Differentiation of fat-poor angiomyolipoma from clear cell renal cell carcinoma in contrast-enhanced MDCT images using quantitative feature classification

Scientific Status Quo of Small Renal Lesions: Diagnostic Assessment and Radiomics

Background: Small renal masses (SRMs) are defined as contrast-enhanced renal lesions less than or equal to 4 cm in maximal diameter, which can be compatible with stage T1a renal cell carcinomas (RCCs). Currently, 50-61% of all renal tumors are found incidentally. Methods: The characteristics of the lesion influence the choice of the type of management, which include several methods SRM of management, including nephrectomy, partial nephrectomy, ablation, observation, and also stereotactic body radiotherapy. Typical imaging methods available for differentiating benign from malignant renal lesions include ultrasound (US), contrast-enhanced ultrasound (CEUS), computed tomography (CT), and magnetic resonance imaging (MRI). Results: Although ultrasound is the first imaging technique used to detect small renal lesions, it has several limitations. CT is the main and most widely used imaging technique for SRM characterization. The main advantages of MRI compared to CT are the better contrast resolution and tissue characterization, the use of functional imaging sequences, the possibility of performing the examination in patients allergic to iodine-containing contrast medium, and the absence of exposure to ionizing radiation. For a correct evaluation during imaging follow-up, it is necessary to use a reliable method for the assessment of renal lesions, represented by the Bosniak classification system. This classification was initially developed based on contrast-enhanced CT imaging findings, and the 2019 revision proposed the inclusion of MRI features; however, the latest classification has not yet received widespread validation. Conclusions: The use of radiomics in the evaluation of renal masses is an emerging and increasingly central field with several applications such as characterizing renal masses, distinguishing RCC subtypes, monitoring response to targeted therapeutic agents, and prognosis in a metastatic context.

Development and external validation of the multichannel deep learning model based on unenhanced CT for differentiating fat-poor angiomyolipoma from renal cell carcinoma: a two-center retrospective study

PURPOSE

There are undetectable levels of fat in fat-poor angiomyolipoma. Thus, it is often misdiagnosed as renal cell carcinoma. We aimed to develop and evaluate a multichannel deep learning model for differentiating fat-poor angiomyolipoma (fp-AML) from renal cell carcinoma (RCC).

METHODS

This two-center retrospective study included 320 patients from the First Affiliated Hospital of Sun Yat-Sen University (FAHSYSU) and 132 patients from the Sun Yat-Sen University Cancer Center (SYSUCC). Data from patients at FAHSYSU were divided into a development dataset (n = 267) and a hold-out dataset (n = 53). The development dataset was used to obtain the optimal combination of CT modality and input channel. The hold-out dataset and SYSUCC dataset were used for independent internal and external validation, respectively.

RESULTS

In the development phase, models trained on unenhanced CT images performed significantly better than those trained on enhanced CT images based on the fivefold cross-validation. The best patient-level performance, with an average area under the receiver operating characteristic curve (AUC) of 0.951 ± 0.026 (mean ± SD), was achieved using the "unenhanced CT and 7-channel" model, which was finally selected as the optimal model. In the independent internal and external validation, AUCs of 0.966 (95% CI 0.919-1.000) and 0.898 (95% CI 0.824-0.972), respectively, were obtained using the optimal model. In addition, the performance of this model was better on large tumors (≥ 40 mm) in both internal and external validation.

CONCLUSION

The promising results suggest that our multichannel deep learning classifier based on unenhanced whole-tumor CT images is a highly useful tool for differentiating fp-AML from RCC.

CT radiomics for differentiating fat poor angiomyolipoma from clear cell renal cell carcinoma: Systematic review and meta-analysis

PURPOSE

Differentiation of fat-poor angiomyolipoma (fp-AMLs) from renal cell carcinoma (RCC) is often not possible from just visual interpretation of conventional cross-sectional imaging, typically requiring biopsy or surgery for diagnostic confirmation. However, radiomics has the potential to characterize renal masses without the need for invasive procedures. Here, we conducted a systematic review on the accuracy of CT radiomics in distinguishing fp-AMLs from RCCs.

METHODS

We conducted a search using PubMed/MEDLINE, Google Scholar, Cochrane Library, Embase, and Web of Science for studies published from January 2011-2022 that utilized CT radiomics to discriminate between fp-AMLs and RCCs. A random-effects model was applied for the meta-analysis according to the heterogeneity level. Furthermore, subgroup analyses (group 1: RCCs vs. fp-AML, and group 2: ccRCC vs. fp-AML), and quality assessment were also conducted to explore the possible effect of interstudy differences. To evaluate CT radiomics performance, the pooled sensitivity, specificity, and diagnostic odds ratio (DOR) were assessed. This study is registered with PROSPERO (CRD42022311034).

RESULTS

Our literature search identified 10 studies with 1456 lesions in 1437 patients. Pooled sensitivity was 0.779 [95% CI: 0.562-0.907] and 0.817 [95% CI: 0.663-0.910] for groups 1 and 2, respectively. Pooled specificity was 0.933 [95% CI: 0.814-0.978]and 0.926 [95% CI: 0.854-0.964] for groups 1 and 2, respectively. Also, our findings showed higher sensitivity and specificity of 0.858 [95% CI: 0.742-0.927] and 0.886 [95% CI: 0.819-0.930] for detecting ccRCC from fp-AML in the unenhanced phase of CT scan as compared to the corticomedullary and nephrogenic phases of CT scan.

CONCLUSION

This study suggested that radiomic features derived from CT has high sensitivity and specificity in differentiating RCCs vs. fp-AML, particularly in detecting ccRCCs vs. fp-AML. Also, an unenhanced CT scan showed the highest specificity and sensitivity as compared to contrast CT scan phases. Differentiating between fp-AML and RCC often is not possible without biopsy or surgery; radiomics has the potential to obviate these invasive procedures due to its high diagnostic accuracy.

Role of AI and Radiomic Markers in Early Diagnosis of Renal Cancer and Clinical Outcome Prediction: A Brief Review

Globally, renal cancer (RC) is the 10th most common cancer among men and women. The new era of artificial intelligence (AI) and radiomics have allowed the development of AI-based computer-aided diagnostic/prediction (AI-based CAD/CAP) systems, which have shown promise for the diagnosis of RC (i.e., subtyping, grading, and staging) and prediction of clinical outcomes at an early stage. This will absolutely help reduce diagnosis time, enhance diagnostic abilities, reduce invasiveness, and provide guidance for appropriate management procedures to avoid the burden of unresponsive treatment plans. This survey mainly has three primary aims. The first aim is to highlight the most recent technical diagnostic studies developed in the last decade, with their findings and limitations, that have taken the advantages of AI and radiomic markers derived from either computed tomography (CT) or magnetic resonance (MR) images to develop AI-based CAD systems for accurate diagnosis of renal tumors at an early stage. The second aim is to highlight the few studies that have utilized AI and radiomic markers, with their findings and limitations, to predict patients' clinical outcome/treatment response, including possible recurrence after treatment, overall survival, and progression-free survival in patients with renal tumors. The promising findings of the aforementioned studies motivated us to highlight the optimal AI-based radiomic makers that are correlated with the diagnosis of renal tumors and prediction/assessment of patients' clinical outcomes. Finally, we conclude with a discussion and possible future avenues for improving diagnostic and treatment prediction performance.

Ultrasound Image-Based Deep Features and Radiomics for the Discrimination of Small Fat-Poor Angiomyolipoma and Small Renal Cell Carcinoma

We evaluated the performance of ultrasound image-based deep features and radiomics for differentiating small fat-poor angiomyolipoma (sfp-AML) from small renal cell carcinoma (SRCC). This retrospective study included 194 patients with pathologically proven small renal masses (diameter ≤4 cm; 67 in the sfp-AML group and 127 in the SRCC group). We obtained 206 and 364 images from the sfp-AML and SRCC groups with experienced radiologist identification, respectively. We extracted 4024 deep features from the autoencoder neural network and 1497 radiomics features from the Pyradiomics toolbox; the latter included first-order, shape, high-order, Laplacian of Gaussian and Wavelet features. All subjects were allocated to the training and testing sets with a ratio of 3:1 using stratified sampling. The least absolute shrinkage and selection operator (LASSO) regression model was applied to select the most diagnostic features. Support vector machine (SVM) was adopted as the discriminative classifier. An optimal feature subset including 45 deep and 7 radiomics features was screened by the LASSO model. The SVM classifier achieved good performance in discriminating between sfp-AMLs and SRCCs, with areas under the curve (AUCs) of 0.96 and 0.85 in the training and testing sets, respectively. The classifier built using deep and radiomics features can accurately differentiate sfp-AMLs from SRCCs on ultrasound imaging.

Artificial intelligence and radiomics in evaluation of kidney lesions: a comprehensive literature review

Radiomics and artificial intelligence (AI) may increase the differentiation of benign from malignant kidney lesions, differentiation of angiomyolipoma (AML) from renal cell carcinoma (RCC), differentiation of oncocytoma from RCC, differentiation of different subtypes of RCC, to predict Fuhrman grade, to predict gene mutation through molecular biomarkers and to predict treatment response in metastatic RCC undergoing immunotherapy. Neural networks analyze imaging data. Statistical, geometrical, textural features derived are giving quantitative data of contour, internal heterogeneity and gray zone features of lesions. A comprehensive literature review was performed, until July 2022. Studies investigating the diagnostic value of radiomics in differentiation of renal lesions, grade prediction, gene alterations, molecular biomarkers and ongoing clinical trials have been analyzed. The application of AI and radiomics could lead to improved sensitivity, specificity, accuracy in detecting and differentiating between renal lesions. Standardization of scanner protocols will improve preoperative differentiation between benign, low-risk cancers and clinically significant renal cancers and holds the premises to enhance the diagnostic ability of imaging tools to characterize renal lesions.

Comparative Analysis for the Distinction of Chromophobe Renal Cell Carcinoma from Renal Oncocytoma in Computed Tomography Imaging Using Machine Learning Radiomics Analysis

Background: ChRCC and RO are two types of rarely occurring renal tumors that are difficult to distinguish from one another based on morphological features alone. They differ in prognosis, with ChRCC capable of progressing and metastasizing, but RO is benign. This means discrimination of the two tumors is of crucial importance. Objectives: The purpose of this research was to develop and comprehensively evaluate predictive models that can discriminate between ChRCC and RO tumors using Computed Tomography (CT) scans and ML-Radiomics texture analysis methods. Methods: Data were obtained from 78 pathologically confirmed renal masses, scanned at two institutions. Data from the two institutions were combined to form a third set resulting in three data cohorts, i.e., cohort 1, 2 and combined. Contrast-enhanced scans were used and the axial cross-sectional slices of each tumor were extracted from the 3D data using a semi-automatic segmentation technique for both 2D and 3D scans. Radiomics features were extracted before and after applying filters and the dimensions of the radiomic features reduced using the least absolute shrinkage and selection operator (LASSO) method. Synthetic minority oversampling technique (SMOTE) was applied to avoid class imbalance. Five ML algorithms were used to train models for predictive classification and evaluated using 5-fold cross-validation. Results: The number of selected features with good model performance was 20, 40 and 6 for cohorts 1, 2 and combined, respectively. The best model performance in cohorts 1, 2 and combined had an excellent Area Under the Curve (AUC) of 1.00 ± 0.000, 1.00 ± 0.000 and 0.87 ± 0.073, respectively. Conclusions: ML-based radiomics signatures are potentially useful for distinguishing ChRCC and RO tumors, with a reliable level of performance for both 2D and 3D scanning.

Predictive Value of CT-Based Radiomics in Distinguishing Renal Angiomyolipomas with Minimal Fat from Other Renal Tumors

Objectives

This study is aimed at determining whether CT-based radiomics models can help differentiate renal angiomyolipomas with minimal fat (AMLmf) from other solid renal tumors.

Methods

This retrospective study included 58 patients with a postoperative pathologically confirmed AMLmf (observation group) and 140 patients with other common renal tumors (control group). Non-contrast-enhanced CT and contrast-enhanced CT data were evaluated. Radiomics features were extracted from manually delineated volume of interest (VOIs). The least absolute shrinkage and selection operator (LASSO) regression was used for feature screening. Five classifiers, including logistic regression, multilayer perceptron (MLP), support vector machine (SVM), k-nearest neighbor (KNN), and logistic regression (LR), were used, with leave-out validation (128 training, 60 testing). The diagnostic performance of the classifier was evaluated and compared by receiver operating characteristic curve (ROC) analysis.

Results

Among the 1029 extracted features, prediction models of AMLmf were composed, by 2, 10, 4, and 9 selected features for precontrast phase (PCP), corticomedullary phase (CMP), nephrographic phase (NP), and excretory phase (EP), respectively. Models of CMP and NP achieved adequate performance after using MLP classifier, with prediction accuracy of 0.767 (AUC 0.85, sensitivity 0.76, and specificity 0.78) and 0.783 (AUC 0.83, sensitivity 0.79, and specificity 0.78), respectively. MLP model of features selected from the combination of the all features had the best diagnostic performance (accuracy 0.8500, sensitivity 0.8095, specificity 0.9444, and AUC 0.9193).

Conclusions

Radiomics features may help to distinguish benign AMLmf from common malignant kidney masses, which may contribute to the selection of interventions for renal tumors.

Novel Imaging Methods for Renal Mass Characterization: A Collaborative Review

CONTEXT

The incidental detection of localized renal masses has been rising steadily, but a significant proportion of these tumors are benign or indolent and, in most cases, do not require treatment. At the present time, a majority of patients with an incidentally detected renal tumor undergo treatment for the presumption of cancer, leading to a significant number of unnecessary surgical interventions that can result in complications including loss of renal function. Thus, there exists a clinical need for improved tools to aid in the pretreatment characterization of renal tumors to inform patient management.

OBJECTIVE

To systematically review the evidence on noninvasive, imaging-based tools for solid renal mass characterization.

EVIDENCE ACQUISITION

The MEDLINE database was systematically searched for relevant studies on novel imaging techniques and interpretative tools for the characterization of solid renal masses, published in the past 10 yr.

EVIDENCE SYNTHESIS

Over the past decade, several novel imaging tools have offered promise for the improved characterization of indeterminate renal masses. Technologies of particular note include multiparametric magnetic resonance imaging of the kidney, molecular imaging with targeted radiopharmaceutical agents, and use of radiomics as well as artificial intelligence to enhance the interpretation of imaging studies. Among these, 99mTc-sestamibi single photon emission computed tomography/computed tomography (CT) for the identification of benign renal oncocytomas and hybrid oncocytic chromophobe tumors, and positron emission tomography/CT imaging with radiolabeled girentuximab for the identification of clear cell renal cell carcinoma, are likely to be closest to implementation in clinical practice.

CONCLUSIONS

A number of novel imaging tools stand poised to aid in the noninvasive characterization of indeterminate renal masses. In the future, these tools may aid in patient management by providing a comprehensive virtual biopsy, complete with information on tumor histology, underlying molecular abnormalities, and ultimately disease prognosis.

PATIENT SUMMARY

Not all renal tumors require treatment, as a significant proportion are either benign or have limited metastatic potential. Several innovative imaging tools have shown promise for their ability to improve the characterization of renal tumors and provide guidance in terms of patient management.

Hyperechoic Renal Masses: Differentiation of Angiomyolipomas from Renal Cell Carcinomas using Tumor Size and Ultrasound Radiomics

A retrospective single-center study was performed to assess the performance of ultrasound image-based texture analysis in differentiating angiomyolipoma (AML) from renal cell carcinoma (RCC) on incidental hyperechoic renal lesions. Ultrasound reports of patients from 2012 to 2017 were queried, and those with a hyperechoic renal mass <5 cm in diameter with further imaging characterization and/or pathological correlation were included. Quantitative texture analysis was performed using a model including 18 texture features. Univariate logistic regression was used to identify texture variables differing significantly between AML and RCC, and the performance of the model was measured using the area under the receiver operating characteristic (ROC) curve. One hundred thirty hyperechoic renal masses in 127 patients characterized as RCCs (25 [19%]) and AMLs (105 [81%]) were included. Size (odds ratio [OR] = 0.12, 95% confidence interval [CI]: 0.04-0.43, p < 0.001) and 4 of 18 texture features, including entropy (OR = 0.09, 95% CI: 0.01-0.81, p = 0.03), gray-level non-uniformity (OR = 0.12, 95% CI: 0.02-0.72, p = 0.02), long-run emphasis (OR = 0.49, 95% CI: 0.27-0.91, p = 0.02) and run-length non-uniformity (OR = 2.18, 95% CI: 1.14-4.16, p = 0.02) were able to differentiate AMLs from RCCs. The area under the ROC curve for the performance of the model, including texture features and size, was 0.945 (p < 0.001). Ultrasound image-based textural analysis enables differentiation of hyperechoic RCCs from AMLs with high accuracy, which improves further when combined with tumor size.

Improving segmentation and classification of renal tumors in small sample 3D CT images using transfer learning with convolutional neural networks

PURPOSE

Computed tomography (CT) images can display internal organs of patients and are particularly suitable for preoperative surgical diagnoses. The increasing demands for computer-aided systems in recent years have facilitated the development of many automated algorithms, especially deep convolutional neural networks, to segment organs and tumors or identify diseases from CT images. However, performances of some systems are highly affected by the amount of training data, while the sizes of medical image data sets, especially three-dimensional (3D) data sets, are usually small. This condition limits the application of deep learning.

METHODS

In this study, given a practical clinical data set that has 3D CT images of 20 patients with renal carcinoma, we designed a pipeline employing transfer learning to alleviate the detrimental effect of the small sample size. A dual-channel fine segmentation network (FS-Net) was constructed to segment kidney and tumor regions, with 210 publicly available 3D images from a competition employed during the training phase. We also built discriminative classifiers to classify the benign and malignant tumors based on the segmented regions, where both handcrafted and deep features were tested.

RESULTS

Our experimental results showed that the Dice values of segmented kidney and tumor regions were 0.9662 and 0.7685, respectively, which were better than those of state-of-the-art methods. The classification model using radiomics features can classify most of the tumors correctly.

CONCLUSIONS

The designed FS-Net was demonstrated to be more effective than simply fine-tuning on the practical small size data set given that the model can borrow knowledge from large auxiliary data without diluting the signal in primary data. For the small data set, radiomics features outperformed deep features in the classification of benign and malignant tumors. This work highlights the importance of architecture design in transfer learning, and the proposed pipeline is anticipated to provide a reference and inspiration for small data analysis.

Deep Convolutional Neural Network Based Analysis of Liver Tissues Using Computed Tomography Images

Liver disease is one of the most prominent causes of the increase in the death rate worldwide. These death rates can be reduced by early liver diagnosis. Computed tomography (CT) is a method for the analysis of liver images in clinical practice. To analyze a large number of liver images, radiologists face problems that sometimes lead to the wrong classifications of liver diseases, eventually resulting in severe conditions, such as liver cancer. Thus, a machine-learning-based method is needed to classify such problems based on their texture features. This paper suggests two different kinds of algorithms to address this challenging task of liver disease classification. Our first method, which is based on conventional machine learning, uses texture features for classification. This method uses conventional machine learning through automated texture analysis and supervised machine learning methods. For this purpose, 3000 clinically verified CT image samples were obtained from 71 patients. Appropriate image classes belonging to the same disease were trained to confirm the abnormalities in liver tissues by using supervised learning methods. Our proposed method correctly quantified asymmetric patterns in CT images using machine learning. We evaluated the effectiveness of the feature vector with the K Nearest Neighbor (KNN), Naive Bayes (NB), Support Vector Machine (SVM), and Random Forest (RF) classifiers. The second algorithm proposes a semantic segmentation model for liver disease identification. Our model is based on semantic image segmentation (SIS) using a convolutional neural network (CNN). The model encodes high-density maps through a specific guided attention method. The trained model classifies CT images into five different categories of various diseases. The compelling results obtained confirm the effectiveness of the proposed model. The study concludes that abnormalities in the human liver could be discriminated and diagnosed by texture analysis techniques, which may also assist radiologists and medical physicists in predicting the severity and proliferation of abnormalities in liver diseases.

CT classification model of pancreatic serous cystic neoplasms and mucinous cystic neoplasms based on a deep neural network

BACKGROUND

At present, numerous challenges exist in the diagnosis of pancreatic SCNs and MCNs. After the emergence of artificial intelligence (AI), many radiomics research methods have been applied to the identification of pancreatic SCNs and MCNs.

PURPOSE

A deep neural network (DNN) model termed Multi-channel-Multiclassifier-Random Forest-ResNet (MMRF-ResNet) was constructed to provide an objective CT imaging basis for differential diagnosis between pancreatic serous cystic neoplasms (SCNs) and mucinous cystic neoplasms (MCNs).

MATERIALS AND METHODS

This study is a retrospective analysis of pancreatic unenhanced and enhanced CT images in 63 patients with pancreatic SCNs and 47 patients with MCNs (3 of which were mucinous cystadenocarcinoma) confirmed by pathology from December 2010 to August 2016. Different image segmented methods (single-channel manual outline ROI image and multi-channel image), feature extraction methods (wavelet, LBP, HOG, GLCM, Gabor, ResNet, and AlexNet) and classifiers (KNN, Softmax, Bayes, random forest classifier, and Majority Voting rule method) are used to classify the nature of the lesion in each CT image (SCNs/MCNs). Then, the comparisons of classification results were made based on sensitivity, specificity, precision, accuracy, F1 score, and area under the receiver operating characteristic curve (AUC), with pathological results serving as the gold standard.

RESULTS

Multi-channel-ResNet (AUC 0.98) was superior to Manual-ResNet (AUC 0.91).CT image characteristics of lesions extracted by ResNet are more representative than wavelet, LBP, HOG, GLCM, Gabor, and AlexNet. Compared to the use of three classifiers alone and Majority Voting rule method, the use of the MMRF-ResNet model exhibits a better evaluation effect (AUC 0.96) for the classification of the pancreatic SCNs and MCNs.

CONCLUSION

The CT image classification model MMRF-ResNet is an effective method to distinguish between pancreatic SCNs and MCNs.

Differentiating Benign from Malignant Renal Tumors Using T2- and Diffusion-Weighted Images: A Comparison of Deep Learning and Radiomics Models Versus Assessment from Radiologists

BACKGROUND

Differentiating benign from malignant renal tumors is important for selection of the most effective treatment.

PURPOSE

To develop magnetic resonance imaging (MRI)-based deep learning (DL) models for differentiation of benign and malignant renal tumors and to compare their discrimination performance with the performance of radiomics models and assessment by radiologists.

STUDY TYPE

Retrospective.

POPULATION

A total of 217 patients were randomly assigned to a training cohort (N = 173) or a testing cohort (N = 44).

FIELD STRENGTH/SEQUENCE

Diffusion-weighted imaging (DWI) and fast spin-echo sequence T2-weighted imaging (T2WI) at 3.0T.

ASSESSMENT

A radiologist manually labeled the region of interest (ROI) on each image. Three DL models using ResNet-18 architecture and three radiomics models using random forest were developed using T2WI alone, DWI alone, and a combination of the two image sets to discriminate between benign and malignant renal tumors. The diagnostic performance of two radiologists was assessed based on professional experience. We also compared the performance of each model and the radiologists.

STATISTICAL TESTS

The area under the receiver operating characteristic (ROC) curve (AUC) was used to assess the performance of each model and the radiologists. P < 0.05 indicated statistical significance.

RESULTS

The AUC of the DL models based on T2WI, DWI, and the combination was 0.906, 0.846, and 0.925 in the testing cohorts, respectively. The AUC of the combination DL model was significantly better than that of the models based on individual sequences (0.925 > 0.906, 0.925 > 0.846). The AUC of the radiomics models based on T2WI, DWI, and the combination was 0.824, 0.742, and 0.826 in the testing cohorts, respectively. The AUC of two radiologists was 0.724 and 0.667 in the testing cohorts.

CONCLUSION

Thus, the MRI-based DL model is useful for differentiating benign from malignant renal tumors in clinic, and the DL model based on T2WI + DWI had the best performance.

LEVEL OF EVIDENCE

3 TECHNICAL EFFICACY STAGE: 2.

Radiomics to better characterize small renal masses

PURPOSE

Radiomics is a specific field of medical research that uses programmable recognition tools to extract objective information from standard images to combine with clinical data, with the aim of improving diagnostic, prognostic, and predictive accuracy beyond standard visual interpretation. We performed a narrative review of radiomic applications that may support improved characterization of small renal masses (SRM). The main focus of the review was to identify and discuss methods which may accurately differentiate benign from malignant renal masses, specifically between renal cell carcinoma (RCC) subtypes and from angiomyolipoma without visible fat (fat-poor AML) and oncocytoma. Furthermore, prediction of grade, sarcomatoid features, and gene mutations would be of importance in terms of potential clinical utility in prognostic stratification and selecting personalised patient management strategies.

METHODS

A detailed search of original articles was performed using the PubMed-MEDLINE database until 20 September 2020 to identify the English literature relevant to radiomics applications in renal tumour assessment. In total, 42 articles were included in the analysis in 3 main categories related to SRM: prediction of benign versus malignant SRM, subtypes, and nuclear grade, and other features of aggressiveness.

CONCLUSION

Overall, studies reported the superiority of radiomics over expert radiological assessment, but were mainly of retrospective design and therefore of low-quality evidence. However, it is clear that radiomics is an attractive modality that has the potential to improve the non-invasive diagnostic accuracy of SRM imaging and prediction of its natural behaviour. Further prospective validation studies of radiomics are needed to augment management algorithms of SRM.

Prediction of chemotherapy response in breast cancer patients at pre-treatment using second derivative texture of CT images and machine learning

•

Textural and second derivative textural features of CT images can be used in conjunction with machine learning models to predict breast cancer response to chemotherapy prior to the start of treatment.

•

The proposed predictive model separates the patients at pre-treatment into two cohorts (responders/non-responders) with significantly different survival.

•

The proposed methodology is a step forward towards the precision oncology paradigm for breast cancer patients.

Although neoadjuvant chemotherapy (NAC) is a crucial component of treatment for locally advanced breast cancer (LABC), only about 70% of patients respond to it. Effective adjustment of NAC for individual patients can significantly improve survival rates of those resistant to standard regimens. Thus, the early prediction of NAC outcome is of great importance in facilitating a personalized paradigm for breast cancer therapeutics. In this study, quantitative computed tomography (qCT) parametric imaging in conjunction with machine learning techniques were investigated to predict LABC tumor response to NAC. Textural and second derivative textural (SDT) features of CT images of 72 patients diagnosed with LABC were analysed before the initiation of NAC to quantify intra-tumor heterogeneity. These quantitative features were processed through a correlation-based feature reduction followed by a sequential feature selection with a bootstrap 0.632+ area under the receiver operating characteristic (ROC) curve (AUC0.632+) criterion. The best feature subset consisted of a combination of one textural and three SDT features. Using these features, an AdaBoost decision tree could predict the patient response with a cross-validated AUC0.632+ accuracy, sensitivity and specificity of 0.88, 85%, 88% and 75%, respectively. This study demonstrates, for the first time, that a combination of textural and SDT features of CT images can be used to predict breast cancer response NAC prior to the start of treatment which can potentially facilitate early therapy adjustments.

Radiomics analysis of contrast-enhanced CT for classification of hepatic focal lesions in colorectal cancer patients: its limitations compared to radiologists

OBJECTIVE

To evaluate diagnostic performance of a radiomics model for classifying hepatic cyst, hemangioma, and metastasis in patients with colorectal cancer (CRC) from portal-phase abdominopelvic CT images.

METHODS

This retrospective study included 502 CRC patients who underwent contrast-enhanced CT and contrast-enhanced liver MRI between January 2005 and December 2010. Portal-phase CT images of training (n = 386) and validation (n = 116) cohorts were used to develop a radiomics model for differentiating three classes of liver lesions. Among multiple handcrafted features, the feature selection was performed using ReliefF method, and random forest classifiers were used to train the selected features. Diagnostic performance of the developed model was compared with that of four radiologists. A subgroup analysis was conducted based on lesion size.

RESULTS

The radiomics model demonstrated significantly lower overall and hemangioma- and metastasis-specific polytomous discrimination index (PDI) (overall, 0.8037; hemangioma-specific, 0.6653; metastasis-specific, 0.8027) than the radiologists (overall, 0.9622-0.9680; hemangioma-specific, 0.9452-0.9630; metastasis-specific, 0.9511-0.9869). For subgroup analysis, the PDI of the radiomics model was different according to the lesion size (< 10 mm, 0.6486; ≥ 10 mm, 0.8264) while that of the radiologists was relatively maintained. For classifying metastasis from benign lesions, the radiomics model showed excellent diagnostic performance, with an accuracy of 84.36% and an AUC of 0.9426.

CONCLUSION

Albeit inferior to the radiologists, the radiomics model achieved substantial diagnostic performance when differentiating hepatic lesions from portal-phase CT images of CRC patients. This model was limited particularly to classifying hemangiomas and subcentimeter lesions.

KEY POINTS

• Albeit inferior to the radiologists, the radiomics model could differentiate cyst, hemangioma, and metastasis with substantial diagnostic performance using portal-phase CT images of colorectal cancer patients. • The radiomics model demonstrated limitations especially in classifying hemangiomas and subcentimeter liver lesions.

A Non-Invasive Scoring System to Differential Diagnosis of Clear Cell Renal Cell Carcinoma (ccRCC) From Renal Angiomyolipoma Without Visible Fat (RAML-wvf) Based on CT Features

Background

Renal angiomyolipoma without visible fat (RAML-wvf) and clear cell renal cell carcinoma (ccRCC) have many overlapping features on imaging, which poses a challenge to radiologists. This study aimed to create a scoring system to distinguish ccRCC from RAML-wvf using computed tomography imaging.

Methods

A total of 202 patients from 2011 to 2019 that were confirmed by pathology with ccRCC (n=123) or RAML (n=79) were retrospectively analyzed by dividing them randomly into a training cohort (n=142) and a validation cohort (n=60). A model was established using logistic regression and weighted to be a scoring system. ROC, AUC, cut-off point, and calibration analyses were performed. The scoring system was divided into three ranges for convenience in clinical evaluations, and the diagnostic probability of ccRCC was calculated.

Results

Four independent risk factors are included in the system: 1) presence of a pseudocapsule, 2) a heterogeneous tumor parenchyma in pre-enhancement scanning, 3) a non-high CT attenuation in pre-enhancement scanning, and 4) a heterogeneous enhancement in CMP. The prediction accuracy had an ROC of 0.978 (95% CI, 0.956–0.999; P=0.011), similar to the primary model (ROC, 0.977; 95% CI, 0.954–1.000; P=0.012). A sensitivity of 91.4% and a specificity of 93.9% were achieved using 4.5 points as the cutoff value. Validation showed a good result (ROC, 0.922; 95% CI, 0.854–0.991, P=0.035). The number of patients with ccRCC in the three ranges (0 to <2 points; 2–4 points; >4 to ≤11 points) significantly increased with increasing scores.

Conclusion

This scoring system is convenient for distinguishing between ccRCC and RAML-wvf using four computed tomography features.

A convention-radiomics CT nomogram for differentiating fat-poor angiomyolipoma from clear cell renal cell carcinoma

This study aimed to construct convention-radiomics CT nomogram containing conventional CT characteristics and radiomics signature for distinguishing fat-poor angiomyolipoma (fp-AML) from clear-cell renal cell carcinoma (ccRCC). 29 fp-AML and 110 ccRCC patients were enrolled and underwent CT examinations in this study. The radiomics-only logistic model was constructed with selected radiomics features by the analysis of variance (ANOVA)/Mann–Whitney (MW), correlation analysis, and Least Absolute Shrinkage and Selection Operator (LASSO), and the radiomics score (rad-score) was computed. The convention-radiomics logistic model based on independent conventional CT risk factors and rad-score was constructed for differentiating. Then the relevant nomogram was developed. Receiver operation characteristic (ROC) curves were calculated to quantify the accuracy for distinguishing. The rad-score of ccRCC was smaller than that of fp-AML. The convention-radioimics logistic model was constructed containing variables of enhancement pattern, V_UP, and rad-score. To the entire cohort, the area under the curve (AUC) of convention-radiomics model (0.968 [95% CI 0.923–0.990]) was higher than that of radiomics-only model (0.958 [95% CI 0.910–0.985]). Our study indicated that convention-radiomics CT nomogram including conventional CT risk factors and radiomics signature exhibited better performance in distinguishing fp-AML from ccRCC.

Automated segmentation of kidney and renal mass and automated detection of renal mass in CT urography using 3D U-Net-based deep convolutional neural network

OBJECTIVES

To develop a 3D U-Net-based deep learning model for automated segmentation of kidney and renal mass, and detection of renal mass in corticomedullary phase of computed tomography urography (CTU).

METHODS

Data on 882 kidneys obtained from CTU data of 441 patients with renal mass were used to learn and evaluate the deep learning model. The CTU data of 35 patients with small renal tumors (diameter ≤ 1.5 cm) were used for additional testing. The ground truth data for the kidney, renal tumor, and cyst were manually annotated on corticomedullary phase images of CTU. The proposed segmentation model for kidney and renal mass was constructed based on a 3D U-Net. The segmentation accuracy was evaluated through the Dice similarity coefficient (DSC). The volume of the maximum 3D volume of interest of renal tumor and cyst in the predicted segmentation by the model was used as an identification indicator, while the detection performance of the model was evaluated by the area under the receiver operation characteristic curve.

RESULTS

The proposed model showed a high accuracy in segmentation of kidney and renal tumor, with average DSC of 0.973 and 0.844, respectively. It performed moderately in the renal cyst segmentation, with an average DSC of 0.536 in the test set. Also, this model showed good performance in detecting renal tumor and cyst.

CONCLUSIONS

The proposed automated segmentation and detection model based on 3D U-Net shows promising results for the segmentation of kidney and renal tumor, and the detection of renal tumor and cyst.

KEY POINTS

• The segmentation model based on 3D U-Net showed high accuracy in segmentation of kidney and renal neoplasm, and good detection performance of renal neoplasm and cyst in corticomedullary phase of CTU. • The segmentation model based on 3D U-Net is a fully automated aided diagnostic tool that could be used to reduce the workload of radiologists and improve the accuracy of diagnosis. • The segmentation model based on 3D U-Net would be helpful to provide quantitative information for diagnosis, treatment, surgical planning, etc.

Review of Value of CT Texture Analysis and Machine Learning in Differentiating Fat-Poor Renal Angiomyolipoma from Renal Cell Carcinoma

The diagnosis of patients with suspected angiomyolipoma relies on the detection of abundant macroscopic intralesional fat, which is always of no use to differentiate fat-poor angiomyolipoma (fp-AML) from renal cell carcinoma and diagnosis of fp-AML excessively depends on individual experience. Texture analysis was proven to be a potentially useful biomarker for distinguishing between benign and malignant tumors because of its capability of providing objective and quantitative assessment of lesions by analyzing features that are not visible to the human eye. This review aimed to summarize the literature on the use of texture analysis to diagnose patients with fat-poor angiomyolipoma vs those with renal cell carcinoma and to evaluate its current application, limitations, and future challenges in order to avoid unnecessary surgical resection.

Radiomic Features and Machine Learning for the Discrimination of Renal Tumor Histological Subtypes: A Pragmatic Study Using Clinical-Routine Computed Tomography

Simple Summary

This study evaluates how advanced image analyses (radiomic features) and machine learning algorithms can help to distinguish subtypes of kidney tumors in computed tomography (CT) images, which is important for further patient treatment. For 201 patients, the image analyses showed a moderate performance, but robustly performed across various imaging centers and even in cases with suboptimal image quality. In particular, distinguishing one specific subtype of kidney tumor (oncocytomas) from other subtypes proves to be challenging. The algorithms presented in this study can help in the clinical decision-making process for kidney tumor patients, for example, to decide whether to perform kidney surgery or not.

Abstract

This study evaluates the diagnostic performance of radiomic features and machine learning algorithms for renal tumor subtype assessment in venous computed tomography (CT) studies from clinical routine. Patients undergoing surgical resection and histopathological assessment of renal tumors at a tertiary referral center between 2012 and 2019 were included. Preoperative venous-phase CTs from multiple referring imaging centers were segmented, and standardized radiomic features extracted. After preprocessing, class imbalance handling, and feature selection, machine learning algorithms were used to predict renal tumor subtypes using 10-fold cross validation, assessed as multiclass area under the curve (AUC). In total, n = 201 patients were included (73.7% male; mean age 66 ± 11 years), with n = 131 clear cell renal cell carcinomas (ccRCC), n = 29 papillary RCC, n = 11 chromophobe RCC, n = 16 oncocytomas, and n = 14 angiomyolipomas (AML). An extreme gradient boosting algorithm demonstrated the highest accuracy (multiclass area under the curve (AUC) = 0.72). The worst discrimination was evident for oncocytomas vs. AML and oncocytomas vs. chromophobe RCC (AUC = 0.55 and AUC = 0.45, respectively). In sensitivity analyses excluding oncocytomas, a random forest algorithm showed the highest accuracy, with multiclass AUC = 0.78. Radiomic feature analyses from venous-phase CT acquired in clinical practice with subsequent machine learning can discriminate renal tumor subtypes with moderate accuracy. The classification of oncocytomas seems to be the most complex with the lowest accuracy.

Value of radiomics in differential diagnosis of chromophobe renal cell carcinoma and renal oncocytoma

Purpose

To explore the value of CT-enhanced quantitative features combined with machine learning for differential diagnosis of renal chromophobe cell carcinoma (chRCC) and renal oncocytoma (RO).

Methods

Sixty-one cases of renal tumors (chRCC = 44; RO = 17) that were pathologically confirmed at our hospital between 2008 and 2018 were retrospectively analyzed. All patients had undergone preoperative enhanced CT scans including the corticomedullary (CMP), nephrographic (NP), and excretory phases (EP) of contrast enhancement. Volumes of interest (VOIs), including lesions on the images, were manually delineated using the RadCloud platform. A LASSO regression algorithm was used to screen the image features extracted from all VOIs. Five machine learning classifications were trained to distinguish chRCC from RO by using a fivefold cross-validation strategy. The performance of the classifier was mainly evaluated by areas under the receiver operating characteristic (ROC) curve and accuracy.

Results

In total, 1029 features were extracted from CMP, NP, and EP. The LASSO regression algorithm was used to screen out the four, four, and six best features, respectively, and eight features were selected when CMP and NP were combined. All five classifiers had good diagnostic performance, with area under the curve (AUC) values greater than 0.850, and support vector machine (SVM) classifier showed a diagnostic accuracy of 0.945 (AUC 0.964 ± 0.054; sensitivity 0.999; specificity 0.800), showing the best performance.

Conclusions

Accurate preoperative differential diagnosis of chRCC and RO can be facilitated by a combination of CT-enhanced quantitative features and machine learning.

Artificial Intelligence in Renal Mass Characterization: A Systematic Review of Methodologic Items Related to Modeling, Performance Evaluation, Clinical Utility, and Transparency

OBJECTIVE. The objective of our study was to systematically review the literature about the application of artificial intelligence (AI) to renal mass characterization with a focus on the methodologic quality items. MATERIALS AND METHODS. A systematic literature search was conducted using PubMed to identify original research studies about the application of AI to renal mass characterization. Besides baseline study characteristics, a total of 15 methodologic quality items were extracted and evaluated on the basis of the following four main categories: modeling, performance evaluation, clinical utility, and transparency items. The qualitative synthesis was presented using descriptive statistics with an accompanying narrative. RESULTS. Thirty studies were included in this systematic review. Overall, the methodologic quality items were mostly favorable for modeling (63%) and performance evaluation (63%). Even so, the studies (57%) more frequently constructed their work on nonrobust features. Furthermore, only a few studies (10%) had a generalizability assessment with independent or external validation. The studies were mostly unsuccessful in terms of clinical utility evaluation (89%) and transparency (97%) items. For clinical utility, the interesting findings were lack of comparisons with both radiologists' evaluation (87%) and traditional models (70%) in most of the studies. For transparency, most studies (97%) did not share their data with the public. CONCLUSION. To bring AI-based renal mass characterization from research to practice, future studies need to improve modeling and performance evaluation strategies and pay attention to clinical utility and transparency issues.

MR texture analysis in differentiating renal cell carcinoma from lipid-poor angiomyolipoma and oncocytoma

OBJECTIVES

To assess the utility of magnetic resonance texture analysis (MRTA) in differentiating renal cell carcinoma (RCC) from lipid-poor angiomyolipoma (lpAML) and oncocytoma.

METHODS

After ethical approval, 42 patients with 54 masses (34 RCC, 14 lpAML and six oncocytomas) who underwent MRI on a 1.5 T scanner (Avanto, Siemens, Erlangen, Germany) between January 2011 and December 2012 were retrospectively included in the study. MRTA was performed on the TexRAD research software (Feedback Plc., Cambridge, UK) using free-hand polygonal region of interest (ROI) drawn on the maximum cross-sectional area of the tumor to generate six first-order statistical parameters. The Mann-Whitney U test was used to look for any statically significant difference. The receiver operating characteristic (ROC) curve analysis was done to select the parameter with the highest class separation capacity [area under the curve (AUC)] for each MRI sequence.

RESULTS

Several texture parameters on MRI showed high-class separation capacity (AUC > 0.8) in differentiating RCC from lpAML and oncocytoma. The best performing parameter in differentiating RCC from lpAML was mean of positive pixels (MPP) at SSF 2 (AUC: 0.891) on DWI b500. In differentiating RCC from oncocytoma, the best parameter was mean at SSF 0 (AUC: 0.935) on DWI b1000.

CONCLUSIONS

MRTA could potentially serve as a useful non-invasive tool for differentiating RCC from lpAML and oncocytoma.

ADVANCES IN KNOWLEDGE

There is limited literature addressing the role of MRTA in differentiating RCC from lpAML and oncocytoma. Our study demonstrated several texture parameters which were useful in this regard.

Machine Learning in Radiomic Renal Mass Characterization: Fundamentals, Applications, Challenges, and Future Directions

OBJECTIVE. The purpose of this study is to provide an overview of the traditional machine learning (ML)-based and deep learning-based radiomic approaches, with focus placed on renal mass characterization. CONCLUSION. ML currently has a very low barrier to entry into general medical practice because of the availability of many open-source, free, and easy-to-use toolboxes. Therefore, it should not be surprising to see its related applications in renal mass characterization. A wider picture of the previous works might be beneficial to move this field forward.

A priori prediction of tumour response to neoadjuvant chemotherapy in breast cancer patients using quantitative CT and machine learning

Response to Neoadjuvant chemotherapy (NAC) has demonstrated a high correlation to survival in locally advanced breast cancer (LABC) patients. An early prediction of responsiveness to NAC could facilitate treatment adjustments on an individual patient basis that would be expected to improve treatment outcomes and patient survival. This study investigated, for the first time, the efficacy of quantitative computed tomography (qCT) parametric imaging to characterize intra-tumour heterogeneity and its application in predicting tumour response to NAC in LABC patients. Textural analyses were performed on CT images acquired from 72 patients before the start of chemotherapy to determine quantitative features of intra-tumour heterogeneity. The best feature subset for response prediction was selected through a sequential feature selection with bootstrap 0.632 + area under the receiver operating characteristic (ROC) curve (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{A}\mathrm{U}\mathrm{C}}_{0.632+}$$\end{document}AUC0.632+) as a performance criterion. Several classifiers were evaluated for response prediction using the selected feature subset. Amongst the applied classifiers an Adaboost decision tree provided the best results with cross-validated \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{A}\mathrm{U}\mathrm{C}}_{0.632+}$$\end{document}AUC0.632+, accuracy, sensitivity and specificity of 0.89, 84%, 80% and 88%, respectively. The promising results obtained in this study demonstrate the potential of the proposed biomarkers to be used as predictors of LABC tumour response to NAC prior to the start of treatment.

Automated classification of solid renal masses on contrast-enhanced computed tomography images using convolutional neural network with decision fusion

OBJECTIVES

To develop a deep learning-based method for automated classification of renal cell carcinoma (RCC) from benign solid renal masses using contrast-enhanced computed tomography (CECT) images.

METHODS

This institutional review board-approved retrospective study evaluated CECT in 315 patients with 77 benign (57 oncocytomas, and 20 fat-poor angiomyolipoma) and 238 malignant (RCC: 123 clear cell, 69 papillary, and 46 chromophobe subtypes) tumors identified consecutively between 2015 and 2017. We employed a decision fusion-based model to aggregate slice level predictions determined by convolutional neural network (CNN) via a majority voting system to evaluate renal masses on CECT. The CNN-based model was trained using 7023 slices with renal masses manually extracted from CECT images of 155 patients, cropped automatically around kidneys, and augmented artificially. We also examined the fully automated approach for renal mass evaluation on CECT. Moreover, a 3D CNN was trained and tested using the same datasets and the obtained results were compared with those acquired from slice-wise algorithms.

RESULTS

For differentiation of RCC versus benign solid masses, the semi-automated majority voting-based CNN algorithm achieved accuracy, precision, and recall of 83.75%, 89.05%, and 91.73% using 160 test cases, respectively. Fully automated pipeline yielded accuracy, precision, and recall of 77.36%, 85.92%, and 87.22% on the same test cases, respectively. 3D CNN reported accuracy, precision, and recall of 79.24%, 90.32%, and 84.21% using 160 test cases, respectively.

CONCLUSIONS

A semi-automated majority voting CNN-based methodology enabled accurate classification of RCC from benign neoplasms among solid renal masses on CECT.

KEY POINTS

• Our proposed semi-automated majority voting CNN-based algorithm achieved accuracy of 83.75% for the diagnosis of RCC from benign solid renal masses on CECT images. • A fully automated CNN-based methodology classified solid renal masses with moderate accuracy of 77.36% using the same test images. • Employing 3D CNN-based methodology yielded slightly lower accuracy for renal mass classification compared with the semi- automated 2D CNN-based algorithm (79.24%).

Deep Learning to Distinguish Benign from Malignant Renal Lesions Based on Routine MR Imaging

PURPOSE

With increasing incidence of renal mass, it is important to make a pretreatment differentiation between benign renal mass and malignant tumor. We aimed to develop a deep learning model that distinguishes benign renal tumors from renal cell carcinoma (RCC) by applying a residual convolutional neural network (ResNet) on routine MR imaging.

EXPERIMENTAL DESIGN

Preoperative MR images (T2-weighted and T1-postcontrast sequences) of 1,162 renal lesions definitely diagnosed on pathology or imaging in a multicenter cohort were divided into training, validation, and test sets (70:20:10 split). An ensemble model based on ResNet was built combining clinical variables and T1C and T2WI MR images using a bagging classifier to predict renal tumor pathology. Final model performance was compared with expert interpretation and the most optimized radiomics model.

RESULTS

Among the 1,162 renal lesions, 655 were malignant and 507 were benign. Compared with a baseline zero rule algorithm, the ensemble deep learning model had a statistically significant higher test accuracy (0.70 vs. 0.56, P = 0.004). Compared with all experts averaged, the ensemble deep learning model had higher test accuracy (0.70 vs. 0.60, P = 0.053), sensitivity (0.92 vs. 0.80, P = 0.017), and specificity (0.41 vs. 0.35, P = 0.450). Compared with the radiomics model, the ensemble deep learning model had higher test accuracy (0.70 vs. 0.62, P = 0.081), sensitivity (0.92 vs. 0.79, P = 0.012), and specificity (0.41 vs. 0.39, P = 0.770).

CONCLUSIONS

Deep learning can noninvasively distinguish benign renal tumors from RCC using conventional MR imaging in a multi-institutional dataset with good accuracy, sensitivity, and specificity comparable with experts and radiomics.

Quantitative Analysis of Multiphase Contrast-Enhanced CT Images: A Pilot Study of Preoperative Prediction of Fat-Poor Angiomyolipoma and Renal Cell Carcinoma

OBJECTIVE. The objective of our study was to preoperatively predict fat-poor angiomyolipoma (fp-AML) and renal cell carcinoma (RCC) by conducting quantitative analysis of contrast-enhanced CT images. MATERIALS AND METHODS. One hundred fifteen patients with a pathologic diagnosis of fp-AML or RCC from a single institution were randomly allocated into a train set (tumor size: mean ± SD, 4.50 ± 2.62 cm) and test set (tumor size: 4.32 ± 2.73 cm) after data augmentation. High-dimensional histogram-based features, texture-based features, and Laws features were first extracted from CT images and were then combined as different combinations sets to construct a logistic prediction model based on the least absolute shrinkage and selection operator procedure for the prediction of fp-AML and RCC. Prediction performances were assessed by classification accuracy, area under the ROC curve (AUC), positive predictive value, negative predictive value, true-positive rate, and false-positive rate (FPR). In addition, we also investigated the effects of different gray-scales of quantitative features on prediction performances. RESULTS. The following combination sets of features achieved satisfying performances in the test set: histogram-based features (mean AUC = 0.8492, mean classification accuracy = 91.01%); histogram-based features and texture-based features (mean AUC = 0.9244, mean classification accuracy = 91.81%); histogram-based features and Laws features (mean AUC = 0.8546, mean classification accuracy = 88.76%); and histogram-based features, texture-based features, and Laws features (mean AUC = 0.8925, mean classification accuracy = 90.36%). The different quantitative gray-scales did not have an obvious effect on prediction performances. CONCLUSION. The integration of histogram-based features with texture-based features and Laws features provided a potential biomarker for the preoperative diagnosis of fp-AML and RCC. The accurate diagnosis of benign or malignant renal masses would help to make the clinical decision for radical surgery or close follow-up.

Radiomics of Renal Masses: Systematic Review of Reproducibility and Validation Strategies

OBJECTIVE. The purpose of this study was to systematically review the radiomics literature on renal mass characterization in terms of reproducibility and validation strategies. MATERIALS AND METHODS. With use of PubMed and Google Scholar, a systematic literature search was performed to identify original research papers assessing the value of radiomics in characterization of renal masses. The data items were extracted on the basis of three main categories: baseline study characteristics, radiomic feature reproducibility strategies, and statistical model validation strategies. RESULTS. After screening and application of the eligibility criteria, a total of 41 papers were included in the study. Almost one-half of the papers (19 [46%]) presented at least one reproducibility analysis. Segmentation variability (18 [44%]) was the main theme of the analyses, outnumbering image acquisition or processing (3 [7%]). No single paper considered slice selection bias. The most commonly used statistical tool for analysis was intraclass correlation coefficient (14 of 19 [74%]), with no consensus on the threshold or cutoff values. Approximately one-half of the papers (22 [54%]) used at least one validation method, with a predominance of internal validation techniques (20 [49%]). The most frequently used internal validation technique was k-fold cross-validation (12 [29%]). Independent or external validation was used in only three papers (7%). CONCLUSION. Workflow characteristics described in the radiomics literature about renal mass characterization are heterogeneous. To bring radiomics from a mere research area to clinical use, the field needs many more papers that consider the reproducibility of radiomic features and include independent or external validation in their workflow.

A CT-based radiomics nomogram for differentiation of renal angiomyolipoma without visible fat from homogeneous clear cell renal cell carcinoma

OBJECTIVES

To develop and validate a radiomics nomogram for preoperative differentiating renal angiomyolipoma without visible fat (AML.wovf) from homogeneous clear cell renal cell carcinoma (hm-ccRCC).

METHODS

Ninety-nine patients with AML.wovf (n = 36) and hm-ccRCC (n = 63) were divided into a training set (n = 80) and a validation set (n = 19). Radiomics features were extracted from corticomedullary phase and nephrographic phase CT images. A radiomics signature was constructed and a radiomics score (Rad-score) was calculated. Demographics and CT findings were assessed to build a clinical factors model. Combined with the Rad-score and independent clinical factors, a radiomics nomogram was constructed. Nomogram performance was assessed with respect to calibration, discrimination, and clinical usefulness.

RESULTS

Fourteen features were used to build the radiomics signature. The radiomics signature showed good discrimination in the training set (AUC [area under the curve], 0.879; 95%; confidence interval [CI], 0.793-0.966) and the validation set (AUC, 0.846; 95% CI, 0.643-1.000). The radiomics nomogram showed good calibration and discrimination in the training set (AUC, 0.896; 95% CI, 0.810-0.983) and the validation set (AUC, 0.949; 95% CI, 0.856-1.000) and showed better discrimination capability (p < 0.05) compared with the clinical factor model (AUC, 0.788; 95% CI, 0.683-0.893) in the training set. Decision curve analysis demonstrated the nomogram outperformed the clinical factors model and radiomics signature in terms of clinical usefulness.

CONCLUSIONS

The CT-based radiomics nomogram, a noninvasive preoperative prediction tool that incorporates the Rad-score and clinical factors, shows favorable predictive efficacy for differentiating AML.wovf from hm-ccRCC, which might assist clinicians in tailoring precise therapy.

KEY POINTS

• Differential diagnosis between AML.wovf and hm-ccRCC is rather difficult by conventional imaging modalities. • A radiomics nomogram integrated with the radiomics signature, demographics, and CT findings facilitates differentiation of AML.wovf from hm-ccRCC with improved diagnostic efficacy. • The CT-based radiomics nomogram might spare unnecessary surgery for AML.wovf.

Radiomics of small renal masses on multiphasic CT: accuracy of machine learning-based classification models for the differentiation of renal cell carcinoma and angiomyolipoma without visible fat

OBJECTIVE

To investigate the discriminative capabilities of different machine learning-based classification models on the differentiation of small (< 4 cm) renal angiomyolipoma without visible fat (AMLwvf) and renal cell carcinoma (RCC).

METHODS

This study retrospectively collected 163 patients with pathologically proven small renal mass, including 118 RCC and 45 AMLwvf patients. Target region of interest (ROI) delineation, followed by texture feature extraction, was performed on a representative slice with the largest lesion area on each phase of the four-phase CT images. Fifteen concatenations of the four-phasic features were fed into 224 classification models (built with 8 classifiers and 28 feature selection methods), classification performances of the 3360 resultant discriminative models were compared, and the top-ranked features were analyzed.

RESULTS

Image features extracted from the unenhanced phase (UP) CT image demonstrated dominant classification performances over features from other three phases. The two discriminative models "SVM + t_score" and "SVM + relief" achieved the highest classification AUC of 0.90. The 10 top-ranked features from UP included 1 shape feature, 5 first-order statistics features, and 4 texture features, where the shape feature and the first-order statistics features showed superior discriminative capabilities in differentiating RCC vs. AMLwvf through the t-SNE visualization.

CONCLUSION

Image features extracted from UP are sufficient to generate accurate differentiation between AMLwvf and RCC using machine learning-based classification model.

KEY POINTS

• Radiomics extracted from unenhanced CT are sufficient to accurately differentiate angiomyolipoma without visible fat and renal cell carcinoma using machine learning-based classification model. • The highest discriminative models achieved an AUC of 0.90 and were based on the analysis of unenhanced CT, alone or in association with images obtained at the nephrographic phase. • Features related to shape and to histogram analysis (first-order statistics) showed superior discrimination compared with gray-level distribution of the image (second-order statistics, commonly called texture features).

A review of the principles of texture analysis and its role in imaging of genitourinary neoplasms

Advances in the management of genitourinary neoplasms have resulted in a trend towards providing patients with personalized care. Texture analysis of medical images, is one of the tools that is being explored to provide information such as detection and characterization of tumors, determining their aggressiveness including grade and metastatic potential and for prediction of survival rates and risk of recurrence. In this article we review the basic principles of texture analysis and then detail its current role in imaging of individual neoplasms of the genitourinary system.

Deep learning and radiomics: the utility of Google TensorFlow™ Inception in classifying clear cell renal cell carcinoma and oncocytoma on multiphasic CT

PURPOSE

Currently, all solid enhancing renal masses without microscopic fat are considered malignant until proven otherwise and there is substantial overlap in the imaging findings of benign and malignant renal masses, particularly between clear cell RCC (ccRCC) and benign oncocytoma (ONC). Radiomics has attracted increased attention for its utility in pre-operative work-up on routine clinical images. Radiomics based approaches have converted medical images into mineable data and identified prognostic imaging signatures that machine learning algorithms can use to construct predictive models by learning the decision boundaries of the underlying data distribution. The TensorFlow™ framework from Google is a state-of-the-art open-source software library that can be used for training deep learning neural networks for performing machine learning tasks. The purpose of this study was to investigate the diagnostic value and feasibility of a deep learning-based renal lesion classifier using open-source Google TensorFlow™ Inception in differentiating ccRCC from ONC on routine four-phase MDCT in patients with pathologically confirmed renal masses.

METHODS

With institutional review board approval for this 1996 Health Insurance Portability and Accountability Act compliant retrospective study and a waiver of informed consent, we queried our institution's pathology, clinical, and radiology databases for histologically proven cases of ccRCC and ONC obtained between January 2000 and January 2016 scanned with a an intravenous contrast-enhanced four-phase renal mass protocol (unenhanced (UN), corticomedullary (CM), nephrographic (NP), and excretory (EX) phases). To extract features to be used for the machine learning model, the entire renal mass was contoured in the axial plane in each of the four phases, resulting in a 3D volume of interest (VOI) representative of the entire renal mass. We investigated thirteen different approaches to convert the acquired VOI data into a set of images that adequately represented each tumor which was used to train the final layer of the neural network model. Training was performed over 4000 iterations. In each iteration, 90% of the data were designated as training data and the remaining 10% served as validation data and a leave-one-out cross-validation scheme was implemented. Accuracy, sensitivity, specificity, positive (PPV) and negative predictive (NPV) values, and CIs were calculated for the classification of the thirteen processing modes.

RESULTS

We analyzed 179 consecutive patients with 179 lesions (128 ccRCC and 51 ONC). The ccRCC cohort had a mean size of 3.8 cm (range 0.8-14.6 cm) and the ONC cohort had a mean lesion size of 3.9 cm (range 1.0-13.1 cm). The highest specificity and PPV (52.9% and 80.3%, respectively) were achieved in the EX phase when we analyzed the single mid-slice of the tumor in the axial, coronal and sagittal plane, and when we increased the number of mid-slices of the tumor to three, with an accuracy of 75.4%, which also increased the sensitivity to 88.3% and the PPV to 79.6%. Using the entire tumor volume also showed that classification performance was best in the EX phase with an accuracy of 74.4%, a sensitivity of 85.8% and a PPV of 80.1%. When the entire tumor volume, plus mid-slices from all phases and all planes presented as tiled images, were submitted to the final layer of the neural network we achieved a PPV of 82.5%.

CONCLUSIONS

The best classification result was obtained in the EX phase among the thirteen classification methods tested. Our proof of concept study is the first step towards understanding the utility of machine learning in the differentiation of ccRCC from ONC on routine CT images. We hope this could lead to future investigation into the development of a multivariate machine learning model which may augment our ability to accurately predict renal lesion histology on imaging.

Reliability of Single-Slice-Based 2D CT Texture Analysis of Renal Masses: Influence of Intra- and Interobserver Manual Segmentation Variability on Radiomic Feature Reproducibility

OBJECTIVE. The objective of our study was to investigate the potential influence of intra- and interobserver manual segmentation variability on the reliability of single-slice-based 2D CT texture analysis of renal masses. MATERIALS AND METHODS. For this retrospective study, 30 patients with clear cell renal cell carcinoma were included from a public database. For intra- and interobserver analyses, three radiologists with varying degrees of experience segmented the tumors from unenhanced CT and corticomedullary phase contrast-enhanced CT (CECT) in different sessions. Each radiologist was blind to the image slices selected by other radiologists and him- or herself in the previous session. A total of 744 texture features were extracted from original, filtered, and transformed images. The intraclass correlation coefficient was used for reliability analysis. RESULTS. In the intraobserver analysis, the rates of features with good to excellent reliability were 84.4-92.2% for unenhanced CT and 85.5-93.1% for CECT. Considering the mean rates of unenhanced CT and CECT, having high experience resulted in better reliability rates in terms of the intraobserver analysis. In the interobserver analysis, the rates were 76.7% for unenhanced CT and 84.9% for CECT. The gray-level cooccurrence matrix and first-order feature groups yielded higher good to excellent reliability rates on both unenhanced CT and CECT. Filtered and transformed images resulted in more features with good to excellent reliability than the original images did on both unenhanced CT and CECT. CONCLUSION. Single-slice-based 2D CT texture analysis of renal masses is sensitive to intra- and interobserver manual segmentation variability. Therefore, it may lead to nonreproducible results in radiomic analysis unless a reliability analysis is considered in the workflow.

A computer-aided diagnosis scheme of breast lesion classification using GLGLM and shape features: Combined-view and multi-classifiers

PURPOSE

To address high false-positive results of FFDM issue, we make the first effort to develop a computer-aided diagnosis (CAD) scheme to analyze and distinguish breast lesions.

METHOD

The breast lesion regions were first segmented and depicted on FFDM images from 106 patients. In this work, 11 gray-level gap-length matrix texture features and 12 shape features were extracted form craniocaudal view and mediolateral oblique view, and then Student's t-test, Fisher-score and Relief-F were introduced to select features. We also investigated the effect of three factors, i.e., discretisation, selection methods and classifier methods, of the classification performance via analysis of variance. Finally, a classification model was constructed. Spearman's correlation coefficient analysis was conducted to assess the internal relevance of features.

RESULTS

The proposed scheme using Student's t-test achieved an area under the receiver operating characteristic curve (AUC) value of 0.923 at 512 bins. The AUC values are 0.884, 0.867, 0.874 and 0.901 for the low gray-level gaps emphasis (LGGE), solidity, extent, and the combined set, respectively. Solidity and extent depicts the correlation coefficient of 0.86 (P < 0.05).

CONCLUSIONS

We present a new CAD scheme based on the contribution of the significant factors. The experimental results demonstrate that the presented scheme can be used to successfully distinguish breast carcinoma lesions and benign fibroadenoma lesions in our FFDM dataset and the MIAS dataset, which may provide a CAD method to assist radiologists in diagnosing and interpreting screening mammograms. Moreover, we found that LGGE, solidity and extent features show great potential for breast lesion classification.

Differentiation of Predominantly Solid Enhancing Lipid-Poor Renal Cell Masses by Use of Contrast-Enhanced CT: Evaluating the Role of Texture in Tumor Subtyping

OBJECTIVE

The purpose of this study was to assess the accuracy of a panel of texture features extracted from clinical CT in differentiating benign from malignant solid enhancing lipid-poor renal masses.

MATERIALS AND METHODS

In a retrospective case-control study of 174 patients with predominantly solid nonmacroscopic fat-containing enhancing renal masses, 129 cases of malignant renal cell carcinoma were found, including clear cell, papillary, and chromophobe subtypes. Benign renal masses-oncocytoma and lipid-poor angiomyolipoma-were found in 45 patients. Whole-lesion ROIs were manually segmented and coregistered from the standard-of-care multiphase contrast-enhanced CT (CECT) scans of these patients. Pathologic diagnosis of all tumors was obtained after surgical resection. CECT images of the renal masses were used as inputs to a CECT texture analysis panel comprising 31 texture metrics derived with six texture methods. Stepwise logistic regression analysis was used to select the best predictor among all candidate predictors from each of the texture methods, and their performance was quantified by AUC.

RESULTS

Among the texture predictors aiding renal mass subtyping were entropy, entropy of fast-Fourier transform magnitude, mean, uniformity, information measure of correlation 2, and sum of averages. These metrics had AUC values ranging from good (0.80) to excellent (0.98) across the various subtype comparisons. The overall CECT-based tumor texture model had an AUC of 0.87 (p < 0.05) for differentiating benign from malignant renal masses.

CONCLUSION

The CT texture statistical model studied was accurate for differentiating benign from malignant solid enhancing lipid-poor renal masses.

Angiomyolipoma of the Kidneys: Current Perspectives and Challenges in Diagnostic Imaging and Image-Guided Therapy

Angiomyolipomas (AML) are benign tumors of the kidneys frequently encountered in radiologic practice in large tertiary centers. In comparison to renal cell carcinomas (RCC), AML are seldom treated unless they are large, undergo malignant transformation or develop complications like acute hemorrhage. The common garden triphasic (classic) AML is an easy diagnosis, however, some variants lack macroscopic fat in which case the radiologic differentiation from RCC becomes challenging. Several imaging features, both qualitative and quantitative, have been described in differentiating the 2 entities. Although minimal fat AML is not entirely a radiologic diagnosis, the suspicion raised on imaging necessitates sampling and potentially avoids an unwanted surgery. Recently a new variant, epitheloid AML has been described which often has atypical imaging features and is at a higher risk for malignant transformation. Apart from the diagnosis, the radiologist also needs to convey information regarding nephrometric scores which help in surgical decision-making. Recently, more and more AMLs are managed with selective arterial embolization and percutaneous ablation, both of which are associated with less morbidity when compared to surgery. The purpose of this article is to review the imaging and pathologic features of classic AML as well as the differentiation of minimal fat AML from RCC. In addition, an overview of nephrometric scoring and image-guided interventions is also provided.

Machine learning-based quantitative texture analysis of CT images of small renal masses: Differentiation of angiomyolipoma without visible fat from renal cell carcinoma

OBJECTIVE

To evaluate the diagnostic performance of machine-learning based quantitative texture analysis of CT images to differentiate small (≤ 4 cm) angiomyolipoma without visible fat (AMLwvf) from renal cell carcinoma (RCC).

METHODS

This single-institutional retrospective study included 58 patients with pathologically proven small renal mass (17 in AMLwvf and 41 in RCC groups). Texture features were extracted from the largest possible tumorous regions of interest (ROIs) by manual segmentation in preoperative three-phase CT images. Interobserver reliability and the Mann-Whitney U test were applied to select features preliminarily. Then support vector machine with recursive feature elimination (SVM-RFE) and synthetic minority oversampling technique (SMOTE) were adopted to establish discriminative classifiers, and the performance of classifiers was assessed.

RESULTS

Of the 42 extracted features, 16 candidate features showed significant intergroup differences (P < 0.05) and had good interobserver agreement. An optimal feature subset including 11 features was further selected by the SVM-RFE method. The SVM-RFE+SMOTE classifier achieved the best performance in discriminating between small AMLwvf and RCC, with the highest accuracy, sensitivity, specificity and AUC of 93.9 %, 87.8 %, 100 % and 0.955, respectively.

CONCLUSION

Machine learning analysis of CT texture features can facilitate the accurate differentiation of small AMLwvf from RCC.

KEY POINTS

• Although conventional CT is useful for diagnosis of SRMs, it has limitations. • Machine-learning based CT texture analysis facilitate differentiation of small AMLwvf from RCC. • The highest accuracy of SVM-RFE+SMOTE classifier reached 93.9 %. • Texture analysis combined with machine-learning methods might spare unnecessary surgery for AMLwvf.