Search

Objectives. To externally validate and assess the accuracy of a previously trained fully automatic nnU-Net CNN algorithm to identify and segment primary neuroblastoma tumors in MR images in a large children cohort. Methods. An international multicenter, multivendor imaging repository of patients with neuroblastic tumors was used to validate the performance of a trained Machine Learning (ML) tool to identify and delineate primary neuroblastoma tumors. The dataset was heterogeneous and completely independent from the one used to train and tune the model, consisting of 300 children with neuroblastic tumors having 535 MR T2-weighted sequences (486 sequences at diagnosis and 49 after finalization of the first phase of chemotherapy). The automatic segmentation algorithm was based on a nnU-Net architecture developed within the PRIMAGE project. For comparison, the segmentation masks were manually edited by an expert radiologist, and the time for the manual editing was recorded. Different overlaps and spatial metrics were calculated to compare both masks. Results. The median Dice Similarity Coefficient (DSC) was high 0.997; 0.944–1.000 (median; Q1–Q3). In 18 MR sequences (6%), the net was not able neither to identify nor segment the tumor. No differences were found regarding the MR magnetic field, type of T2 sequence, or tumor location. No significant differences in the performance of the net were found in patients with an MR performed after chemotherapy. The time for visual inspection of the generated masks was 7.9 ± 7.5 (mean ± Standard Deviation (SD)) seconds. Those cases where manual editing was needed (136 masks) required 124 ± 120 s. Conclusions. The automatic CNN was able to locate and segment the primary tumor on the T2-weighted images in 94% of cases. There was an extremely high agreement between the automatic tool and the manually edited masks. This is the first study to validate an automatic segmentation model for neuroblastic tumor identification and segmentation with body MR images. The semi-automatic approach with minor manual editing of the deep learning segmentation increases the radiologist’s confidence in the solution with a minor workload for the radiologist.

Tumor segmentation is one of the key steps in imaging processing. The goals of this study were to assess the inter-observer variability in manual segmentation of neuroblastic tumors and to analyze whether the state-of-the-art deep learning architecture nnU-Net can provide a robust solution to detect and segment tumors on MR images. A retrospective multicenter study of 132 patients with neuroblastic tumors was performed. Dice Similarity Coefficient (DSC) and Area Under the Receiver Operating Characteristic Curve (AUC ROC) were used to compare segmentation sets. Two more metrics were elaborated to understand the direction of the errors: the modified version of False Positive (FPRm) and False Negative (FNR) rates. Two radiologists manually segmented 46 tumors and a comparative study was performed. nnU-Net was trained-tuned with 106 cases divided into five balanced folds to perform cross-validation. The five resulting models were used as an ensemble solution to measure training (n = 106) and validation (n = 26) performance, independently. The time needed by the model to automatically segment 20 cases was compared to the time required for manual segmentation. The median DSC for manual segmentation sets was 0.969 (±0.032 IQR). The median DSC for the automatic tool was 0.965 (±0.018 IQR). The automatic segmentation model achieved a better performance regarding the FPRm. MR images segmentation variability is similar between radiologists and nnU-Net. Time leverage when using the automatic model with posterior visual validation and manual adjustment corresponds to 92.8%.

Simple Summary There is growing interest in applying quantitative diffusion techniques to magnetic resonance imaging for cancer diagnosis and treatment. These measurements are used as a surrogate marker of tumor cellularity and aggressiveness, although there may be factors that introduce some bias to these approaches. Thus, we explored a novel methodology based on confidence habitats and voxel uncertainty to improve the power of the apparent diffusion coefficient to discriminate between benign and malignant neuroblastic tumor profiles in children. We were able to show this offered an improved sensitivity and negative predictive value relative to standard voxel-based methodologies. Background/Aim: In recent years, the apparent diffusion coefficient (ADC) has been used in many oncology applications as a surrogate marker of tumor cellularity and aggressiveness, although several factors may introduce bias when calculating this coefficient. The goal of this study was to develop a novel methodology (Fit-Cluster-Fit) based on confidence habitats that could be applied to quantitative diffusion-weighted magnetic resonance images (DWIs) to enhance the power of ADC values to discriminate between benign and malignant neuroblastic tumor profiles in children. Methods: Histogram analysis and clustering-based algorithms were applied to DWIs from 33 patients to perform tumor voxel discrimination into two classes. Voxel uncertainties were quantified and incorporated to obtain a more reproducible and meaningful estimate of ADC values within a tumor habitat. Computational experiments were performed by smearing the ADC values in order to obtain confidence maps that help identify and remove noise from low-quality voxels within high-signal clustered regions. The proposed Fit-Cluster-Fit methodology was compared with two other methods: conventional voxel-based and a cluster-based strategy. Results: The cluster-based and Fit-Cluster-Fit models successfully differentiated benign and malignant neuroblastic tumor profiles when using values from the lower ADC habitat. In particular, the best sensitivity (91%) and specificity (89%) of all the combinations and methods explored was achieved by removing uncertainties at a 70% confidence threshold, improving standard voxel-based sensitivity and negative predictive values by 4% and 10%, respectively. Conclusions: The Fit-Cluster-Fit method improves the performance of imaging biomarkers in classifying pediatric solid tumor cancers and it can probably be adapted to dynamic signal evaluation for any tumor.