Wilms tumor is a common pediatric solid tumor. To evaluate tumor response to chemotherapy and decide whether nephron-sparing surgery is possible, tumor volume measurements based on magnetic resonance imaging (MRI) are important. Currently, radiological volume measurements are based on measuring tumor dimensions in three directions. Manual segmentation-based volume measurements might be more accurate, but this process is time-consuming and user-dependent. The aim of this study was to investigate whether manual segmentation-based volume measurements are more accurate and to explore whether these segmentations can be automated using deep learning. We included the MRI images of 45 Wilms tumor patients (age 0–18 years). First, we compared radiological tumor volumes with manual segmentation-based tumor volume measurements. Next, we created an automated segmentation method by training a nnU-Net in a five-fold cross-validation. Segmentation quality was validated by comparing the automated segmentation with the manually created ground truth segmentations, using Dice scores and the 95th percentile of the Hausdorff distances (HD95). On average, manual tumor segmentations result in larger tumor volumes. For automated segmentation, the median dice was 0.90. The median HD95 was 7.2 mm. We showed that radiological volume measurements underestimated tumor volume by about 10% when compared to manual segmentation-based volume measurements. Deep learning can potentially be used to replace manual segmentation to benefit from accurate volume measurements without time and observer constraints.

Background: Medical images play an important role in diagnosis and treatment of pediatric solid tumors. The field of radiology, pathology, and other image‐based diagnostics are getting increasingly important and advanced. This indicates a need for advanced image processing technology such as Deep Learning (DL).Aim: Our review focused on the use of DL in multidisciplinary imaging in pediatric surgical oncology.Methods: A search was conducted within three databases (Pubmed, Embase, and Scopus), and 2056 articles were identified. Three separate screenings were performed for each identified subfield.Results: In total, we identified 36 articles, divided between radiology (n = 22), pathology (n = 9), and other image‐based diagnostics (n = 5). Four types of tasks were identified in our review: classification, prediction, segmentation, and synthesis. General statements about the studies'’ performance could not be made due to the inhomogeneity of the included studies. To implement DL in pediatric clinical practice, both technical validation and clinical validation are of uttermost importance.Conclusion: In conclusion, our review provided an overview of all DL research in the field of pediatric surgical oncology. The more advanced status of DL in adults should be used as guide to move the field of DL in pediatric oncology further, to keep improving the outcomes of children with cancer.

Patient-specific 3D models of neuroblastoma and relevant anatomy are useful tools for surgical planning. However, these models do not represent the heterogenous biology of neuroblastoma. This heterogeneity is visualized with the ADC and ¹²³I-MIGB-SPECT-CT imaging. Combining these multi-modal data into preoperative 3D heatmaps, may allow differentiation of the areas of vital and non-vital tumor tissue. We developed a workflow to create multi-modal preoperative 3D models for neuroblastoma surgery. We included 7 patients who underwent neuroblastoma surgery between 2022 and 2023. We developed 3D models based on the contrast enhanced T1-weighted MRI scans. Subsequently, we aligned the corresponding ADC and ¹²³I-MIBG-SPECT-CT images using rigid transformation. We estimated registration precision using the Dice score and the target registration error (TRE). 3D heatmaps were computed based on ADC and ¹²³I-MIBG uptake. The registration algorithm had a median Dice score of 0.81 (0.75-0.90) for ADC and 0.77 (0.65-0.91) for ¹²³I-MIBG-SPECT. For the ADC registration, the median TRE of renal vessels was 4.90 mm (0.86-10.18) and of the aorta 4.67 mm (1.59-12.20). For the ¹²³I -MIBG-SPECT imaging the TRE of the renal vessels was 5.52 mm (1.71-10.97) and 5.28 mm (3.33-16.77) for the aorta. We successfully developed a registration workflow to create multi-modal 3D models which allows the surgeon to visualize the tumor and its biological behavior in relation to the surrounding tissue. Future research will include linking of pathological results to imaging data, to validate these multi-modal 3D models. Level IV. Clinical Research.

Surgery plays a key role in treating neuroblastoma. To assist surgical planning, anatomical 3D models derived from the segmentation of anatomical structures on MRI scans are often used. Automation using deep learning can make segmentations less time-consuming and more reliable. We organized the Surgical Planning in PedIatric Neuroblastoma (SPPIN) challenge, to stimulate developments and benchmarking of automatic segmentation of neuroblastoma on MRI. SPPIN is the first segmentation challenge in extracranial pediatric oncology. Nine teams provided a valid submission. Evaluation was based on the Dice similarity coefficient (Dice score), the 95th percentile of the Hausdorff distance (HD95), and the volumetric similarity (VS). A combination of these scores determined the ranking of the teams. The spread in the median evaluation scores per team was large (Dice: 0.21–0.82; HD95: 63.31–7.69; VS: 0.31–0.91). The top-performing team achieved a median Dice score of 0.82 (with an HD95 of 7.69 mm and a VS of 0.91) using a large, pre-trained model. However, in the pre-operative segmentations, significantly lower evaluation scores were observed. Our results indicate that pre-training might be useful in small, pediatric datasets. Although the general results of the winning team were high, they were insufficient to use for surgical planning in small, pre-operative tumors.

Renal mass segmentation has important potential to enhance the clinical workflow, especially in settings requiring quantitative assessments. Kidney volume could serve as an important biomarker for renal diseases, with changes in volume correlating directly with kidney function. Currently, clinical practice often relies on subjective visual assessment for evaluating kidney size and kidney lesions, including tumors and cysts, which are typically staged based on diameter, volume, and anatomical location. To support a more objective and reproducible approach, this research aims to develop a robust, thoroughly validated renal mass segmentation algorithm, named Renal-Net. We employ publicly available training datasets and leverage the state-of-the-art medical image segmentation framework nnU-Net. Validation is conducted using both proprietary and public test datasets, with segmentation performance quantified by Dice coefficient and the 95th percentile Hausdorff distance. Furthermore, we analyze robustness across subgroups based on patient sex, age, CT contrast phases, and tumor histologic subtypes. Our findings demonstrate that our segmentation algorithm, trained exclusively on publicly available data, generalizes effectively to external test sets and outperforms existing state-of-the-art models across all tested datasets. Subgroup analyses reveal consistent high performance, indicating strong robustness and reliability. The developed algorithm and associated code are publicly accessible at https://github.com/DIAGNijmegen/oncology-kidney-abnormality-segmentation

Creating 3D models based on pre-operative MRI of patients with a Wilms tumor (WT) can aid surgical planning. However, creating these models requires manual delineation (segmentation) of the MRI imaging. Deep learning can automate this, but most validations of these segmentation methods are retrospective. This article prospective evaluation of a WT segmentation method in a clinical workflow aimed at creating 3D models for surgical planning. A deep learning-based segmentation method was developed and retrospectively validated on a dataset of 56 patients. This method, based on nnU-Net, segmented both kidney and tumor and was implemented within the current clinical workflow. It was tested prospectively on 10 consecutive patients with WT. The performance of this method was quantified using the Dice score between the automated and corrected segmentations, the time required for the various steps of the clinical workflow and an analysis of segmentation errors. In 2/10 patients the automated segmentation was sufficient to be used directly. In 8/10 patients the automated segmentation needed corrections (with a median correction time of 11 min). The median Dice score for kidney was 1.00 (range: 0.58-1.00) and for tumor 0.98 (range: 0.00-1.00). The segmentation errors identified most often were an under-segmentation of tumor borders (n = 3) and the incorrect identification of the tumor/kidney border (n = 3). The implementation of an automated segmentation method for creating 3D models of patients with a WT is feasible in current clinical workflows. In 20% of the patients, no corrections were needed and for most other patients, corrections could be applied in less than 15 min. None.