Volumetric Features Research Articles

Front Cover

AbstractIn our work we present an optofluidic‐based microscope on chip for 3D imaging of Drosophila embryos with dual color capabilities. The cover image shows the device during experimental validation along with the acquired images of two different embryos which are automatically optically sectioned while flowing in the microfluidic channel. The frames reveal internal volumetric features of the specimens.Further details can be found in the article by Roberto Memeo, Petra Paiè, Federico Sala, Michele Castriotta, Chiara Guercio, Thomas Vaccari, Roberto Osellame, Andrea Bassi, and Francesca Bragheri (e202000396). image

A novel technology to integrate imaging and clinical markers for non-invasive diagnosis of lung cancer

This study presents a non-invasive, automated, clinical diagnostic system for early diagnosis of lung cancer that integrates imaging data from a single computed tomography scan and breath bio-markers obtained from a single exhaled breath to quickly and accurately classify lung nodules. CT imaging and breath volatile organic compounds data were collected from 47 patients. Spherical Harmonics-based shape features to quantify the shape complexity of the pulmonary nodules, 7th-Order Markov Gibbs Random Field based appearance model to describe the spatial non-homogeneities in the pulmonary nodule, and volumetric features (size) of pulmonary nodules were calculated from CT images. 27 VOCs in exhaled breath were captured by a micro-reactor approach and quantied using mass spectrometry. CT and breath markers were input into a deep-learning autoencoder classifier with a leave-one-subject-out cross validation for nodule classification. To mitigate the limitation of a small sample size and validate the methodology for individual markers, retrospective CT scans from 467 patients with 727 pulmonary nodules, and breath samples from 504 patients were analyzed. The CAD system achieved 97.8% accuracy, 97.3% sensitivity, 100% specificity, and 99.1% area under curve in classifying pulmonary nodules.

Baseline Volumetric T2 Relaxation Time Histogram Analysis: Can It Be Used to Predict the Response to Intravenous Methylprednisolone Therapy in Patients With Thyroid-Associated Ophthalmopathy?

ObjectivePrediction of therapy response to intravenous methylprednisolone pulses (ivMP) is crucial for thyroid-associated ophthalmopathy (TAO). Image histograms may offer sensitive imaging biomarkers for therapy effect prediction. This study aimed to investigate whether pretherapeutic, multiparametric T2 relaxation time(T2RT) histogram features of extraocular muscles (EOMs) can be used to predict therapy response.Materials and MethodsForty-five active and moderate-severe TAO patients, who were treated with standard ivMP and underwent orbital MRI before therapy, were retrospectively included in this study. The patients were divided into responsive (n = 24, 48 eyes) and unresponsive group(n = 21, 42 eyes) according to clinical evaluation. Baseline clinical features of patients and histogram-derived T2RT parameters of the EOMs were analyzed and compared. Logistic regression model was conducted to determine independent predictors, and a histogram features nomogram was formulated for personalized prediction.ResultsResponsive group displayed lower values for 5th, 10th percentiles (P < 0.050, respectively), and higher values for 75th, 90th, and 95th percentiles, skewness, entropy, and inhomogeneity (P < 0.050, respectively) than unresponsive group. Multivariate logistic regression analysis showed that 95th percentile of >88.1 [odds ratio (OR) = 12.078; 95% confidence interval (CI) = 3.98–36.655, p < 0.001], skewness of >0.31 (OR = 3.935; 95% CI = 2.28–6.788, p < 0.001) and entropy of >3.41 (OR = 4.375; 95% CI = 2.604–7.351, p < 0.001) were independent predictors for favorable response. The nomogram integration of three independent predictors demonstrated optimal predictive efficiency, with a C-index of 0.792.ConclusionsPre-treatment volumetric T2RT histogram features of EOMs could function to predict the response to ivMP in patients with TAO. The nomogram based on histogram features facilitates the selection of patients who will derive maximal benefit from ivMP.

Structural Integrity of Steel Pipeline with Clusters of Corrosion Defects.

The main goal of this paper is to evaluate the burst pressure and structural integrity of a steel pipeline based on in-line inspection results, in respect to the grouping criteria of closely spaced volumetric surface features. In the study, special attention is paid to evaluation of data provided from the diagnostics using an axial excitation magnetic flux leakage technology in respect to multiple defects grouping. Standardized clustering rules were applied to the corrosion pits taken from an in-line inspection of the gas transmission pipeline. Basic rules of interaction of pipe wall metal losses are expressed in terms of longitudinal and circumferential spacing of the features in the colony. The effect of interactions of the detected anomalies on the tube residual strength evaluated according to the Det Norske Veritas Recommended practice was investigated in the current study. In the presented case, groups of closely-spaced defects behaved similarly as individual flaws with regard to their influence on burst pressure and pipeline failure probability.

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.

Automatic cardiac cine MRI segmentation and heart disease classification

Cardiac cine magnetic resonance imaging (MRI) continues to be recognized as an established modality for non-invasive assessment of the function and structure of the cardiovascular system. Making full use of fully convolutional neural networks CNNs ability to operate pixel-wise classification, cine MRI sequences can be segmented and volumetric features of three key heart structures are computed for disease prediction. The three key heart structures are the left ventricle cavity, right ventricle cavity and the left ventricle myocardium. In this paper, we suggest an automated pipeline for both cardiac segmentation and diagnosis. The study was conducted on a dataset of 150 patients from Dijon Hospital in the context of the post-2017 Medical Image Computing and Computer Assisted Intervention MICCAI, Automated Cardiac Diagnosis Challenge (ACDC). The challenge consists in two phases: (i) a segmentation contest, where performance is evaluated on dice overlap coefficient and Hausdorff distance metrics, and a (ii) diagnosis contest for heart disease classification. For this aim, we propose the use of a deep learning based network for segmentation of the three key cardiac structures within short-axis cine MRI sequences and a classifier ensemble for heart disease classification. The deep learning segmentation network is a UNet fully convolutional neural network variant with fewer trainable parameters. The classifier ensemble consists in combining three classifiers, namely a multilayer perceptron, a random forest and a support vector machine. Before feeding the segmentation network, a preliminary step consists in localizing heart region and cropping input images to a restricted region of interest (ROI). This is achieved by a signal processing based approach and aims at reducing multi-class imbalance and computational load. We achieved nearly state of the art accuracy performances for both the segmentation and disease classification challenges. Reporting a mean dice overlap coefficient of 0.92 for the three cardiac structures segmentation, along with good limits of agreement for the various derived clinical indices, leading to an accuracy of 0.92 for the disease classification on unseen data.

Volumetric Feature-Based Alzheimer’s Disease Diagnosis From sMRI Data Using a Convolutional Neural Network and a Deep Neural Network

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder that is mostly prevalent in people older than 65 years. The hippocampus is a widely studied region of interest (ROI) for a number of reasons, such as memory function analysis, stress development observation and neurological disorder investigation. Moreover, hippocampal volume atrophy is known to be linked with Alzheimer’s disease. On the other hand, several biomarkers, such as amyloid beta ( $\text{a}\beta _{42}$ ) protein, tau, phosphorylated tau and hippocampal volume atrophy, are being used to diagnose AD. In this research work, we have proposed a method to diagnose AD based on slice-wise volumetric features extracted from the left and right hippocampi of structural magnetic resonance imaging (sMRI) data. The proposed method is an aggregation of a convolutional neural network (CNN) model with a deep neural network (DNN) model. The left and right hippocampi have been localized automatically using a two-stage ensemble Hough-CNN. The localized hippocampal positions are used to extract ( $80\times 80\times $ 80 voxels) 3-D patches. The 2-D slices are then separated from the 3-D patches along axial, sagittal, and coronal views. The pre-processed 2-D patches are used to extract volumetric features from each slice by using a discrete volume estimation convolutional neural network (DVE-CNN) model. The extracted volumetric features have been used to train and test the classification network. The proposed approach has achieved average weighted classification accuracies of 94.82% and 94.02% based on the extracted volumetric features attributed to the left and right hippocampi, respectively. In addition, it has achieved area under the curve (AUC) values of 92.54% and 90.62% for the left and right hippocampi, respectively. Our method has outperformed the other methods by a certain margin in the same dataset.

Can Skeletal Joint Positional Ordering Influence Action Recognition on Spectrally Graded CNNs: A Perspective on Achieving Joint Order Independent Learning

3D skeletal based action recognition is being practiced with features extracted from joint positional sequence modeling on deep learning frameworks. However, the spatial ordering of skeletal joints during the entire action recognition lifecycle is found to be fixed across datasets and frameworks. Intuition inspired us to investigate through experimentation, the influence of multiple random skeletal joint ordered features on the performance of deep learning systems. Therefore, the argument: can joint order independent learning for skeletal action recognition practicable? If practicable, the goal is to discover how many different types of randomly ordered joint feature representations are sufficient for training deep networks. Implicitly, we further investigated on multiple features and deep networks that recorded highest performance on jumbled joints. This work proposes a novel idea of learning skeletal joint volumetric features on a spectrally graded CNN to achieve joint order independence. Intuitively, we propose 4 joint features called as quad joint volumetric features (QJVF), which are found to offer better spatio temporal relationships between time series joint data when compared to existing features. Consequently, we propose a Spectrally graded Convolutional Neural Network (SgCNN) to characterize spatially divergent features extracted from jumbled skeletal joints. Finally, evaluation of the proposed hypothesis has been experimented on our 3D skeletal action KLHA3D102, KLYOGA3D datasets along with benchmarks, HDM05, CMU and NTU RGB D. The results demonstrated that the joint order independent feature learning is achievable on CNNs trained on quantified spatio temporal feature maps extracted from randomly shuffled skeletal joints from action sequences.

Design and Performance of Scaled-Up Microwave Dryer for Seaweed Drying

Direct sunlight is commonly used to dry fresh seaweed by artisanal farmers in Indonesia due to its low cost and ease of handling. Nevertheless, this method poses some drawbacks such as lengthy duration, weather dependency and quality degradation. The application of microwave technology in food processing has progressed dramatically, including in the drying process. The microwave drying method is more efficient and can shorten the processing time. This study aimed to describe a large-scale microwave dryer (MD) design and performance to assist the fresh seaweed drying process. The design concept applies microwave energy with a volumetric heating feature to accelerate the seaweed drying process without damaging its functional groups. The MD dimensions were 2410 (l) x 270 (w) x 210 (h) mm with a dryer capacity of up to six kilograms, an enlarged cavity and multiple magnetrons. The main components of the MD were cavity, air circulation system, drying system and control system. According to the performance testing, the MD’s optimum performance was at a power level setting of P7 and six kilograms load. At this setting, we obtained a dried seaweed with a moisture ratio of 0.68±0.05, drying rate of 30.29±1.32 g/min, specific energy consumption of 3.96±0.08 MJ/kg H2O and energy efficiency of 58.45±2.65%. The total power of the P7 setting operation required 2.00 kW. Fourier-Transform Infrared (FTIR) spectra showed that the functional groups of the dried seaweed were unaltered.

Multi-stage learning for segmentation of aortic dissections using a prior aortic anatomy simplification.

Aortic dissection (AD) is a life-threatening cardiovascular disease with a high mortality rate. The accurate and generalized 3-D reconstruction of AD from CT-angiography can effectively assist clinical procedures and surgery plans, however, is clinically unavaliable due to the lacking of efficient tools. In this study, we presented a novel multi-stage segmentation framework for type B AD to extract true lumen (TL), false lumen (FL) and all branches (BR) as different classes. Two cascaded neural networks were used to segment the aortic trunk and branches and to separate the dual lumen, respectively. An aortic straightening method was designed based on the prior vascular anatomy of AD, simplifying the curved aortic shape before the second network. The straightening-based method achieved the mean Dice scores of 0.96, 0.95 and 0.89 for TL, FL, and BR on a multi-center dataset involving 120 patients, outperforming the end-to-end multi-class methods and the multi-stage methods without straightening on the dual-lumen segmentation, even using different network architectures. Both the global volumetric features of the aorta and the local characteristics of the primary tear could be better identified and quantified based on the straightening. Comparing to previous deep learning methods dealing with AD segmentations, the proposed framework presented advantages in segmentation accuracy.

A physics-guided modular deep-learning based automated framework for tumor segmentation in PET

An important need exists for reliable positron emission tomography (PET) tumor-segmentation methods for tasks such as PET-based radiation-therapy planning and reliable quantification of volumetric and radiomic features. To address this need, we propose an automated physics-guided deep-learning-based three-module framework to segment PET images on a per-slice basis. The framework is designed to help address the challenges of limited spatial resolution and lack of clinical training data with known ground-truth tumor boundaries in PET. The first module generates PET images containing highly realistic tumors with known ground-truth using a new stochastic and physics-based approach, addressing lack of training data. The second module trains a modified U-net using these images, helping it learn the tumor-segmentation task. The third module fine-tunes this network using a small-sized clinical dataset with radiologist-defined delineations as surrogate ground-truth, helping the framework learn features potentially missed in simulated tumors. The framework was evaluated in the context of segmenting primary tumors in 18F-fluorodeoxyglucose (FDG)-PET images of patients with lung cancer. The framework’s accuracy, generalizability to different scanners, sensitivity to partial volume effects (PVEs) and efficacy in reducing the number of training images were quantitatively evaluated using Dice similarity coefficient (DSC) and several other metrics. The framework yielded reliable performance in both simulated (DSC: 0.87 (95% confidence interval (CI): 0.86, 0.88)) and patient images (DSC: 0.73 (95% CI: 0.71, 0.76)), outperformed several widely used semi-automated approaches, accurately segmented relatively small tumors (smallest segmented cross-section was 1.83 cm2), generalized across five PET scanners (DSC: 0.74 (95% CI: 0.71, 0.76)), was relatively unaffected by PVEs, and required low training data (training with data from even 30 patients yielded DSC of 0.70 (95% CI: 0.68, 0.71)). In conclusion, the proposed automated physics-guided deep-learning-based PET-segmentation framework yielded reliable performance in delineating tumors in FDG-PET images of patients with lung cancer.

Numerical simulation of propane MILD combustion in a lab-scale cylindrical furnace

This paper reports a CFD numerical study of propane (C3H8) combustion inside a laboratory-scale cylindrical furnace under moderate or intense low oxygen dilution (MILD) regime. The reliability of the present numerical simulation was firstly validated against previous experiment (Energy & Fuels, 2018, 32(8): 8817–8829), and the C3H8 MILD combustion characteristics were thoroughly examined through a comparison with its traditional combustion counterpart, in terms of temperature distribution, heat release behavior, CO formation, and NO formation as well as reduction mechanisms. The results show that, MILD combustion produces a lower overall heat flux on furnace walls than traditional combustion, while the proportion of convection heat transfer increases remarkably, suggesting the different heat transfer behaviors between the two combustion modes. In addition, both MILD and traditional combustion modes exhibit only the positive heat release region inside the furnace, however, C3H8 consumption pathway indicates that traditional combustion features strong C3H8 pyrolysis in the early ignition stage (x = 80 mm), which is not found in MILD combustion. Moreover, the formation of CO in traditional combustion is dominated by the oxidation of larger hydrocarbon molecule in the intense reaction region (x = 240 mm), while CO in MILD combustion comes from the oxidation of smaller hydrocarbon molecules in both early ignition and intense reaction regions, showing its volumetric reaction feature. Furthermore, the relative contributions of the prompt route as well as the N2O and NNH intermediate routes to final NO emission are enhanced in MILD combustion comparing to those in traditional combustion, however, these routes still can be neglected in both traditional and MILD combustion due to their significantly low yields. For NO reduction, reburning route plays a more important role than SNCR route in both traditional and MILD combustion, and is even more profound in the latter. Overall, thermal-route still dominates the NO formation for both combustion modes in the present furnace.

MD-NDNet: a multi-dimensional convolutional neural network for false-positive reduction in pulmonary nodule detection

Pulmonary nodule false-positive reduction is of great significance for automated nodule detection in clinical diagnosis of low-dose computed tomography (LDCT) lung cancer screening. Due to individual intra-nodule variations and visual similarities between true nodules and false positives as soft tissues in LDCT images, the current clinical practices remain subject to shortcomings of potential high-risk and time-consumption issues. In this paper, we propose a multi-dimensional nodule detection network (MD-NDNet) for automatic nodule false-positive reduction using deep convolutional neural network (DCNNs). The underlying method collaboratively integrates multi-dimensional nodule information to complementarily and comprehensively extract nodule inter-plane volumetric correlation features using three-dimensional CNNs (3D CNNs) and spatial nodule correlation features from sagittal, coronal, and axial planes using two-dimensional CNNs (2D CNNs) with attention module. To incorporate different sizes and shapes of nodule candidates, a multi-scale ensemble strategy is employed for probability aggregation with weights. The proposed method is evaluated on the LUNA16 challenge dataset in ISBI 2016 with ten-fold cross-validation. Experiment results show that the proposed framework achieves classification performance with a CPM score of 0.9008. All of these indicate that our method enables an efficient, accurate and reliable pulmonary nodule detection for clinical diagnosis.

Machine‐learning‐based Alzheimer's disease dementia score using structural MRI neurodegeneration patterns: Independent validation on ADNI, AIBL, OASIS and MIRIAD

AbstractBackgroundBiomarkers for dementia of Alzheimer's type (DAT) are sought to facilitate accurate prediction of the disease onset, ideally predating the onset of cognitive deterioration. T1‐weighted magnetic resonance imaging (MRI) is a commonly used neuroimaging modality for measuring brain structure in‐vivo, potentially providing information enabling the design of biomarkers for DAT.MethodWe propose a novel biomarker using structural MRI volume‐based features to compute a similarity score for the individual's structural patterns relative to those observed in the DAT group. We employed ensemble learning framework that combines structural features in most discriminative ROIs to create an aggregate measure of neurodegeneration in the brain. This classifier is trained on 423 stable normal control (NC) and 330 DAT subjects, where clinical diagnosis is likely to have the highest certainty, and then validated on several independent datasets for subjects at various stages along the longitudinal trajectory from NC to DAT.ResultIndependent validation on 8834 un‐seen images from ADNI, AIBL, OASIS, MIRIAD Alzheimer's disease (AD) databases showed promising potential to predict the development of DAT depending on the time‐to‐conversion (TTC). Classification performance on stable versus progressive mild cognitive impairment (MCI) groups achieved an AUC of 0.81 for TTC of 6 months, and 0.73 for TTC of up to 7 years, achieving the state‐of‐the‐art results.ConclusionWe propose and validate a novel ensemble‐learning based biomarker score based on 3D topographic patterns of MRI‐observed structural volumetric features. The output score, indicating similarity to patterns seen in DAT, provides an intuitive measure of how closely the individual's brain features resemble the DAT group. This score can be used for assessing the presence of AD structural atrophy patterns in normal aging and MCI stages, as well as monitoring the progression of the individual's brain along the disease course.

MRI Radiomics for Assessment of Molecular Subtype, Pathological Complete Response, and Residual Cancer Burden in Breast Cancer Patients Treated With Neoadjuvant Chemotherapy

There are limited data on pretreatment imaging features that can predict response to neoadjuvant chemotherapy (NAC). To extract volumetric pretreatment MRI radiomics features and assess corresponding associations with breast cancer molecular subtypes, pathological complete response (pCR), and residual cancer burden (RCB) in patients treated with NAC. In this IRB-approved study, clinical and pretreatment MRI data from patients with biopsy-proven breast cancer who received NAC between September 2009 and July 2016 were retrospectively analyzed. Tumors were manually identified and semi-automatically segmented on first postcontrast images. Morphological and three-dimensional textural features were computed, including unfiltered and filtered image data, with spatial scaling factors (SSF) of 2, 4, and 6 mm. Wilcoxon rank-sum tests and area under the receiver operating characteristic curve were used for statistical analysis. Two hundred and fifty nine patients with unilateral breast cancer, including 73 (28.2%) HER2+, 112 (43.2%) luminal, and 74 (28.6%) triple negative breast cancers (TNBC), were included. There was a significant difference in the median volume (p = 0.008), median longest axial tumor diameter (p = 0.009), and median longest volumetric diameter (p = 0.01) among tumor subtypes. There was also a significant difference in minimum signal intensity and entropy among the tumor subtypes with SSF = 4 mm (p = 0.009 and p = 0.02 respectively) and SSF = 6 mm (p = 0.007 and p < 0.001 respectively). Additionally, sphericity (p = 0.04) in HER2+ tumors and entropy with SSF = 2, 4, 6 mm (p = 0.004, 0.02, 0.047 respectively) in luminal tumors were significantly associated with pCR. Multiple features demonstrated significant association (p < 0.05) with pCR in TNBC and with RCB in luminal tumors and TNBC, with standard deviation of intensity with SSF = 6 mm achieving the highest AUC (AUC = 0.734) for pCR in TNBC. MRI radiomics features are associated with different molecular subtypes of breast cancer, pCR, and RCB. These features may be noninvasive imaging biomarkers to identify cancer subtype and predict response to NAC.

Automated feature quantification of Lipiodol as imaging biomarker to predict therapeutic efficacy of conventional transarterial chemoembolization of liver cancer

Conventional transarterial chemoembolization (cTACE) is a guideline-approved image-guided therapy option for liver cancer using the radiopaque drug-carrier and micro-embolic agent Lipiodol, which has been previously established as an imaging biomarker for tumor response. To establish automated quantitative and pattern-based image analysis techniques of Lipiodol deposition on 24 h post-cTACE CT as biomarker for treatment response. The density of Lipiodol deposits in 65 liver lesions was automatically quantified using Hounsfield Unit thresholds. Lipiodol deposition within the tumor was automatically assessed for patterns including homogeneity, sparsity, rim, and peripheral deposition. Lipiodol deposition was correlated with enhancing tumor volume (ETV) on baseline and follow-up MRI. ETV on baseline MRI strongly correlated with Lipiodol deposition on 24 h CT (p < 0.0001), with 8.22% ± 14.59 more Lipiodol in viable than necrotic tumor areas. On follow-up, tumor regions with Lipiodol showed higher rates of ETV reduction than areas without Lipiodol (p = 0.0475) and increasing densities of Lipiodol enhanced this effect. Also, homogeneous (p = 0.0006), non-sparse (p < 0.0001), rim deposition within sparse tumors (p = 0.045), and peripheral deposition (p < 0.0001) of Lipiodol showed improved response. This technical innovation study showed that an automated threshold-based volumetric feature characterization of Lipiodol deposits is feasible and enables practical use of Lipiodol as imaging biomarker for therapeutic efficacy after cTACE.

Is age an additional factor in the treatment of elderly patients with glioblastoma? A new stratification model: an Italian Multicenter Study.

Approximately half of glioblastoma (GBM) cases develop in geriatric patients, and this trend is destined to increase with the aging of the population. The optimal strategy for management of GBM in elderly patients remains controversial. The aim of this study was to assess the role of surgery in the elderly (≥ 65 years old) based on clinical, molecular, and imaging data routinely available in neurosurgical departments and to assess a prognostic survival score that could be helpful in stratifying the prognosis for elderly GBM patients. Clinical, radiological, surgical, and molecular data were retrospectively analyzed in 322 patients with GBM from 9 neurosurgical centers. Univariate and multivariate analyses were performed to identify predictors of survival. A random forest approach (classification and regression tree [CART] analysis) was utilized to create the prognostic survival score. Survival analysis showed that overall survival (OS) was influenced by age as a continuous variable (p = 0.018), MGMT (p = 0.012), extent of resection (EOR; p = 0.002), and preoperative tumor growth pattern (evaluated with the preoperative T1/T2 MRI index; p = 0.002). CART analysis was used to create the prognostic survival score, forming six different survival groups on the basis of tumor volumetric, surgical, and molecular features. Terminal nodes with similar hazard ratios were grouped together to form a final diagram composed of five classes with different OSs (p < 0.0001). EOR was the most robust influencing factor in the algorithm hierarchy, while age appeared at the third node of the CART algorithm. The ability of the prognostic survival score to predict death was determined by a Harrell's c-index of 0.75 (95% CI 0.76-0.81). The CART algorithm provided a promising, thorough, and new clinical prognostic survival score for elderly surgical patients with GBM. The prognostic survival score can be useful to stratify survival risk in elderly GBM patients with different surgical, radiological, and molecular profiles, thus assisting physicians in daily clinical management. The preliminary model, however, requires validation with future prospective investigations. Practical recommendations for clinicians/surgeons would strengthen the quality of the study; e.g., surgery can be considered as a first therapeutic option in the workflow of elderly patients with GBM, especially when the preoperative estimated EOR is greater than 80%.

Value of volumetric and textural analysis in predicting the treatment response in patients with locally advanced rectal cancer.

The aim of this study was to assess the value of baseline 18F-FDG PET/CT in predicting the response to neoadjuvant chemo-radiotherapy (NCRT) in patients with locally advanced rectal cancer (LARC) via the volumetric and texture data obtained from 18F-FDG PET/CT images. In total, 110 patients who had undergone NCRT after initial PET/CT and followed by surgical resection were included in this study. Patients were divided into two groups randomly as a train set (n: 88) and test set (n: 22). Pathological response using three-point tumor regression grade (TRG) and metastatic lymph nodes in PET/CT images were determined. TRG1 were accepted as responders and TRG2-3 as non-responders. Region of interest for the primary tumors was drawn and volumetric features (metabolic tumor volume (MTV) and total lesion glycolysis (TLG)) and texture features were calculated. In train set, the relationship between these features and TRG was investigated with Mann-Whitney U test. Receiver operating curve analysis was performed for features with p < 0.05. Correlation between features were evaluated with Spearman correlation test, features with correlation coefficient < 0.8 were evaluated with the logistic regression analysis for creating a model. The model obtained was tested with a test set that has not been used in modeling before. In train set 32 (36.4%) patients were responders. The rate of visually detected metastatic lymph node at baseline PET/CT was higher in non-responders than responders (71.4% and 46.9%, respectively, p = 0.022). There was a statistically significant difference between TLG, MTV, SHAPE_compacity, NGLDMcoarseness, GLRLM_GLNU, GLRLM_RLNU, GLZLM_LZHGE and GLZLM_GLNU between responders and non-responders. MTV and NGLDMcoarseness demonstrated the most significance (p = 0.011). A multivariate logistic regression analysis that included MTV, coarseness, GLZLM_LZHGE and lymph node metastasis was performed. Multivariate analysis demonstrated MTV and lymph node metastasis were the most meaningful parameters. The model's AUC was calculated as 0.714 (p = 0.001,0.606-0.822, 95% CI). In test set, AUC was determined 0.838 (p = 0.008,0.671-1.000, 95% CI) in discriminating non-responders. Although there were points where textural features were found to be significant, multivariate analysis revealed no diagnostic superiority over MTV in predicting treatment response.In this study, it was thought higher MTV value and metastatic lymph nodes in PET/CT images could be a predictor of low treatment response in patients with LARC.

Diagnosis of Invasive Lung Adenocarcinoma Based on Chest CT Radiomic Features of Part-Solid Pulmonary Nodules: A Multicenter Study.

Background Solid components of part-solid nodules (PSNs) at CT are reflective of invasive adenocarcinoma, but studies describing radiomic features of PSNs and the perinodular region are lacking. Purpose To develop and to validate radiomic signatures diagnosing invasive lung adenocarcinoma in PSNs compared with the Brock, clinical-semantic features, and volumetric models. Materials and Methods This retrospective multicenter study (https://ClinicalTrials.gov, NCT03872362) included 291 patients (median age, 60 years; interquartile range, 55-65 years; 191 women) from January 2013 to October 2017 with 297 PSN lung adenocarcinomas split into training (n = 229) and test (n = 68) data sets. Radiomic features were extracted from the different regions (gross tumor volume [GTV], solid, ground-glass, and perinodular). Random-forest models were trained using clinical-semantic, volumetric, and radiomic features, and an online nodule calculator was used to compute the Brock model. Performances of models were evaluated using standard metrics such as area under the curve (AUC), accuracy, and calibration. The integrated discrimination improvement was applied to assess model performance changes after the addition of perinodular features. Results The radiomics model based on ground-glass and solid features yielded an AUC of 0.98 (95% confidence interval [CI]: 0.96, 1.00) on the test data set, which was significantly higher than the Brock (AUC, 0.83 [95% CI: 0.72, 0.94]; P = .007), clinical-semantic (AUC, 0.90 [95% CI: 0.83, 0.98]; P = .03), volumetric GTV (AUC, 0.87 [95% CI: 0.78, 0.96]; P = .008), and radiomics GTV (AUC, 0.88 [95% CI: 0.80, 0.96]; P = .01) models. It also achieved the best accuracy (93% [95% CI: 84%, 98%]). Both this model and the model with added perinodular features showed good calibration, whereas adding perinodular features did not improve the performance (integrated discrimination improvement, -0.02; P = .56). Conclusion Separating ground-glass and solid CT radiomic features of part-solid nodules was useful in diagnosing the invasiveness of lung adenocarcinoma, yielding a better predictive performance than the Brock, clinical-semantic, volumetric, and radiomics gross tumor volume models. Online supplemental material is available for this article. See also the editorial by Nishino in this issue. Published under a CC BY 4.0 license.

MODELLING OF LAND SURFACE TEMPERATURE USING GRAY LEVEL CO-OCCURRENCE MATRIX AND RANDOM FOREST REGRESSION

Abstract. Modelling of land surface temperature (LST) is conducted to be able to explain the spatial and temporal variations of LST using a set of explanatory variables. LST in a previous study was modelled as a linear function of vegetation cover and built up cover as quantified by the normalized difference vegetation index (NDVI) and the normalized difference built-up index (NDBI), respectively, and other variables, namely, albedo, solar radiation (SR), surface area-volume ratio (SVR), and skyview factor (SVF). SVF requires a digital surface model of sufficient resolution while SVR computation needs 3D volumetric features representing buildings as input. These inputs are typically not readily available. In addition, NDVI and NDBI do not fully describe the spatial variability of vegetation and built-up cover within an LST pixel. In this study, PlanetScope images (3m resolution) were processed to provide soil-adjusted vegetation index (SAVI) and VgNIR Built-up Index (VgNIR-BI) layers. The following gray level co-occurrence matrices (GLCM) were generated from SAVI and VgNIR-BI: Mean, Variance, Homogeneity, Contrast, Dissimilarity, Entropy, Second Moment, and Correlation. Random Forest regression was run for several cases with different combinations of GLCM features and non-GLCM variables. Using GLCM features alone yielded less satisfactory models. However, the use of additional GLCM features in combination with other variables resulted in lower MSE and a slight increase in R2. Considering NDBI, NDVI, SAVI_GLCM_contrast, VgNIR-BI_GLCM_contrast, VgNIR-BI_GLCM_dissimilarity, and SAVI_GLCM_contrast only, the RF model yielded an MSE=1.657 and validation R2=0.822. While this 6-variable model’s performance is slightly less, the need for DSM and 3D building models which are necessary for the generation of SVF and SVR layers is eliminated. Exploratory regression (ER) was also conducted. The best 6-variable ER model (Adj. R2=0.79) consists of SVR, NDBI, NDVI, SAVI_GLCM_second_moment, VgNIR-BI_GLCM_mean, and VgNIR-BI_GLCM_entropy. In comparison, OLS regression using the 6 non-GLCM variables yielded an Adj. R2=0.691. The results of RFR and ER both indicate the value of GLCM features in providing valuable information to the models of LST. LST is best described through a combination of GLCM features describing relatively homogenous areas (i.e., dominant land cover or low-frequency areas) and the more heterogenous areas (i.e., edges or high-frequency areas) and non-GLCM variables.