Combined molecular subtyping, grading, and segmentation of glioma using multi-task deep learning.

BACKGROUND: Accurate characterization of glioma is crucial for clinical decision making. A delineation of the tumor is also desirable in the initial decision stages but is time-consuming. Previously, deep learning methods have been developed that can either non-invasively predict the genetic or histological features of glioma, or that can automatically delineate the tumor, but not both tasks at the same time. Here, we present our method that can predict the molecular subtype and grade, while simultaneously providing a delineation of the tumor. METHODS: We developed a single multi-task convolutional neural network that uses the full 3D, structural, preoperative MRI scans to predict the IDH mutation status, the 1p/19q co-deletion status, and the grade of a tumor, while simultaneously segmenting the tumor. We trained our method using a patient cohort containing 1508 glioma patients from 16 institutes. We tested our method on an independent dataset of 240 patients from 13 different institutes. RESULTS: In the independent test set, we achieved an IDH-AUC of 0.90, an 1p/19q co-deletion AUC of 0.85, and a grade AUC of 0.81 (grade II/III/IV). For the tumor delineation, we achieved a mean whole tumor Dice score of 0.84. CONCLUSIONS: We developed a method that non-invasively predicts multiple, clinically relevant features of glioma. Evaluation in an independent dataset shows that the method achieves a high performance and that it generalizes well to the broader clinical population. This first-of-its-kind method opens the door to more generalizable, instead of hyper-specialized, AI methods.

Predicting the 1p/19q Codeletion Status of Presumed Low-Grade Glioma with an Externally Validated Machine Learning Algorithm.

PURPOSE: Patients with 1p/19q codeleted low-grade glioma (LGG) have longer overall survival and better treatment response than patients with 1p/19q intact tumors. Therefore, it is relevant to know the 1p/19q status. To investigate whether the 1p/19q status can be assessed prior to tumor resection, we developed a machine learning algorithm to predict the 1p/19q status of presumed LGG based on preoperative MRI. EXPERIMENTAL DESIGN: Preoperative brain MR images from 284 patients who had undergone biopsy or resection of presumed LGG were used to train a support vector machine algorithm. The algorithm was trained on the basis of features extracted from post-contrast T1-weighted and T2-weighted MR images and on patients' age and sex. The performance of the algorithm compared with tissue diagnosis was assessed on an external validation dataset of MR images from 129 patients with LGG from The Cancer Imaging Archive (TCIA). Four clinical experts also predicted the 1p/19q status of the TCIA MR images. RESULTS: The algorithm achieved an AUC of 0.72 in the external validation dataset. The algorithm had a higher predictive performance than the average of the neurosurgeons (AUC 0.52) but lower than that of the neuroradiologists (AUC of 0.81). There was a wide variability between clinical experts (AUC 0.45-0.83). CONCLUSIONS: Our results suggest that our algorithm can noninvasively predict the 1p/19q status of presumed LGG with a performance that on average outperformed the oncological neurosurgeons. Evaluation on an independent dataset indicates that our algorithm is robust and generalizable.

Stereotactic radiosurgery for brain AVMs: role of interobserver variation in target definition on digital subtraction angiography.

PURPOSE: We evaluated the extent of interobserver variation in contouring arteriovenous malformations (AVMs) on digital subtraction angiography (DSA) with respect to volume, spatial localization, and dosimetry and correlated our findings with the clinical outcome. METHODS AND MATERIALS: Thirty-one patients who had undergone radiosurgery for brain AVMs were studied. Six clinicians independently contoured the nidus on the original DSA. As a measure of variation, the ratio between the volumes of agreement and the corresponding encompassing volumes, as well as the absolute positional shift between the individual target volumes were derived. Using the original treatment plan, the dosimetric coverage of the individually contoured volumes with standard collimators was compared with a similar plan using dynamic conformal arcs. RESULTS: The mean contoured nidus volume was 3.6 +/- 5.6 cm3. The mean agreement ratio was 0.45 +/- 0.18 for all possible pairs of observers. The mean absolute positional shift between individually contoured volumes was 2.8 +/- 2.6 mm. These differences were more marked in previously treated groups and tended to be more pronounced in those with treatment failure. The mean coverage of the individual volumes by the 80% prescription isodose was 88.1% +/- 3.2% using conventional collimators and 78.9% +/- 4.4% using dynamic conformal arcs (p = 0.001). CONCLUSION: Substantial interobserver variations exist when contouring brain AVMs on DSA for the purpose of radiosurgical planning. Such variations may result in underdosage to the AVM and, thereby, contribute to treatment failure. The consequences of contouring variations may increase with the use of more conformal radiosurgical techniques.