Joint Modeling of RNAseq and Radiomics Data for Glioma Molecular Characterization and Prediction

RNA sequencing (RNAseq) is a recent technology that profiles gene expression by measuring the relative frequency of the RNAseq reads. RNAseq read counts data is increasingly used in oncologic care and while radiology features (radiomics) have also been gaining utility in radiology practice such as disease diagnosis, monitoring, and treatment planning. However, contemporary literature lacks appropriate RNA-radiomics (henceforth, radiogenomics) joint modeling where RNAseq distribution is adaptive and also preserves the nature of RNAseq read counts data for glioma grading and prediction. The Negative Binomial (NB) distribution may be useful to model RNAseq read counts data that addresses potential shortcomings. In this study, we propose a novel radiogenomics-NB model for glioma grading and prediction. Our radiogenomics-NB model is developed based on differentially expressed RNAseq and selected radiomics/volumetric features which characterize tumor volume and sub-regions. The NB distribution is fitted to RNAseq counts data, and a log-linear regression model is assumed to link between the estimated NB mean and radiomics. Three radiogenomics-NB molecular mutation models (e.g., IDH mutation, 1p/19q codeletion, and ATRX mutation) are investigated. Additionally, we explore gender-specific effects on the radiogenomics-NB models. Finally, we compare the performance of the proposed three mutation prediction radiogenomics-NB models with different well-known methods in the literature: Negative Binomial Linear Discriminant Analysis (NBLDA), differentially expressed RNAseq with Random Forest (RF-genomics), radiomics and differentially expressed RNAseq with Random Forest (RF-radiogenomics), and Voom-based count transformation combined with the nearest shrinkage classifier (VoomNSC). Our analysis shows that the proposed radiogenomics-NB model significantly outperforms (ANOVA test, p < 0.05) for prediction of IDH and ATRX mutations and offers similar performance for prediction of 1p/19q codeletion, when compared to the competing models in the literature, respectively.

Evaluation of Rockfall Hazard Based On UAV Technology And 3D Rockfall Simulations

In this study, the rockfall hazard in Hacıabdullah village located in the Central Anatolia region of Turkey was assessed with three-dimensional (3D) rockfall analyses based on unmanned aerial vehicle (UAV) technology. With several rockfall disasters experienced in the village, the final event occurred as severe rockfall in 2008 and several houses were evacuated due to rockfall risk after this event. In order for the rockfall hazard to be assessed close to reality in the study area, rocks with falling potential identified in the field were assessed using high-resolution digital surface model (DSM) data produced from images obtained with UAV. During field studies, 17 rocks with fall hazard were identified and dimensional measurements were performed. According to dimensional values, the geometric and volumetric features of each rock were assessed close to reality with the RAMMS 3D rockfall modelling program. As a result of modelling, the kinetic energies of the rocks were identified to reach up to 3476 kJ, with velocities of up to 23.1 m/s and bounce heights of up to 14.57 m. On steep slopes rocks do not travel very long distances; however, in gentle slopes, they were identified to be able to roll very long distances. Rocks that do not move very far from the source are; in other words, where the fall process is dominant, may create damage on roads mainly. However, those with the feature of rolling, in other words, blocks which can travel long distances from the source area, have the potential to cause great damage to settlement areas, roads and trees. According to the hazard map, modelling of rock blocks numbered R6, R12, R13, R14, R15, R16 and R17 showed settlement units were within the high and moderate risk areas.

Machine learning-optimized Combinatorial MRI scale (COMRISv2) correlates highly with cognitive and physical disability scales in Multiple Sclerosis patients

Composite MRI scales of central nervous system tissue destruction correlate stronger with clinical outcomes than their individual components in multiple sclerosis (MS) patients. Using machine learning (ML), we previously developed Combinatorial MRI scale (COMRISv1) solely from semi-quantitative (semi-qMRI) biomarkers. Here, we asked how much better COMRISv2 might become with the inclusion of quantitative (qMRI) volumetric features and employment of more powerful ML algorithm.The prospectively acquired MS patients, divided into training (n=172) and validation (n=83) cohorts underwent brain MRI imaging and clinical evaluation. Neurological examination was transcribed to NeurEx app that automatically computes disability scales. qMRI features were computed by LesionTOADS algorithm. Modified random forest pipeline selected biomarkers for optimal model(s) in the training cohort.COMRISv2 models validated moderate correlation with cognitive disability (Rho = 0.674; Linh’s concordance coefficient [CCC] = 0.458; p<0.001) and strong correlations with physical disability (Spearman Rho = 0.830-0.852; CCC = 0.789-0.823; p<0.001). The NeurEx led to the strongest COMRISv2 model. Addition of qMRI features enhanced performance only of cognitive disability model, likely because semi-qMRI biomarkers measure infratentorial injury with greater accuracy.COMRISv2 models predict most granular clinical scales in MS with remarkable criterion validity, expanding scientific utilization of cohorts with missing clinical data.

Lung Nodule Classification Using Biomarkers, Volumetric Radiomics, and 3D CNNs

We present a hybrid algorithm to estimate lung nodule malignancy that combines imaging biomarkers from Radiologist’s annotation with image classification of CT scans. Our algorithm employs a 3D Convolutional Neural Network (CNN) as well as a Random Forest in order to combine CT imagery with biomarker annotation and volumetric radiomic features. We analyze and compare the performance of the algorithm using only imagery, only biomarkers, combined imagery + biomarkers, combined imagery + volumetric radiomic features, and finally the combination of imagery + biomarkers + volumetric features in order to classify the suspicion level of nodule malignancy. The National Cancer Institute (NCI) Lung Image Database Consortium (LIDC) IDRI dataset is used to train and evaluate the classification task. We show that the incorporation of semi-supervised learning by means of K-Nearest-Neighbors (KNN) can increase the available training sample size of the LIDC-IDRI, thereby further improving the accuracy of malignancy estimation of most of the models tested although there is no significant improvement with the use of KNN semi-supervised learning if image classification with CNNs and volumetric features is combined with descriptive biomarkers. Unexpectedly, we also show that a model using image biomarkers alone is more accurate than one that combines biomarkers with volumetric radiomics, 3D CNNs, and semi-supervised learning. We discuss the possibility that this result may be influenced by cognitive bias in LIDC-IDRI because malignancy estimates were recorded by the same radiologist panel as biomarkers, as well as future work to incorporate pathology information over a subset of study participants.

Deep Learning Achieves Neuroradiologist-Level Performance in Detecting Hydrocephalus

PurposeWe aim to develop automated detection of hydrocephalus requiring treatment in a heterogeneous patient population referred for MRI brain scans, and compare performance to that of neuroradiologists.Materials and MethodsWe leveraged 496 clinical MRI brain scans (259 hydrocephalus) collected retrospectively at a single clinical site from patients aged 2–90 years (mean 54) referred for any reason. Sixteen MRI scans (ten hydrocephalus) were segmented semi-automatically in 3D to delineate ventricles, extraventricular CSF, and brain tissues. A 3D CNN was trained on these segmentations and subsequently used to automatically segment the remaining 480 scans. To detect hydrocephalus, volumetric features such as volumes of ventricles and temporal horns were computed from the segmentation and were used to train a linear classifier. Machine performance was evaluated in a diagnosis dataset where hydrocephalus was confirmed as requiring surgical intervention, and compared to four neuroradiologists on a random subset of 240 scans. The pipeline was tested on a separate screening dataset of 205 scans collected from a routine clinical population aged 1–95 years (mean 56) to predict the majority reading from four neuroradiologists using images alone.ResultsWhen compared to the neuroradiologists at a matched sensitivity, the machine did not show a significant difference in specificity (proportions test, p > 0.05). The machine demonstrated comparable performance in independent diagnosis and screening datasets. Overall ROC performance compared favorably with the state-of-the-art (AUC 0.82–0.93).ConclusionHydrocephalus can be detected automatically from MRI in a heterogeneous patient population with performance equivalent to that of neuroradiologists.Summary statementA two-stage automated pipeline was developed to segment head MRI and extract volumetric features to accurately and efficiently detect hydrocephalus that required shunting and achieved performance comparable to that of trained neuroradiologists.Key PointsWe developed a state-of-the-art 3D deep convolutional network to perform fully automated segmentation of the ventricles, extraventricular cerebrospinal fluid, and brain tissues in anisotropic MRI brain scans in a heterogeneous patient population.Volumetric features extracted from anatomical segmentations can be used to classify hydrocephalus (which may require neurosurgical intervention) vs. non-hydrocephalus.When tested in an independent dataset, the network achieved performance comparable to that of expert neuroradiologists.

How to Achieve Efficiency and Accuracy in Discrete Element Simulation of Asphalt Mixture: A DRF-Based Equivalent Model for Asphalt Sand Mortar

The clump-based discrete element model is one of the asphalt mixture simulation methods, which has the potential to not only predict mixture performance but also simulate particle movement during compaction, transporting, and other situations. However, modelling of asphalt sand mortar in this method remains to be a problem due to computing capacity. Larger-sized balls (generally 2.0–2.36 mm) were usually used to model the smaller particles and asphalt binder, but this replacement may result in the mixture’s unrealistic volumetric features. More specifically, replacing original elements with equal volume but larger size particles will increase in buck volume and then different particle contacting states. The major objective of this research is to provide a solution to the dilemma situation through an improved equivalent model of the smaller particles and asphalt binders. The key parameter of the equivalent model is the diameter reduction factor (DRF), which was proposed in this research to minimize the effects of asphalt mortar’s particle replacement modelling. To determine DRF, the DEM-based analysis was conducted to evaluate several mixture features, including element overlap ratio, ball-wall contact number, and the average wall stress. Through this study, it was observed that when the original glued ball diameters are ranging from 2.00 mm and 2.36 mm, the diameter reduction factor changes from 0.82 to 0.86 for AC mixtures and 0.80 to 0.84 for SMA mixtures. The modelling method presented in this research is suitable not only for asphalt mixtures but also for the other particulate mix with multisize particles.

Deep learning for glioblastoma segmentation using preoperative magnetic resonance imaging identifies volumetric features associated with survival

Background Measurement of volumetric features is challenging in glioblastoma. We investigate whether volumetric features derived from preoperative MRI using a convolutional neural network–assisted segmentation is correlated with survival. Methods Preoperative MRI of 120 patients were scored using Visually Accessible Rembrandt Images (VASARI) features. We trained and tested a multilayer, multi-scale convolutional neural network on multimodal brain tumour segmentation challenge (BRATS) data, prior to testing on our dataset. The automated labels were manually edited to generate ground truth segmentations. Network performance for our data and BRATS data was compared. Multivariable Cox regression analysis corrected for multiple testing using the false discovery rate was performed to correlate clinical and imaging variables with overall survival. Results Median Dice coefficients in our sample were (1) whole tumour 0.94 (IQR, 0.82–0.98) compared to 0.91 (IQR, 0.83–0.94 p = 0.012), (2) FLAIR region 0.84 (IQR, 0.63–0.95) compared to 0.81 (IQR, 0.69–0.8 p = 0.170), (3) contrast-enhancing region 0.91 (IQR, 0.74–0.98) compared to 0.83 (IQR, 0.78–0.89 p = 0.003) and (4) necrosis region were 0.82 (IQR, 0.47–0.97) compared to 0.67 (IQR, 0.42–0.81 p = 0.005). Contrast-enhancing region/tumour core ratio (HR 4.73 [95% CI, 1.67–13.40], corrected p = 0.017) and necrotic core/tumour core ratio (HR 8.13 [95% CI, 2.06–32.12], corrected p = 0.011) were independently associated with overall survival. Conclusion Semi-automated segmentation of glioblastoma using a convolutional neural network trained on independent data is robust when applied to routine clinical data. The segmented volumes have prognostic significance.

Association of metabolic and genetic heterogeneity in head and neck squamous cell carcinoma with prognostic implications: integration of FDG PET and genomic analysis

Purpose The linkage between the genetic and phenotypic heterogeneity of the tumor has not been thoroughly evaluated. Herein, we investigated how the genetic and metabolic heterogeneity features of the tumor are associated with each other in head and neck squamous cell carcinoma (HNSC). We further assessed the prognostic significance of those features. Methods The mutant-allele tumor heterogeneity (MATH) score (n = 508), a genetic heterogeneity feature, and tumor glycolysis feature (GlycoS) (n = 503) were obtained from the HNSC dataset in the cancer genome atlas (TCGA). We identified matching patients (n = 33) who underwent 18F-fluorodeoxyglucose positron emission tomography (FDG PET) from the cancer imaging archive (TCIA) and obtained the following information from the primary tumor: metabolic, metabolic-volumetric, and metabolic heterogeneity features. The association between the genetic and metabolic features and their prognostic values were assessed. Results Tumor metabolic heterogeneity and metabolic-volumetric features showed a mild degree of association with MATH (n = 25, ρ = 0.4~0.5, P < 0.05 for all features). The patients with higher FDG PET features and MATH died sooner. Combination of MATH and tumor metabolic heterogeneity features showed a better stratification of prognosis than MATH. Also, higher MATH and GlycoS were associated with significantly worse overall survival (n = 499, P = 0.002 and 0.0001 for MATH and GlycoS, respectively). Furthermore, both MATH and GlycoS independently predicted overall survival after adjusting for clinicopathologic features and the other (P = 0.015 and 0.006, respectively). Conclusion Both tumor metabolic heterogeneity and metabolic-volumetric features assessed by FDG PET showed a mild degree of association with genetic heterogeneity in HNSC. Both metabolic and genetic heterogeneity features were predictive of survival and there was an additive prognostic value when the metabolic and genetic heterogeneity features were combined. Also, MATH and GlycoS were independent prognostic factors in HNSC; they can be used for precise prognostication once validated.