University of Twente Student Theses


Multi-modal 3-Dimensional visualization of pediatric neuroblastoma : aiding surgical planning beyond anatomical information

Simons, D.C. (2023) Multi-modal 3-Dimensional visualization of pediatric neuroblastoma : aiding surgical planning beyond anatomical information.

Full text not available from this repository.

Full Text Status:Access to this publication is restricted
Abstract:Purpose: Patient-specific 3-dimensional (3D) models of neuroblastoma and relevant anatomy are useful tools for surgical planning, particularly of tumors encasing blood vessels or vital organs. However, these models do not represent the heterogeneous biology of neuroblastoma. Clinically, this heterogeneity is visualized with the Apparent Diffusion Coefficient (ADC) and 123I-MIBG-SPECT-CT scans. By combining this multi-modal data to form a preoperative 3D model, we may allow delineation of areas of viable and non-viable tumor tissue. In this study, we developed a workflow to create multi-modal preoperative 3D models for neuroblastoma surgery and to link the imaging to pathology results. With the latter we can identify whether the heterogeneous imaging reflects tumor viability and whether this can be used for surgical planning. Methods: We included patients who underwent neuroblastoma surgery in 2022-2023. We developed 3D models based on gadolinium enhanced T1-weighted MRI scans. Subsequently, we aligned these with corresponding ADC and 123I-MIBG-SPECT-CT images with a rigid transformation. We estimated registration precision via Dice coefficient on the unaffected kidney and with the target registration error (TRE) of various anatomical structures. 3D risk models were computed based on ADC values and 123I-MIBG uptake by thresholding. Preoperative imaging data was correlated to the histopathology by using a patient specific 3D printed cutting guide. The accuracy of the different risk areas was evaluated by denoting whether vital tumor tissue was present or not, with histopathology. Results: The registration algorithm had a mean Dice coefficient for the kidney of 0.83 (STD=0.06) and 0.80 (STD=0.12) for the ADC and the 123I-MIBG-SPECT-CT respectively. For the ADC registration, the mean TRE of the center of mass of the kidney, renal vessels and aorta was 5.38 (STD=3.76), 5.42 mm (STD=2.83) and 5.32 mm (STD=4.11) respectively. For the 123I-MIBG-SPECT-CT imaging the TRE was 5.44 mm (STD=5.32), 3.66 mm (STD=3.56) and 4.39 mm (STD=5.51) respectively. Risk models were successfully created for the registered ADC and 123I-MIBG-SPECT-CT scans. High-risk areas on the ADC risk models contained in 70.0% only vital tumor tissue and the low-risk areas contained in 68.4% only non-viable tissue. For 123I-MIBG risk models this was 77.8% and 60.0% respectively. Conclusion: We successfully developed a registration workflow to develop and validate multi-modal 3D models of neuroblastoma with histopathology. With this, surgeons are able to visualize the tumor and potentially its biological behavior in relation to its surrounding tissue. More evidence must be gained to evaluate the accuracy and reliability of multi-modal 3D models in relation to the pathology results.
Item Type:Essay (Master)
Faculty:TNW: Science and Technology
Subject:44 medicine
Programme:Technical Medicine MSc (60033)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page