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.

Computer-aided methods have shown added value for diagnosing and predicting brain disorders and can thus support decision making in clinical care and treatment planning. This chapter will provide insight into the type of methods, their working, their input data –such as cognitive tests, imaging, and genetic data– and the types of output they provide. We will focus on specific use cases for diagnosis, i.e., estimating the current “condition” of the patient, such as early detection and diagnosis of dementia, differential diagnosis of brain tumors, and decision making in stroke. Regarding prediction, i.e., estimation of the future “condition” of the patient, we will zoom in on use cases such as predicting the disease course in multiple sclerosis and predicting patient outcomes after treatment in brain cancer. Furthermore, based on these use cases, we will assess the current state-of-the-art methodology and highlight current efforts on benchmarking of these methods and the importance of open science therein. Finally, we assess the current clinical impact of computer-aided methods and discuss the required next steps to increase clinical impact.

Histopathological growth patterns (HGPs) are independent prognosticators for colorectal liver metastases (CRLM). Currently, HGPs are determined postoperatively. In this study, we evaluated radiomics for preoperative prediction of HGPs on computed tomography (CT), and its robustness to segmentation and acquisition variations. Patients with pure HGPs [i.e. 100% desmoplastic (dHGP) or 100% replacement (rHGP)] and a CT-scan who were surgically treated at the Erasmus MC between 2003–2015 were included retrospectively. Each lesion was segmented by three clinicians and a convolutional neural network (CNN). A prediction model was created using 564 radiomics features and a combination of machine learning approaches by training on the clinician’s and testing on the unseen CNN segmentations. The intra-class correlation coefficient (ICC) was used to select features robust to segmentation variations; ComBat was used to harmonize for acquisition variations. Evaluation was performed through a 100 × random-split cross-validation. The study included 93 CRLM in 76 patients (48% dHGP; 52% rHGP). Despite substantial differences between the segmentations of the three clinicians and the CNN, the radiomics model had a mean area under the curve of 0.69. ICC-based feature selection or ComBat yielded no improvement. Concluding, the combination of a CNN for segmentation and radiomics for classification has potential for automatically distinguishing dHGPs from rHGP, and is robust to segmentation and acquisition variations. Pending further optimization, including extension to mixed HGPs, our model may serve as a preoperative addition to postoperative HGP assessment, enabling further exploitation of HGPs as a biomarker.

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.

Purpose We aimed to derive a "robustness recipe" giving the range robustness (RR) and setup robustness (SR) settings (ie, the error values) that ensure adequate clinical target volume (CTV) coverage in oropharyngeal cancer patients for given Gaussian distributions of systematic setup, random setup, and range errors (characterized by standard deviations of Σ, σ, and ρ, respectively) when used in minimax worst-case robust intensity modulated proton therapy (IMPT) optimization. Methods and Materials For the analysis, contoured computed tomography (CT) scans of 9 unilateral and 9 bilateral patients were used. An IMPT plan was considered robust if, for at least 98% of the simulated fractionated treatments, 98% of the CTV received 95% or more of the prescribed dose. For fast assessment of the CTV coverage for given error distributions (ie, different values of Σ, σ, and ρ), polynomial chaos methods were used. Separate recipes were derived for the unilateral and bilateral cases using one patient from each group, and all 18 patients were included in the validation of the recipes. Results Treatment plans for bilateral cases are intrinsically more robust than those for unilateral cases. The required RR only depends on the ρ, and SR can be fitted by second-order polynomials in Σ and σ. The formulas for the derived robustness recipes are as follows: Unilateral patients need SR = -0.15Σ² + 0.27σ² + 1.85Σ - 0.06σ + 1.22 and RR=3% for ρ = 1% and ρ = 2%; bilateral patients need SR = -0.07Σ² + 0.19σ² + 1.34Σ - 0.07σ + 1.17 and RR=3% and 4% for ρ = 1% and 2%, respectively. For the recipe validation, 2 plans were generated for each of the 18 patients corresponding to Σ = σ = 1.5 mm and ρ = 0% and 2%. Thirty-four plans had adequate CTV coverage in 98% or more of the simulated fractionated treatments; the remaining 2 had adequate coverage in 97.8% and 97.9%. Conclusions Robustness recipes were derived that can be used in minimax robust optimization of IMPT treatment plans to ensure adequate CTV coverage for oropharyngeal cancer patients.