Preoperative Brain Tumor Imaging: Models and Software for Segmentation and Standardized Reporting

Abstract

For patients suffering from brain tumor, prognosis estimation and treatment decisions are made by a multidisciplinary team based on a set of preoperative MR scans. Currently, the lack of standardized and automatic methods for tumor detection and generation of clinical reports, incorporating a wide range of tumor characteristics, represents a major hurdle. In this study, we investigate the most occurring brain tumor types: glioblastomas, lower grade gliomas, meningiomas, and metastases, through four cohorts of up to 4,000 patients. Tumor segmentation models were trained using the AGU-Net architecture with different preprocessing steps and protocols. Segmentation performances were assessed in-depth using a wide-range of voxel and patient-wise metrics covering volume, distance, and probabilistic aspects. Finally, two software solutions have been developed, enabling an easy use of the trained models and standardized generation of clinical reports: Raidionics and Raidionics-Slicer. Segmentation performances were quite homogeneous across the four different brain tumor types, with an average true positive Dice ranging between 80 and 90%, patient-wise recall between 88 and 98%, and patient-wise precision around 95%. In conjunction to Dice, the identified most relevant other metrics were the relative absolute volume difference, the variation of information, and the Hausdorff, Mahalanobis, and object average symmetric surface distances. With our Raidionics software, running on a desktop computer with CPU support, tumor segmentation can be performed in 16–54 s depending on the dimensions of the MRI volume. For the generation of a standardized clinical report, including the tumor segmentation and features computation, 5–15 min are necessary. All trained models have been made open-access together with the source code for both software solutions and validation metrics computation. In the future, a method to convert results from a set of metrics into a final single score would be highly desirable for easier ranking across trained models. In addition, an automatic classification of the brain tumor type would be necessary to replace manual user input. Finally, the inclusion of post-operative segmentation in both software solutions will be key for generating complete post-operative standardized clinical reports.

Segmentation of glioblastomas in early post-operative multi-modal MRI with deep neural networks

“…On a second level, the output segmentation masks can be used to differentiate between patients with remnant tumor after surgery and gross total resection patients, with increasing balanced accuracy performance as more sequences are added to the model inputs. Our early post-operative glioblastoma segmentation models have been made freely available in the Raidionics environment 41 .…”

“…The selected architectures are the nnU-Net, state-of-the-art for pre-operative glioblastoma segmentation, and AGU-Net, an architecture developed for pre-operative segmentation of brain tumors. These architectures have both demonstrated excellent performance on pre-operative segmentation in previous studies on pre-operative brain tumor segmentation [27][28][29] , and they exhibit different strengths and weaknesses. The automatic results are compared with manual segmentations, using different combinations of MRI scans in a large dataset consisting of paired pre-and early post-operative MRI scans from 956 patients in 12 medical centers in Europe and the United States.…”

“…Extensive validation studies are presented to identify the best architecture configuration, quantify the performances and ability to generalize, and highlight potential relevance for use in clinical practice. Finally, the best performing models are made publicly available and integrated into the open software Raidionics 29 .…”

The relationship between pathological brain activity and functional network connectivity in glioma patients

“…To define these peritumoral regions, tumor masks were either manually drawn in, slice by slice [LD], on post-gadolinium T1-weighted and FLAIR anatomical images, 28 or automatically segmented using a neural network algorithm and visually checked. 29 Using all 210 cortical regions of the Brainnetome atlas 30 , individual regions were considered part of the peritumoral area when at least 12% of the region's volume overlapped with the tumor mask (Supplementary materials for more information). The contralateral homologue of the peritumoral area (2) was defined as the same atlas region(s) where the tumor was located but in the contralateral hemisphere.…”

Unraveling the Efficiency of Non-Navigated 2D Intraoperative Ultrasound in Glioma Surgery: Challenging the Demand for Increased Technological Sophistication in Intraoperative Imaging

“…The Radionics software toolkit 14 was utilized as the cornerstone for conducting tumor location analysis and gauging the potential for resectability for the glioblastoma subgroup of patients. This method underwent a series of preprocessing stages.…”