional tools with radio-opaque markers for C-arm self-calibration. The proposed method significantly reduces 2D backprojection error and 3D RMSE, enabling accurate 3D vascular reconstruction.