Quantitative Computed Tomography Texture Analysis Research Articles

Impact of quantitative CT texture analysis on the outcome of CT-guided bone biopsy

Texture analysis can provide new imaging-based biomarkers. Texture analysis derived from computed tomography (CT) might be able to better characterize patients undergoing CT-guided percutaneous bone biopsy. The present study evaluated this and correlated texture features with bioptic outcome in patients undergoing CT-guided bone biopsy. Overall, 123 patients (89 female patients, 72.4 %) were included into the present study. All patients underwent CT-guided percutaneous bone biopsy with an 11 Gauge coaxial needle. Clinical parameters and quantitative imaging features were investigated. Random forest classifier was used to predict a positive biopsy result. Overall, 69 patients had osteolytic metastasis (56.1 %) and 54 had osteoblastic metastasis (43.9 %). The overall positive biopsy rate was 72 %. The developed radiomics model demonstrated a prediction accuracy of a positive biopsy result with an AUC of 0.75 [95 %CI 0.65 – 0.85]. In a subgroup of breast cancer patients, the model achieved an AUC of 0.85 [95 %CI 0.73 – 0.96]. In the subgroup of non-breast cancer patients, the signature achieved an AUC of 0.80 [95 %CI 0.60 – 0.99]. Quantitative CT imaging findings comprised of conventional and texture features can aid to predict the bioptic result of CT-guided bone biopsies. The developed radiomics signature aids in clinical decision-making, and could identify patients at risk for a negative biopsy.

Quantitative CT Texture Analysis of COVID-19 Hospitalized Patients during 3-24-Month Follow-Up and Correlation with Functional Parameters.

To quantitatively evaluate CT lung abnormalities in COVID-19 survivors from the acute phase to 24-month follow-up. Quantitative CT features as predictors of abnormalities' persistence were investigated. Patients who survived COVID-19 were retrospectively enrolled and underwent a chest CT at baseline (T0) and 3 months (T3) after discharge, with pulmonary function tests (PFTs). Patients with residual CT abnormalities repeated the CT at 12 (T12) and 24 (T24) months after discharge. A machine-learning-based software, CALIPER, calculated the CT percentage of the whole lung of normal parenchyma, ground glass (GG), reticulation (Ret), and vascular-related structures (VRSs). Differences (Δ) were calculated between time points. Receiver operating characteristic (ROC) curve analyses were performed to test the baseline parameters as predictors of functional impairment at T3 and of the persistence of CT abnormalities at T12. The cohort included 128 patients at T0, 133 at T3, 61 at T12, and 34 at T24. The GG medians were 8.44%, 0.14%, 0.13% and 0.12% at T0, T3, T12 and T24. The Ret medians were 2.79% at T0 and 0.14% at the following time points. All Δ significantly differed from 0, except between T12 and T24. The GG and VRSs at T0 achieved AUCs of 0.73 as predictors of functional impairment, and area under the curves (AUCs) of 0.71 and 0.72 for the persistence of CT abnormalities at T12. CALIPER accurately quantified the CT changes up to the 24-month follow-up. Resolution mostly occurred at T3, and Ret persisting at T12 was almost unchanged at T24. The baseline parameters were good predictors of functional impairment at T3 and of abnormalities' persistence at T12.

Newly developed artificial intelligence algorithm for COVID-19 pneumonia: utility of quantitative CT texture analysis for prediction of favipiravir treatment effect

PurposeUsing CT findings from a prospective, randomized, open-label multicenter trial of favipiravir treatment of COVID-19 patients, the purpose of this study was to compare the utility of machine learning (ML)-based algorithm with that of CT-determined disease severity score and time from disease onset to CT (i.e., time until CT) in this setting.Materials and methodsFrom March to May 2020, 32 COVID-19 patients underwent initial chest CT before enrollment were evaluated in this study. Eighteen patients were randomized to start favipiravir on day 1 (early treatment group), and 14 patients on day 6 of study participation (late treatment group). In this study, percentages of ground-glass opacity (GGO), reticulation, consolidation, emphysema, honeycomb, and nodular lesion volumes were calculated as quantitative indexes by means of the software, while CT-determined disease severity was also visually scored. Next, univariate and stepwise regression analyses were performed to determine relationships between quantitative indexes and time until CT. Moreover, patient outcomes determined as viral clearance in the first 6 days and duration of fever were compared for those who started therapy within 4, 5, or 6 days as time until CT and those who started later by means of the Kaplan–Meier method followed by Wilcoxon’s signed-rank test.Results% GGO and % consolidation showed significant correlations with time until CT (p < 0.05), and stepwise regression analyses identified both indexes as significant descriptors for time until CT (p < 0.05). When divided all patients between time until CT of 4 days and that of more than 4 days, accuracy of the combined quantitative method (87.5%) was significantly higher than that of the CT disease severity score (62.5%, p = 0.008).ConclusionML-based CT texture analysis is equally or more useful for predicting time until CT for favipiravir treatment on COVID-19 patients than CT disease severity score.

A comprehensive texture feature analysis framework of renal cell carcinoma: pathological, prognostic, and genomic evaluation based on CT images.

We tried to realize accurate pathological classification, assessment of prognosis, and genomic molecular typing of renal cell carcinoma by CT texture feature analysis. To determine whether CT texture features can perform accurate pathological classification and evaluation of prognosis and genomic characteristics in renal cell carcinoma. Patients with renal cell carcinoma from five open-source cohorts were analyzed retrospectively in this study. These data were randomly split to train and test machine learning algorithms to segment the lesion, predict the histological subtype, tumor stage, and pathological grade. Dice coefficient and performance metrics such as accuracy and AUC were calculated to evaluate the segmentation and classification model. Quantitative decomposition of the predictive model was conducted to explore the contribution of each feature. Besides, survival analysis and the statistical correlation between CT texture features, pathological, and genomic signatures were investigated. A total of 569 enhanced CT images of 443 patients (mean age 59.4, 278 males) were included in the analysis. In the segmentation task, the mean dice coefficient was 0.96 for the kidney and 0.88 for the cancer region. For classification of histologic subtype, tumor stage, and pathological grade, the model was on a par with radiologists and the AUC was 0.83 [Formula: see text] 0.1, 0.80 [Formula: see text] 0.1, and 0.77 [Formula: see text] 0.1 at 95% confidence intervals, respectively. Moreover, specific quantitative CT features related to clinical prognosis were identified. A strong statistical correlation (R2 = 0.83) between the feature crosses and genomic characteristics was shown. The structural equation modeling confirmed significant associations between CT features, pathological (β = - 0.75), and molecular subtype (β = - 0.30). The framework illustrates high performance in the pathological classification of renal cell carcinoma. Prognosis and genomic characteristics can be inferred by quantitative image analysis. • The analytical framework exhibits high-performance pathological classification of renal cell carcinoma and is on a par with human radiologists. • Quantitative decomposition of the predictive model shows that specific texture features contribute to histologic subtype and tumor stage classification. • Structural equation modeling shows the associations of genomic characteristics to CT texture features. Overall survival and molecular characteristics can be inferred by quantitative CT texture analysis in renal cell carcinoma.

Quantitative CT texture analysis in predicting PD-L1 expression in locally advanced or metastatic NSCLC patients

PurposeThe assessment of Programmed death-ligand 1 (PD-L1) expression has become a game changer in the treatment of patients with advanced non-small cell lung cancer (NSCLC). We aimed to investigate the ability of Radiomics applied to computed tomography (CT) in predicting PD-L1 expression in patients with advanced NSCLC.MethodsBy applying texture analysis, we retrospectively analyzed 72 patients with advanced NSCLC. The datasets were randomly split into a training cohort (2/3) and a validation cohort (1/3). Forty radiomic features were extracted by manually drawing tumor volumes of interest (VOIs) on baseline contrast-enhanced CT. After selecting features on the training cohort, two predictive models were created using binary logistic regression, one for PD-L1 values ≥ 50% and the other for values between 1 and 49%. The two models were analyzed with ROC curves and tested in the validation cohort.ResultsThe Radiomic Score (Rad-Score) for PD-L1 values ≥ 50%, which consisted of Skewness and Low Gray-Level Zone Emphasis (GLZLM_LGZE), presented a cut-off value of − 0.745 with an area under the curve (AUC) of 0.811 and 0.789 in the training and validation cohort, respectively. The Rad-Score for PD-L1 values between 1 and 49% consisted of Sphericity, Skewness, Conv_Q3 and Gray Level Non-Uniformity (GLZLM_GLNU), showing a cut-off value of 0.111 with AUC of 0.763 and 0.806 in the two population, respectively.ConclusionRad-Scores obtained from CT texture analysis could be useful for predicting PD-L1 expression and guiding the therapeutic choice in patients with advanced NSCLC.

Performance of quantitative CT texture analysis in differentiation of gastric tumors.

To examine the computed tomography (CT) images of patients with a diagnosis of gastric tumor by texture analysis and to investigate its place in differential diagnosis. Contrast enhanced venous phase CT images of 163 patients with pathological diagnosis of gastric adenocarcinoma (n = 125), gastric lymphoma (n = 12) and gastrointestinal stromal tumors (n = 26) were retrospectively analyzed. Pixel size adjustment, gray-level discretization and gray-level normalization procedures were applied as pre-processing steps. Region of interest (ROI) was determined from the axial slice that represented the largest lesion area and a total of 40 texture features were calculated for each patient. Texture features were compared between the tumor subtypes and between adenocarcinoma grades. Statistically significant texture features were combined into a single parameter by logistic regression analysis. The sensitivity and specificity of these features and the combined parameter were measured to differentiate tumor subtypes by receiver-operating characteristic curve (ROC) analysis. Classifications between adenocarcinoma versus lymphoma, adenocarcinoma vs. gastrointestinal stromal tumor (GIST) and well-differentiated adenocarcinoma versus poorly differentiated adenocarcinoma using texture features yielded successful results with high sensitivity (98, 91, 96%, respectively) and specificity (75, 77, 80%, respectively). CT texture analysis is a non-invasive promising method for classifying gastric tumors and predicting gastric adenocarcinoma differentiation.

Effect of phase of enhancement on texture analysis in renal masses evaluated with non-contrast-enhanced, corticomedullary, and nephrographic phase-enhanced CT images.

To compare texture analysis (TA) features of solid renal masses on renal protocol (non-contrast enhanced [NECT], corticomedullary [CM], nephrographic [NG]) CT. A total of 177 consecutive solid renal masses (116 renal cell carcinoma [RCC]; 51 clear cell [cc], 40 papillary, 25 chromophobe, and 61 benign masses; 49 oncocytomas, 12 fat-poor angiomyolipomas) with three-phase CT between 2012 and 2017 were studied. Two blinded radiologists independently assessed tumor heterogeneity (5-point Likert scale) and segmented tumors. TA features (N = 25) were compared between groups and between phases. Accuracy (area under the curve [AUC]) for RCC versus benign and cc-RCC versus other masses was compared. Subjectively, tumor heterogeneity differed between phases (p < 0.01) and between tumors within the same phase (p = 0.03 [NECT] and p < 0.01 [CM, NG]). Inter-observer agreement was moderate to substantial (intraclass correlation coefficient = 0.55-0.73). TA differed in 92.0% (23/25) features between phases (p < 0.05) except for GLNU and f6. More TA features differed significantly on CM (80.0% [20/25]) compared with NG (40.0% [10/25]) and NECT (16.0% [4/25]) (p < 0.01). For RCC versus benign, AUCs of texture features did not differ comparing CM and NG (p > 0.05), but were higher for 20% (5/25) and 28% (7/25) of features comparing CM and NG with NECT (p < 0.05). For cc-RCC versus other, 36% (9/25) and 40% (10/25) features on CM had higher AUCs compared with NECT and NG images (p < 0.05). Texture analysis of renal masses differs, when evaluated subjectively and quantitatively, by phase of CT enhancement. The corticomedullary phase had the highest discriminatory value when comparing masses and for differentiating cc-RCC from other masses. • Subjectively evaluated renal tumor heterogeneity on CT differs by phase of enhancement. • Quantitative CT texture analysis features in renal tumors differ by phases of enhancement with the corticomedullary phase showing the highest number and most significant differences compared with non-contrast-enhanced and nephrographic phase images. • For diagnosis of clear cell RCC, corticomedullary phase texture analysis features had improved accuracy of classification in approximately 40% of features studied compared with non-contrast-enhanced and nephrographic phase images.

Pancreatic Ductal Adenocarcinoma: Machine Learning-Based Quantitative Computed Tomography Texture Analysis For Prediction Of Histopathological Grade.

PurposeTo assess the performance of combining computed tomography (CT) texture analysis with machine learning for discriminating different histopathological grades of pancreatic ductal adenocarcinoma (PDAC).MethodsFrom July 2012 to August 2017, this retrospective study comprised 56 patients with confirmed histopathological PDAC (32 men, 24 women, mean age 64.04±7.82 years) who had undergone preoperative contrast-enhanced CT imaging within 1 month before surgery. Two radiologists blinded to the histopathological outcome independently segmented lesions for quantitative texture analysis. Histogram features, co-occurrence, and run-length texture were calculated. A support-vector machine was constructed to predict the pathological grade of PDAC based on preoperative texture features.ResultsPathological analysis confirmed 37 low-grade PDAC (five well-differentiated/grade I and 32 moderately differentiated/grade II) and 19 high-grade PDAC (19 poorly differentiated/grade III) tumors. There were no significant differences in clinical or biological characteristics between patients with high-grade and low-grade tumors (P>0.05). There were significant differences between low-grade PDAC and high-grade PDAC on nine histogram features, seven run-length features, and two co-occurrence features. Cluster shade was the most important predictor (sensitivity 0.315). Using these texture features, the support-vector machine achieved 86% accuracy, 78% sensitivity, 95% and specificity.ConclusionMachine learning–based CT texture analysis accurately predicted histopathological differentiation grade of PDAC based on preoperative texture features, leading to maximization patient survival and achievement of personalized precision treatment.

Early prediction of clinical outcomes in patients with aneurysmal subarachnoid hemorrhage using computed tomography texture analysis

Radiological evaluation of subarachnoid hemorrhage (SAH) is often subject to interobserver variability. The aim of this study was to retrospectively detect computed tomography (CT) texture parameters in the early postictal state to predict cerebral vasospasm, delayed cerebral ischemia (DCI), and functional outcome in aneurysmal SAH using quantitative CT texture analysis (CTTA) via a commercially available software program and routine CT images. 40 patients with aneurysmal SAH surgically treated at the Keio University Hospital during a four-year period were analyzed. CT texture analyses were performed using a commercially available software program (Synapse Vincent). The following texture parameters of blood clots in the subarachnoid space and cerebral edema were assessed: mean CT value, entropy, skewness, and kurtosis. The mean CT value of blood clots in the subarachnoid space was significantly associated with cerebral vasospasm, DCI, and functional outcome. The mean CT value ≥ 49.64 Hounsfield units (HU) predicted cerebral vasospasm with a sensitivity and specificity of 85.7% and 61.5%, respectively (area under the curve [AUC] = 0.758). The mean CT value ≥ 49.95 HU predicted DCI with a sensitivity and specificity of 100% and 60.6%, respectively (AUC = 0.810). The mean CT value ≥ 53.00 HU predicted poor functional outcome with a sensitivity and specificity of 56.3% and 91.7%, respectively (AUC = 0.747). CTTA using a commercially available software program demonstrated that the mean CT value of clots in the subarachnoid space in the early postictal state could predict vasospasm, DCI, and clinical outcome with a high sensitivity and specificity.

Machine learning-based texture analysis for differentiation of large adrenal cortical tumours on CT

To compare the efficacy of computed tomography (CT) texture analysis and conventional evaluation by radiologists for differentiation between large adrenal adenomas and carcinomas. Quantitative CT texture analysis was used to evaluate 54 histopathologically proven adrenal masses (mean size=5.9 cm; range=4.1-10 cm) from 54 patients referred to Anderson Cancer Center from January 2002 through April 2014. The patient group included 32 women (mean age at mass evaluation=59 years) and 22 men (mean age at mass evaluation=61 years). Adrenal lesions seen on precontrast and venous-phase CT images were labelled by three different readers, and the labels were used to generate intensity- and geometry-based textural features. The textural features and the attenuation values were considered as input values for a random forest-based classifier. Similarly, the adrenal lesions were classified by two different radiologists based on morphological criteria. Prediction accuracy and interobserver agreement were compared. The textural predictive model achieved a mean accuracy of 82%, whereas the mean accuracy for the radiologists was 68.5% (p<0.0001). The interobserver agreements between the predictive model and radiologists 1 and 2 were 0.44 (p<0.0005; 95% confidence interval [CI]: 0.25-0.62) and 0.47 (p<0.0005; 95% CI: 0.28-0.66), respectively. The Dice similarity coefficient between the readers' image labels was 0.875±0.04. CT texture analysis of large adrenal adenomas and carcinomas is likely to improve CT evaluation of adrenal cortical tumours.

Quantitative CT texture analysis for diagnosing systemic sclerosis: Effect of iterative reconstructions and radiation doses.

To test whether texture analysis (TA) can discriminate between Systemic Sclerosis (SSc) and non-SSc patients in computed tomography (CT) with different radiation doses and reconstruction algorithms.In this IRB-approved retrospective study, 85 CT scans at different radiation doses [49 standard dose CT (SDCT) with a volume CT dose index (CTDIvol) of 4.86 ± 2.1 mGy and 36 low-dose (LDCT) with a CTDIvol of 2.5 ± 1.5 mGy] were selected; 61 patients had Ssc ("cases"), and 24 patients had no SSc ("controls"). CT scans were reconstructed with filtered-back projection (FBP) and with sinogram-affirmed iterative reconstruction (SAFIRE) algorithms. 304 TA features were extracted from each manually drawn region-of-interest at 6 pre-defined levels: at the midpoint between lung apices and tracheal carina, at the level of the tracheal carina, and 4 between the carina and pleural recesses. Each TA feature was averaged between these 6 pre-defined levels and was used as input in the machine learning algorithm artificial neural network (ANN) with backpropagation (MultilayerPerceptron) for differentiating between SSc and non-SSc patients.Results were compared regarding correctly/incorrectly classified instances and ROC-AUCs.ANN correctly classified individuals in 93.8% (AUC = 0.981) of FBP-LDCT, in 78.5% (AUC = 0.859) of FBP-SDCT, in 91.1% (AUC = 0.922) of SAFIRE3-LDCT and 75.7% (AUC = 0.815) of SAFIRE3-SDCT, in 88.1% (AUC = 0.929) of SAFIRE5-LDCT and 74% (AUC = 0.815) of SAFIRE5-SDCT.Quantitative TA-based discrimination of CT of SSc patients is possible showing highest discriminatory power in FBP-LDCT images.

Quantitative computed tomography texture analysis: can it improve diagnostic accuracy to differentiate malignant lymph nodes?

Background and objectiveMediastinal lymph node (LN) staging in individuals with non-small-cell lung cancer plays an important role in staging and treatment planning. This study aimed to assess the accuracy of computed tomography (CT) texture analysis (CTTA) in differentiating benign and malignant mediastinal LNs.MethodsPathologically confirmed malignant and benign mediastinal LN samples, obtained using endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA), were retrospectively reviewed, in addition to chest CT and 18-fluorodeoxyglucose (FDG) uptake positron emission tomography (PET) data. For each LN, CTTA was performed using “AVIEW” software (Coreline Soft, Republic of Korea) by drawing a region of interest.ResultsA total of 132 LNs from 80 patients were included and classified into two groups according to pathology results: malignant (n = 61) and benign (n = 71). In EBUS, size > 1 cm, round shape, heterogeneous echogenicity, and presence of coagulation necrosis sign were more prevalent in malignant than in benign LNs; length was the only feature that distinguished the two groups. Among CTTA features, compactness and normalized standard deviation (SD) showed differences between the two groups. The ability to distinguish malignant LNs was higher using high standard uptake value (SUV) on FDG PET/CT (SUVmax ≥ 5) and normalized SD on CTTA (area under the receiver operating characteristic curve 0.739 versus 0.742, respectively); however, normalized SD demonstrated very low sensitivity despite high specificity.ConclusionsCTTA may be helpful in distinguishing between benign and malignant LNs; however, the diagnostic value was not high. Therefore, integrated evaluation with other imaging modalities is needed.

The value of quantitative CT texture analysis in differentiation of angiomyolipoma without visible fat from clear cell renal cell carcinoma on four-phase contrast-enhanced CT images

To investigate the diagnostic performance and usefulness of texture analysis in differentiating angiomyolipoma (AML) without visible fat from clear cell renal cell carcinoma (ccRCC) on four-phase contrast-enhanced computed tomography (CECT). Seventeen patients with AML without visible fat and 50 patients with ccRCC of size ≤4.5 cm who had also undergone preoperative four-phase CECT were included in this study. The histogram, grey-level co-occurrence matrix (GLCM), and grey-level run length matrix (GLRLM) were evaluated. Sequential feature selection (SFS) and support vector machine (SVM) classifier with leave-one-out cross validation were used. Using the SFS and SVM classifiers, five texture features were selected; mean (unenhanced), standard deviation (unenhanced and excretory), cluster prominence (nephrographic), and long-run high grey-level emphasis (corticomedullary). Diagnostic performance of the five selected texture features for all CT phases was as follows: 82% sensitivity, 76% specificity, 85% accuracy, and 85 area under the receiver operating characteristic curve (AUC). In the subgroup analysis, the AUCs of each phase were significantly >0.5 (p<0.05). In the pairwise comparison of AUCs between four phases, there were no significant differences between the four phases except the unenhanced and corticomedullary phases (p=0.015), i.e., the unenhanced phase showed slightly higher AUC than the corticomedullary phase. Texture analysis of small renal masses (≤4.5 cm) on four-phase CECT can accurately differentiate AML without visible fat from ccRCC and showed good diagnostic performance for both the unenhanced and enhanced phases.

733 - Can quantitative computed tomography texture analysis be used for evaluating stage and histologic grade of upper tract urothelial carcinoma?

733 - Can quantitative computed tomography texture analysis be used for evaluating stage and histologic grade of upper tract urothelial carcinoma?

Influence of segmentation margin on machine learning-based high-dimensional quantitative CT texture analysis: a reproducibility study on renal clear cell carcinomas.

To determine the possible influence of segmentation margin on each step (feature reproducibility, selection, and classification) of the machine learning (ML)-based high-dimensional quantitative computed tomography (CT) texture analysis (qCT-TA) of renal clear cell carcinomas (RcCCs). For this retrospective study, 47 patients with RcCC were included from a public database. Two segmentations were obtained by two radiologists for each tumour: (i) contour-focused and (ii) margin shrinkage of 2mm. Texture features were extracted from original, filtered, and transformed CT images. Feature selection was done using a correlation-based algorithm. The ML classifier was k-nearest neighbours. Classifications were performed with and without using synthetic minority over-sampling technique. Reference standard was nuclear grade (low versus high). Intraclass correlation coefficient (ICC), Pearson's correlation coefficient, Wilcoxon signed-ranks test, and McNemar's test were used in the analysis. The segmentation with margin shrinkage of 2mm (772 of 828; 93.2%) yielded more texture features with excellent reproducibility (ICC ≥ 0.9) than contour-focused segmentation (714 of 828; 86.2%), p < 0.0001. The feature selection algorithms resulted in different feature subsets for two segmentation datasets with only one common feature. All ML-based models based on contour-focused segmentation (area under the curve [AUC] range, 0.865-0.984) performed better than those with margin shrinkage of 2mm (AUC range, 0.745-0.887), p < 0.05. Each step of the ML-based high-dimensional qCT-TA was susceptible to a slight change of 2mm in segmentation margin. Despite yielding fewer features with excellent reproducibility, use of the contour-focused segmentation provided better classification performance for distinguishing nuclear grade. • Each step of a machine learning (ML)-based high-dimensional quantitative computed tomography texture analysis (qCT-TA) is sensitive to even a slight change of 2mm in segmentation margin. • Despite yielding fewer texture features with excellent reproducibility, performing the segmentation focusing on the outermost boundary of the tumours provides better classification performance in ML-based qCT-TA of renal clear cell carcinomas for distinguishing nuclear grade. • Findings of an ML-based high-dimensional qCT-TA may not be reproducible in clinical practice even using the same feature selection algorithm and ML classifier unless the possible influence of the segmentation margin is considered.

Radiogenomics in Clear Cell Renal Cell Carcinoma: Machine Learning-Based High-Dimensional Quantitative CT Texture Analysis in Predicting PBRM1 Mutation Status.

The purpose of this study is to evaluate the potential value of machine learning (ML)-based high-dimensional quantitative CT texture analysis in predicting the mutation status of the gene encoding the protein polybromo-1 (PBRM1) in patients with clear cell renal cell carcinoma (RCC). In this retrospective study, 45 patients with clear cell RCC (29 without the PBRM1 mutation and 16 with the PBRM1 mutation) were identified in The Cancer Genome Atlas-Kidney Renal Clear Cell Carcinoma database. To create stable ML models and balanced classes, the data were augmented to a total of 161 labeled segmentations (87 without the PBRM1 mutation and 74 with the PBRM1 mutation) by obtaining three to five different samples per patient. Texture features were extracted from corticomedullary phase contrast-enhanced CT images with the use of an open-source software package for the extraction of radiomic data from medical images. Reproducibility analysis (intraclass correlation) was performed by two radiologists. Attribute selection and model optimization were done using a wrapper-based classifier-specific algorithm with nested cross-validation. ML classifiers were an artificial neural network (ANN) algorithm and a random forest (RF) algorithm. The models were validated using 10-fold cross-validation. The reference standard was the PBRM1 mutation status. The main performance metric was the AUC value. Of 828 extracted texture features, 759 had excellent reproducibility. Using 10 selected features, the ANN algorithm correctly classified 88.2% (142 of 161) of the clear cell RCCs in terms of PBRM1 mutation status (AUC value, 0.925). Using five selected features, the RF algorithm correctly classified 95.0% (153 of 161) of the clear cell RCCs (AUC value, 0.987). Overall, the RF algorithm performed better than the ANN algorithm (z score = -2.677; p = 0.007). ML-based high-dimensional quantitative CT texture analysis might be a feasible and potential method for predicting PBRM1 mutation status in patients with clear cell RCC.

Can quantitative CT texture analysis be used to differentiate subtypes of renal cell carcinoma?

To investigate whether computed tomography (CT) texture analysis (TA) can be used to differentiate non-clear-cell renal cell carcinoma (non-ccRCC) from clear-cell RCC (ccRCC) and classify non-ccRCC subtypes. One hundred ccRCC and 27 non-ccRCC (12 papillary and 15 chromophobe) were analysed. Texture parameters quantified from multiphasic CT images were compared for the objectives. Receiver operating characteristic (ROC) analysis was performed and the area under the ROC curve (AUC) was calculated. The optimal discriminative texture parameters were used to produce support vector machine (SVM) classifiers. Diagnostic accuracy and 10-fold cross-validation was performed. Compared to ccRCC, non-ccRCC had significantly lower mean grey-level intensity (mean), standard deviation (SD), entropy, mean of positive pixels (MPP), and higher kurtosis (p<0.001). A model incorporating SD, entropy, MPP, and kurtosis produced an AUC of 0.94±0.03 with an accuracy of 87% (sensitivity=89%, specificity=92%) to identify non-ccRCC from ccRCC. Compared to chromophobe RCC, papillary RCC had significantly lower mean and MPP (p=0.002). A model incorporating SD, MPP, and skewness resulted in an AUC of 0.96±0.04 with an accuracy of 78% (sensitivity=87%, specificity=92%) to differentiate between papillary and chromophobe RCC. CT TA could potentially be used as a less invasive tool to classify histological subtypes of RCC.

Clear Cell Renal Cell Carcinoma: Machine Learning-Based Quantitative Computed Tomography Texture Analysis for Prediction of Fuhrman Nuclear Grade.

To evaluate the performance of quantitative computed tomography (CT) texture analysis using different machine learning (ML) classifiers for discriminating low and high nuclear grade clear cell renal cell carcinomas (cc-RCCs). This retrospective study included 53 patients with pathologically proven 54 cc-RCCs (31 low-grade [grade 1 or 2]; 23 high-grade [grade 3 or 4]). In one patient, two synchronous cc-RCCs were included in the analysis. Mean age was 57.5 years. Thirty-four (64.1%) patients were male and 19 were female (35.9%). Mean tumour size based on the maximum diameter was 57.4 mm (range, 16-145 mm). Forty patients underwent radical nephrectomy and 13 underwent partial nephrectomy. Following pre-processing steps, two-dimensional CT texture features were extracted using portal-phase contrast-enhanced CT. Reproducibility of texture features was assessed with the intra-class correlation coefficient (ICC). Nested cross-validation with a wrapper-based algorithm was used in feature selection and model optimisation. The ML classifiers were support vector machine (SVM), multilayer perceptron (MLP, a sort of neural network), naïve Bayes, k-nearest neighbours, and random forest. The performance of the classifiers was compared by certain metrics. Among 279 texture features, 241 features with an ICC equal to or higher than 0.80 (excellent reproducibility) were included in the further feature selection process. The best model was created using SVM. The selected subset of features for SVM included five co-occurrence matrix (ICC range, 0.885-0.998), three run-length matrix (ICC range, 0.889-0.992), one gradient (ICC = 0.998), and four Haar wavelet features (ICC range, 0.941-0.997). The overall accuracy, sensitivity (for detecting high-grade cc-RCCs), specificity (for detecting high-grade cc-RCCs), and overall area under the curve of the best model were 85.1%, 91.3%, 80.6%, and 0.860, respectively. The ML-based CT texture analysis can be a useful and promising non-invasive method for prediction of low and high Fuhrman nuclear grade cc-RCCs. • Based on the percutaneous biopsy literature, ML-based CT texture analysis has a comparable predictive performance with percutaneous biopsy. • Highest predictive performance was obtained with use of the SVM. • SVM correctly classified 85.1% of cc-RCCs in terms of nuclear grade, with an AUC of 0.860.

Textural differences between renal cell carcinoma subtypes: Machine learning-based quantitative computed tomography texture analysis with independent external validation

ObjectiveTo develop externally validated, reproducible, and generalizable models for distinguishing three major subtypes of renal cell carcinomas (RCCs) using machine learning-based quantitative computed tomography (CT) texture analysis (qCT-TA). Materials and methodsSixty-eight RCCs were included in this retrospective study for model development and internal validation. Another 26 RCCs were included from public databases (The Cancer Genome Atlas-TCGA) for independent external validation. Following image preparation steps (reconstruction, resampling, normalization, and discretization), 275 texture features were extracted from unenhanced and corticomedullary phase CT images. Feature selection was firstly done with reproducibility analysis by three radiologists, and; then, with a wrapper-based classifier-specific algorithm. A nested cross-validation was performed for feature selection and model optimization. Base classifiers were the artificial neural network (ANN) and support vector machine (SVM). Base classifiers were also combined with three additional algorithms to improve generalizability performance. Classifications were done with the following groups: (i), non-clear cell RCC (non-cc-RCC) versus clear cell RCC (cc-RCC) and (ii), cc-RCC versus papillary cell RCC (pc-RCC) versus chromophobe cell RCC (chc-RCC). Main performance metric for comparisons was the Matthews correlation coefficient (MCC). ResultsNumber of the reproducible features is smaller for the unenhanced images (93 out of 275) compared to the corticomedullary phase images (232 out of 275). Overall performance metrics of the machine learning-based qCT-TA derived from corticomedullary phase images were better than those of unenhanced images. Using corticomedullary phase images, ANN with adaptive boosting algorithm performed best for discrimination of non-cc-RCCs from cc-RCCs (MCC = 0.728) with an external validation accuracy, sensitivity, and specificity of 84.6%, 69.2%, and 100%, respectively. On the other hand, the performance of the machine learning-based qCT-TA is rather poor for distinguishing three major subtypes. The SVM with bagging algorithm performed best for discrimination of pc-RCC from other RCC subtypes (MCC = 0.804) with an external validation accuracy, sensitivity, and specificity of 69.2%, 71.4%, and 100%, respectively. ConclusionsMachine learning-based qCT-TA can distinguish non-cc-RCCs from cc-RCCs with a satisfying performance. On the other hand, the performance of the method for distinguishing three major subtypes is rather poor. Corticomedullary phase CT images provide much more valuable texture parameters than unenhanced images.

Texture analysis of small renal cell carcinomas at MDCT for predicting relevant histologic and protein biomarkers.

To assess CT texture features of small renal cell carcinomas (≤ 4cm) for association with key pathologic features including protein biomarkers. Quantitative CT texture analysis (CTTA) of small renal cancers (≤ 4cm) was performed on non-contrast and portal venous phase abdominal MDCT scans with an ROI drawn at the largest cross-sectional diameter of the tumor using commercially available software. Texture parameters including mean pixel attenuation, the standard deviation (SD) of the pixel distribution histogram, entropy, the mean of positive pixels, the skewness (i.e., asymmetry) of the pixel histogram, kurtosis (i.e., peakness) of the pixel histogram, and the percentage of positive pixels were correlated with pathologic data from surgical resection, including histology and nuclear grade, as well as microarray analysis in a subset (n = 40) including Ki67 index, CRP, and neovascularization (CD105/CD31). Portal venous phase images were availablein 249 patients (105women, 144men;mean age,56.7years) with tumors ≤ 4cm (mean, median, range, ± SD; 2.66, 2.60, 0.3-4.0 ± 0.85cm). CT texture featuresof standard deviation, mean of the positive pixels,and entropy of the pixel histogram were significantly associated with histologic cell type (clear vs. non-clear;p < 0.001). Entropy and mean of the positive pixels also showed an association with nuclear grade, although not statistically significant.In themicroarray analysissubset, kurtosis of the pixel histogram was associated with CD105/CD31(p = 0.05). SD also showed some association with CD 105 positivity (p = 0.02) and CAIX expression (p = 0.01). Non-contrastCT images were availablein 174 patients (72women,102men;mean age,57.5years). Although the association with histology was not as strong as on the portal venous phase, in the subset of patients with microarray data, SD was found to correlate with CRP (p = 0.08), kurtosis with CRP (p = 0.004), CD105/CD31 (p = 0.002),and with Ki 67 index (p < 0.001). CT texture features were significantly associated withimportanthistopathologic features insmallrenalcancers. These non-invasive measures can be performed retrospectively and may provide useful information when determining follow-up and treatment of small renalcancers.