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

The influence of manual segmentation strategies and different phases selection on machine learning-based computed tomography in renal tumors: a systematic review and meta-analysis

BACKGROUND

Delineating the region/volume of interest (ROI/VOI) and selecting the phases are of importance in developing machine learning (ML). The results will change when choosing different methods of drawing the ROI/VOI and selecting different phases. However, there is no related standard for delineating the ROI/VOI and selecting the phases in renal tumors to develop ML based on computed tomography (CT).

METHODS

The PubMed and Web of Science were searched for related studies published until March 1, 2023. Inclusion criteria were studies that developed ML models in renal tumors from CT images. And the binary diagnostic accuracy data were extracted to obtain the outcomes, such as sensitivity (SE), specificity (SP), accuracy (ACC), and area under the curve (AUC).

RESULTS

Twenty-three papers were included in the meta-analysis with a pooled SE of 87% (95% CI 85-88%), SP of 82% (95% CI 79-85%), and AUC of 91% (95% CI 89-93%) in phases; a pooled SE of 82% (95% CI 80-84%), SP of 85% (95% CI 83-86%), and AUC of 90% (95% CI 88-93%) in phases combined with delineating strategies, respectively. In all different combinations, the contour-focused and single phase produce the highest AUC of 93% (95% CI 90-95%). In subgroup analyses (sample size, year of publication, and geographical distribution), the performance was acceptable on phases and phases combined strategies.

CONCLUSIONS

To explore the effect of manual segmentation strategies and different phases selection on ML-based CT, we find that the method of single phase (CMP or NP) combined with contour-focused was considered a better strategy compared to the other strategies.

Quantitative Measurement on Contrast-Enhanced CT Distinguishes Small Clear Cell Renal Cell Carcinoma From Benign Renal Tumors: A Multicenter Study

RATIONALE AND OBJECTIVES

To evaluate the potential of quantitative measurements on contrast-enhanced CT (CECT) in differentiating small (≤4 cm) clear cell renal cell carcinoma (ccRCC) from benign renal tumors, including fat-poor angiomyolipoma (fpAML) and renal oncocytoma (RO).

MATERIALS AND METHODS

244 patients with pathologically confirmed ccRCC (n = 184) and benign renal tumors (fpAML, n = 50; RO, n = 10) were randomly assigned into training cohort (n = 193) and test cohort 1 (n = 51), while external test cohort 2 (n = 50) was from another hospital. Quantitative parameters were obtained from CECT (unenhanced phase, UP; corticomedullary phase, CMP; nephrographic phase, NP; excretory phase, EP) by measuring attenuation of renal mass and cortex and subsequently calculated. Univariable and multivariable logistic regression analyses were performed to evaluate the association between these parameters and ccRCC. Finally, the constructed models were compared with radiologists' diagnoses.

RESULTS

In univariable analysis, UP-related parameters, particularly UPC-T (cortex minus tumor attenuation on UP), demonstrated AUC of 0.766 in training cohort, 0.901 in test cohort 1, 0.805 in test cohort 2. The heterogeneity-related parameter SD (standard deviation) showed AUC of 0.781, 0.834, and 0.875 respectively. In multivariable analysis, model 1 incorporating UPC-T, NPC-T (cortex minus tumor attenuation on NP), CMPT-UPT (tumor attenuation on CMP minus UP), and SD yielded AUC of 0.866, 0.923, and 0.949 respectively. When compared with radiologists, multivariate models demonstrated higher accuracy (0.800-0.860) and sensitivity (0.794-0.971) than radiologists' assessments (accuracy: 0.700-0.720, sensitivity: 0.588-0.706).

CONCLUSION

Quantitative measurements on CECT, particularly UP- and heterogeneity-related parameters, have potential to discriminate ccRCC and benign renal tumors (fpAML, RO).

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.

Multiphase CT radiomics nomogram for preoperatively predicting the WHO/ISUP nuclear grade of small (< 4 cm) clear cell renal cell carcinoma

BACKGROUND

Small (< 4 cm) clear cell renal cell carcinoma (ccRCC) is the most common type of small renal cancer and its prognosis is poor. However, conventional radiological characteristics obtained by computed tomography (CT) are not sufficient to predict the nuclear grade of small ccRCC before surgery.

METHODS

A total of 113 patients with histologically confirmed ccRCC were randomly assigned to the training set (n = 67) and the testing set (n = 46). The baseline and CT imaging data of the patients were evaluated statistically to develop a clinical model. A radiomics model was created, and the radiomics score (Rad-score) was calculated by extracting radiomics features from the CT images. Then, a clinical radiomics nomogram was developed using multivariate logistic regression analysis by combining the Rad-score and critical clinical characteristics. The receiver operating characteristic (ROC) curve was used to evaluate the discrimination of small ccRCC in both the training and testing sets.

RESULTS

The radiomics model was constructed using six features obtained from the CT images. The shape and relative enhancement value of the nephrographic phase (REV of the NP) were found to be independent risk factors in the clinical model. The area under the curve (AUC) values for the training and testing sets for the clinical radiomics nomogram were 0.940 and 0.902, respectively. Decision curve analysis (DCA) revealed that the radiomics nomogram model was a better predictor, with the highest degree of coincidence.

CONCLUSION

The CT-based radiomics nomogram has the potential to be a noninvasive and preoperative method for predicting the WHO/ISUP grade of small ccRCC.

CT-derived textural analysis parameters discriminate high-attenuation renal cysts from solid renal neoplasms

AIM

To assess the utility of textural features on computed tomography (CT) to differentiate high-attenuation cysts from solid renal neoplasms among indeterminate renal lesions detected incidentally on CT.

MATERIALS AND METHODS

Patients were included if they had an indeterminate renal lesion on CT that was subsequently characterised on ultrasound or magnetic resonance imaging (MRI). Up to three lesions per patient were included if they had a size ≥10 mm and density of 20-70 HU on unenhanced CT or any single phase of contrast-enhanced CT. Cases were categorised as benign or most likely benign cysts (Bosniak II and IIF) versus indeterminate (Bosniak III), mixed solid and cystic (Bosniak IV), or solid renal lesions. A random forest model was generated using 95 textural parameters and four clinical parameters for each lesion.

RESULTS

Two hundred and thirty-four patients were included who had a total of 278 lesions. Of these, 193 (69%) were benign or most likely benign cysts and 85 (31%) were indeterminate, mixed cystic and solid, or solid renal lesions. The random forest model had an area under the curve of 0.71 (95% confidence interval [CI]: 0.65, 0.78), with a sensitivity and specificity of 81.2% and 38.9%, respectively.

CONCLUSION

A multivariate model including textural and clinical parameters had moderate overall performance for discriminating benign or likely benign cysts from indeterminate, mixed solid and cystic, or solid renal lesions. This study serves as a proof of concept and may reduce the need for further follow-up by characterising a significant portion of indeterminate lesions on CT as benign.

A CT-Based Clinical, Radiological and Radiomic Machine Learning Model for Predicting Malignancy of Solid Renal Tumors (UroCCR-75)

BACKGROUND

Differentiating benign from malignant renal tumors is important for patient management, and it may be improved by quantitative CT features analysis including radiomic.

PURPOSE

This study aimed to compare performances of machine learning models using bio-clinical, conventional radiologic and 3D-radiomic features for the differentiation of benign and malignant solid renal tumors using pre-operative multiphasic contrast-enhanced CT examinations.

MATERIALS AND METHODS

A unicentric retrospective analysis of prospectively acquired data from a national kidney cancer database was conducted between January 2016 and December 2020. Histologic findings were obtained by robotic-assisted partial nephrectomy. Lesion images were semi-automatically segmented, allowing for a 3D-radiomic features extraction in the nephrographic phase. Conventional radiologic parameters such as shape, content and enhancement were combined in the analysis. Biological and clinical features were obtained from the national database. Eight machine learning (ML) models were trained and validated using a ten-fold cross-validation. Predictive performances were evaluated comparing sensitivity, specificity, accuracy and AUC.

RESULTS

A total of 122 patients with 132 renal lesions, including 111 renal cell carcinomas (RCCs) (111/132, 84%) and 21 benign tumors (21/132, 16%), were evaluated (58 +/- 14 years, men 74%). Unilaterality (100/111, 90% vs. 13/21, 62%; p = 0.02), necrosis (81/111, 73% vs. 8/21, 38%; p = 0.02), lower values of tumor/cortex ratio at portal time (0.61 vs. 0.74, p = 0.01) and higher variation of tumor/cortex ratio between arterial and portal times (0.22 vs. 0.05, p = 0.008) were associated with malignancy. A total of 35 radiomics features were selected, and "intensity mean value" was associated with RCCs in multivariate analysis (OR = 0.99). After ten-fold cross-validation, a C5.0Tree model was retained for its predictive performances, yielding a sensitivity of 95%, specificity of 42%, accuracy of 87% and AUC of 0.74.

CONCLUSION

Our machine learning-based model combining clinical, radiologic and radiomics features from multiphasic contrast-enhanced CT scans may help differentiate benign from malignant solid renal tumors.

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.

Update on the Applications of Radiomics in Diagnosis, Staging, and Recurrence of Intrahepatic Cholangiocarcinoma

BACKGROUND

This paper offers an assessment of radiomics tools in the evaluation of intrahepatic cholangiocarcinoma.

METHODS

The PubMed database was searched for papers published in the English language no earlier than October 2022.

RESULTS

We found 236 studies, and 37 satisfied our research criteria. Several studies addressed multidisciplinary topics, especially diagnosis, prognosis, response to therapy, and prediction of staging (TNM) or pathomorphological patterns. In this review, we have covered diagnostic tools developed through machine learning, deep learning, and neural network for the recurrence and prediction of biological characteristics. The majority of the studies were retrospective.

CONCLUSIONS

It is possible to conclude that many performing models have been developed to make differential diagnosis easier for radiologists to predict recurrence and genomic patterns. However, all the studies were retrospective, lacking further external validation in prospective and multicentric cohorts. Furthermore, the radiomics models and the expression of results should be standardized and automatized to be applicable in clinical practice.

Contrast-enhanced CT radiomics improves the prediction of abdominal aortic aneurysm progression

OBJECTIVES

To determine if three-dimensional (3D) radiomic features of contrast-enhanced CT (CECT) images improve prediction of rapid abdominal aortic aneurysm (AAA) growth.

METHODS

This longitudinal cohort study retrospectively analyzed 195 consecutive patients (mean age, 72.4 years ± 9.1) with a baseline CECT and a subsequent CT or MR at least 6 months later. 3D radiomic features were measured for 3 regions of the AAA, viz. the vessel lumen only; the intraluminal thrombus (ILT) and aortic wall only; and the entire AAA sac (lumen, ILT, and wall). Multiple machine learning (ML) models to predict rapid growth, defined as the upper tercile of observed growth (> 0.25 cm/year), were developed using data from 60% of the patients. Diagnostic accuracy was evaluated using the area under the receiver operating characteristic curve (AUC) in the remaining 40% of patients.

RESULTS

The median AAA maximum diameter was 3.9 cm (interquartile range [IQR], 3.3-4.4 cm) at baseline and 4.4 cm (IQR, 3.7-5.4 cm) at the mean follow-up time of 3.2 ± 2.4 years (range, 0.5-9 years). A logistic regression model using 7 radiomic features of the ILT and wall had the highest AUC (0.83; 95% confidence interval [CI], 0.73-0.88) in the development cohort. In the independent test cohort, this model had a statistically significantly higher AUC than a model including maximum diameter, AAA volume, and relevant clinical factors (AUC = 0.78, 95% CI, 0.67-0.87 vs AUC = 0.69, 95% CI, 0.57-0.79; p = 0.04).

CONCLUSION

A radiomics-based method focused on the ILT and wall improved prediction of rapid AAA growth from CECT imaging.

KEY POINTS

• Radiomic analysis of 195 abdominal CECT revealed that an ML-based model that included textural features of intraluminal thrombus (if present) and aortic wall improved prediction of rapid AAA progression compared to maximum diameter. • Predictive accuracy was higher when radiomic features were obtained from the thrombus and wall as opposed to the entire AAA sac (including lumen), or the lumen alone. • Logistic regression of selected radiomic features yielded similar accuracy to predict rapid AAA progression as random forests or support vector machines.

CT of the urinary tract revisited

Computed tomography (CT) of the abdomen is usually appropriate for the initial imaging of many urinary tract diseases, due to its wide availability, fast scanning and acquisition of thin slices and isotropic data, that allow the creation of multiplanar reformatted and three-dimensional reconstructed images of excellent anatomic details. Non-enhanced CT remains the standard imaging modality for assessing renal colic. The technique allows the detection of nearly all types of urinary calculi and the estimation of stone burden. CT is the primary diagnostic tool for the characterization of an indeterminate renal mass, including both cystic and solid tumors. It is also the modality of choice for staging a primary renal tumor. Urolithiasis and urinary tract malignancies represent the main urogenic causes of hematuria. CT urography (CTU) improves the visualization of both the upper and lower urinary tract and is recommended for the investigation of gross hematuria and microscopic hematuria, in patients with predisposing factors for urologic malignancies. CTU is highly accurate in the detection and staging of upper tract urothelial malignancies. CT represents the most commonly used technique for the detection and staging of bladder carcinoma and the diagnostic efficacy of CT staging improves with more advanced disease. Nevertheless, it has limited accuracy in differentiating non-muscle invasive bladder carcinoma from muscle-invasive bladder carcinoma. In this review, clinical indications and the optimal imaging technique for CT of the urinary tract is reviewed. The CT features of common urologic diseases, including ureterolithiasis, renal tumors and urothelial carcinomas are discussed.

Radiogenomics in Renal Cancer Management-Current Evidence and Future Prospects

Renal cancer management is challenging from diagnosis to treatment and follow-up. In cases of small renal masses and cystic lesions the differential diagnosis of benign or malignant tissues has potential pitfalls when imaging or even renal biopsy is applied. The recent artificial intelligence, imaging techniques, and genomics advancements have the ability to help clinicians set the stratification risk, treatment selection, follow-up strategy, and prognosis of the disease. The combination of radiomics features and genomics data has achieved good results but is currently limited by the retrospective design and the small number of patients included in clinical trials. The road ahead for radiogenomics is open to new, well-designed prospective studies, with large cohorts of patients required to validate previously obtained results and enter clinical practice.

Differentiating Benign From Malignant Cystic Renal Masses: A Feasibility Study of Computed Tomography Texture-Based Machine Learning Algorithms

OBJECTIVE

The Bosniak classification attempts to predict the likelihood of renal cell carcinoma (RCC) among cystic renal masses but is subject to interobserver variability and often requires multiphase imaging. Artificial intelligence may provide a more objective assessment. We applied computed tomography texture-based machine learning algorithms to differentiate benign from malignant cystic renal masses.

METHODS

This is an institutional review board-approved, Health Insurance Portability and Accountability Act-compliant retrospective study of 147 patients (mean age, 62.4 years; range, 28-89 years; 94 men) with 144 cystic renal masses (93 benign, 51 RCC); 69 were pathology proven (51 RCC, 18 benign), and 75 were considered benign based on more than 4 years of stability at follow-up imaging. Using a single image from a contrast-enhanced abdominal computed tomography scan, mean, SD, mean value of positive pixels, entropy, skewness, and kurtosis radiomics features were extracted. Random forest, multivariate logistic regression, and support vector machine models were used to classify each mass as benign or malignant with 10-fold cross validation. Receiver operating characteristic curves assessed algorithm performance in the aggregated test data.

RESULTS

For the detection of malignancy, sensitivity, specificity, positive predictive value, negative predictive value, and area under the curve were 0.61, 0.87, 0.72, 0.80, and 0.79 for the random forest model; 0.59, 0.87, 0.71, 0.79, and 0.80 for the logistic regression model; and 0.55, 0.86, 0.68, 0.78, and 0.76 for the support vector machine model.

CONCLUSION

Computed tomography texture-based machine learning algorithms show promise in differentiating benign from malignant cystic renal masses. Once validated, these may serve as an adjunct to radiologists' assessments.

Machine learning based multi-label classification of single or mixed-composition urinary stones in in vivo spectral CT

BACKGROUND

Urinary stones comprise both single and mixed compositions. Knowledge of the stone composition helps the urologists choose appropriate medical interventions for patients. The parameters from the spectral computerized tomography (CT) analysis have potential values for identification of the urinary stone compositions.

PURPOSE

The present study aims to identify the compositions of urinary stones in vivo using parameters from spectral CT and machine learning, based on multi-label classification modeling.

METHODS

This retrospective study collected 252 urinary stone samples with single/mixed compositions (including carbapatite [CP], calcium oxalate monohydrate [COM], calcium oxalate dehydrate [COD], uric acid [UA], and struvite [STR]), which were confirmed by ex vivo infrared spectroscopy. Parameters were extracted from an energy spectrum analysis (ESA) of the spectral CT, including the effective atomic number (Z_eff ), Z_eff histogram, CT values at a given x-ray energy level, and material densities. These ESA parameters were utilized for composition analysis via a multi-label classification fusion framework, where 250 multi-label models were built and the classification decisions from the top performance models were integrated by a multi-criterion weighted fusion (MCWF) approach in order to reach a consensus prediction. An example-based metric A c c e x a m $Ac{c_{exam}}$ and label-based metric A c c l a b e l $Ac{c_{label}}$ were used for global and label-wise accuracy evaluations, respectively. The top-ranked parameters associated with discriminating the stone composition were also identified.

RESULTS

The multi-label classification fusion framework achieved an overall A c c e x a m $Ac{c_{exam}}$ of 81.2%, with A c c l a b e l $Ac{c_{label}}$ of 86.7% (CP), 90.6% (COM), 80.6% (COD), 95.0% (UA), and 94.4% (STR) for each composition on the independent testing cohort 1, and A c c e x a m $Ac{c_{exam}}$ of 76.4% with A c c l a b e l $Ac{c_{label}}$ of 80.5% (CP), 88.7% (COM), 74.9% (COD), 94.4% (UA), and 98.5% (STR) on the independent testing cohort 2.

CONCLUSION

The parameters extracted from the ESA on spectral CT can be utilized to characterize single or mixed stone compositions via multi-label classification modeling. The generalization capability of the proposed methodology still requires further verification.

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.

Prediction of early recurrence of hepatocellular carcinoma after liver transplantation based on computed tomography radiomics nomogram

BACKGROUND

Early recurrence results in poor prognosis of patients with hepatocellular carcinoma (HCC) after liver transplantation (LT). This study aimed to explore the value of computed tomography (CT)-based radiomics nomogram in predicting early recurrence of patients with HCC after LT.

METHODS

A cohort of 151 patients with HCC who underwent LT between December 2013 and July 2019 were retrospectively enrolled. A total of 1218 features were extracted from enhanced CT images. The least absolute shrinkage and selection operator algorithm (LASSO) logistic regression was used for dimension reduction and radiomics signature building. The clinical model was constructed after the analysis of clinical factors, and the nomogram was constructed by introducing the radiomics signature into the clinical model. The predictive performance and clinical usefulness of the three models were evaluated using receiver operating characteristic (ROC) curve analysis and decision curve analysis (DCA), respectively. Calibration curves were plotted to assess the calibration of the nomogram.

RESULTS

There were significant differences in radiomics signature among early recurrence patients and non-early recurrence patients in the training cohort (P < 0.001) and validation cohort (P < 0.001). The nomogram showed the best predictive performance, with the largest area under the ROC curve in the training (0.882) and validation (0.917) cohorts. Hosmer-Lemeshow testing confirmed that the nomogram showed good calibration in the training (P = 0.138) and validation (P = 0.396) cohorts. DCA showed if the threshold probability is within 0.06-1, the nomogram had better clinical usefulness than the clinical model.

CONCLUSIONS

Our CT-based radiomics nomogram can preoperatively predict the risk of early recurrence in patients with HCC after LT.

The effect of CT texture-based analysis using machine learning approaches on radiologists' performance in differentiating focal-type autoimmune pancreatitis and pancreatic duct carcinoma

PURPOSE

To develop a support vector machine (SVM) classifier using CT texture-based analysis in differentiating focal-type autoimmune pancreatitis (AIP) and pancreatic duct carcinoma (PD), and to assess the radiologists' diagnostic performance with or without SVM.

MATERIALS AND METHODS

This retrospective study included 50 patients (20 patients with focal-type AIP and 30 patients with PD) who underwent dynamic contrast-enhanced CT. Sixty-two CT texture-based features were extracted from 2D images of the arterial and portal phase CTs. We conducted data compression and feature selections using principal component analysis (PCA) and produced the SVM classifier. Four readers participated in this observer performance study and the statistical significance of differences with and without the SVM was assessed by receiver operating characteristic (ROC) analysis.

RESULTS

The SVM performance indicated a high performance in differentiating focal-type AIP and PD (AUC = 0.920). The AUC for all 4 readers increased significantly from 0.827 to 0.911 when using the SVM outputs (p = 0.010). The AUC for inexperienced readers increased significantly from 0.781 to 0.905 when using the SVM outputs (p = 0.310). The AUC for experienced readers increased from 0.875 to 0.912 when using the SVM outputs, however, there was no significant difference (p = 0.018).

CONCLUSION

The use of SVM classifier using CT texture-based features improved the diagnostic performance for differentiating focal-type AIP and PD on CT.

Current and emerging artificial intelligence applications for pediatric abdominal imaging

Artificial intelligence (AI) uses computers to mimic cognitive functions of the human brain, allowing inferences to be made from generally large datasets. Traditional machine learning (e.g., decision tree analysis, support vector machines) and deep learning (e.g., convolutional neural networks) are two commonly employed AI approaches both outside and within the field of medicine. Such techniques can be used to evaluate medical images for the purposes of automated detection and segmentation, classification tasks (including diagnosis, lesion or tissue characterization, and prediction), and image reconstruction. In this review article we highlight recent literature describing current and emerging AI methods applied to abdominal imaging (e.g., CT, MRI and US) and suggest potential future applications of AI in the pediatric population.

An Optimized Radiomics Model Based on Automated Breast Volume Scan Images to Identify Breast Lesions: Comparison of Machine Learning Methods: Comparison of Machine Learning Methods

OBJECTIVES

To develop and test an optimized radiomics model based on multi-planar automated breast volume scan (ABVS) images to identify malignant and benign breast lesions.

METHODS

Patients (n = 200) with breast lesions who underwent ABVS examinations were included. For each patient, 208 radiomics features were extracted from the ABVS images, including axial plane and coronal plane. Recursive feature elimination, random forest, and chi-square test were used to select features. A support vector machine, logistic regression, and extreme gradient boosting were utilized as classifiers to differentiate malignant and benign breast lesions. The area under the curve, sensitivity, specificity, accuracy, and precision was used to evaluate the performance of the radiomics models. Generalization of the radiomics models was verified through 5-fold cross-validation.

RESULTS

For a single plane or a combination of planes, a combination of recursive feature elimination, and support vector machine yielded the best performance when identifying breast lesions. The machine learning models based on a combination of planes performed better than those based on a single plane. Regarding the axial plane and coronal plane, the machine learning model using a combination of recursive feature elimination and support vector machine yielded the optimal identification performance: average area under the curve (0.857 ± 0.058, 95% confidence interval, 0.763-0.957); the average values of sensitivity, specificity, accuracy, and precision were 87.9, 68.2, 80.7, and 82.9%, respectively.

CONCLUSIONS

The optimized radiomics model based on ABVS images can provide valuable information for identifying benign and malignant breast lesions preoperatively and guide the accurate clinical treatment. Further external validation is required.

Multiphasic CT-Based Radiomics Analysis for the Differentiation of Benign and Malignant Parotid Tumors

Objective

This study aims to investigate the value of machine learning models based on clinical-radiological features and multiphasic CT radiomics features in the differentiation of benign parotid tumors (BPTs) and malignant parotid tumors (MPTs).

Methods

This retrospective study included 312 patients (205 cases of BPTs and 107 cases of MPTs) who underwent multiphasic enhanced CT examinations, which were randomly divided into training (N = 218) and test (N = 94) sets. The radiomics features were extracted from the plain, arterial, and venous phases. The synthetic minority oversampling technique was used to balance minority class samples in the training set. Feature selection methods were done using the least absolute shrinkage and selection operator (LASSO), mutual information (MI), and recursive feature extraction (RFE). Two machine learning classifiers, support vector machine (SVM), and logistic regression (LR), were then combined in pairs with three feature selection methods to build different radiomics models. Meanwhile, the prediction performances of different radiomics models based on single phase (plain, arterial, and venous phase) and multiphase (three-phase combination) were compared to determine which model construction method and phase were more discriminative. In addition, clinical models based on clinical-radiological features and combined models integrating radiomics features and clinical-radiological features were established. The prediction performances of the different models were evaluated by the area under the receiver operating characteristic (ROC) curve (AUC) and the drawing of calibration curves.

Results

Among the 24 established radiomics models composed of four different phases, three feature selection methods, and two machine learning classifiers, the LASSO-SVM model based on a three-phase combination had the optimal prediction performance with AUC (0.936 [95% CI = 0.866, 0.976]), sensitivity (0.78), specificity (0.90), and accuracy (0.86) in the test set, and its prediction performance was significantly better than with the clinical model based on LR (AUC = 0.781, p = 0.012). In the test set, the combined model based on LR had a lower AUC than the optimal radiomics model (AUC = 0.933 vs. 0.936), but no statistically significant difference (p = 0.888).

Conclusion

Multiphasic CT-based radiomics analysis showed a machine learning model based on clinical-radiological features and radiomics features has the potential to provide a valuable tool for discriminating benign from malignant parotid tumors.

Predicting Kirsten Rat Sarcoma Virus Gene Mutation Status in Patients With Colorectal Cancer by Radiomics Models Based on Multiphasic CT

Objective

To develop and validate radiomics models based on multiphasic CT in predicting Kirsten rat sarcoma virus (KRAS) gene mutation status in patients with colorectal cancer (CRC).

Materials and Methods

A total of 231 patients with pathologically confirmed CRC were retrospectively enrolled and randomly divided into training(n=184) and test groups(n=47) in a ratio of 4:1. A total of 1316 quantitative radiomics features were extracted from non-contrast phase (NCP), arterial-phase (AP) and venous-phase (VP) CT for each patient. Four steps were applied for feature selection including Spearman correlation analysis, variance threshold, least absolute contraction and selection operator, and multivariate stepwise regression analysis. Clinical and pathological characteristics were also assessed. Subsequently, three classification methods, logistic regression (LR), support vector machine (SVM) and random tree (RT) algorithm, were applied to develop seven groups of prediction models (NCP, AP, VP, AP+VP, AP+VP+NCP, AP&amp;VP, AP&amp;VP&amp;NCP) for KRAS mutation prediction. The performance of these models was evaluated by receiver operating characteristics curve (ROC) analysis.

Results

Among the three groups of single-phase models, the AP model, developed by LR algorithm, showed the best prediction performance with an AUC value of 0.811 (95% CI:0.685–0.938) in the test cohort. Compared with the single-phase models, the dual-phase (AP+VP) model with the LR algorithm showed better prediction performance (AUC=0.826, 95% CI:0.700-0.952). The performance of multiphasic (AP+VP+NCP) model with the LR algorithm (AUC=0.811, 95%CI: 0.679-0.944) is comparable to the model with the SVM algorithm (AUC=0.811, 95%CI: 0.695-0.918) in the test cohort, but the sensitivity, specificity, and accuracy of the multiphasic (AP+VP+NCP) model with the LR algorithm were 0.810, 0.808, 0.809 respectively, which were highest among these seven groups of prediction models in the test cohort.

Conclusion

The CT radiomics models have the potential to predict KRAS mutation in patients with CRC; different phases may affect the predictive efficacy of radiomics model, of which arterial-phase CT is more informative. The combination of multiphasic CT images can further improve the performance of radiomics model.

Utility of radiomics features of diffusion-weighted magnetic resonance imaging for differentiation of fat-poor angiomyolipoma from clear cell renal cell carcinoma: model development and external validation

PURPOSE

To investigate the utility of radiomics features of diffusion-weighted magnetic resonance imaging (DW-MRI) to differentiate fat-poor angiomyolipoma (fpAML) from clear cell renal cell carcinoma (ccRCC).

MATERIALS AND METHODS

This multi-institutional study included two cohorts with pathologically confirmed renal tumors: 65 patients with ccRCC and 18 with fpAML in the model development cohort, and 17 with ccRCC and 13 with fpAML in the external validation cohort. All patients underwent magnetic resonance imaging (MRI) including DW-MRI. Radiomics analysis was used to extract 39 imaging features from the apparent diffusion coefficient (ADC) map. The radiomics features were analyzed with unsupervised hierarchical cluster analysis. A random forest (RF) model was used to identify radiomics features important for differentiating fpAML from ccRCC in the development cohort. The diagnostic performance of the RF model was evaluated in the development and validation cohorts.

RESULTS

The cases in the developmental cohort were classified into three groups with different frequencies of fpAML by cluster analysis of radiomics features. RF analysis of the development cohort showed that the mean ADC value was important for differentiating fpAML from ccRCC, as well as higher-texture features including gray-level run length matrix (GLRLM)_long-run low gray-level enhancement (LRLGE), and GLRLM_low gray-level run emphasis (LGRE). The area under the curve values of the development [0.90, 95% confidence interval (CI) 0.80-1.00] and validation cohorts (0.87, 95% CI 0.74-1.00) were similar (P = 0.91).

CONCLUSION

The radiomics features of ADC maps are useful for differentiating fpAML from ccRCC.

Artificial Intelligence in Kidney Cancer

Artificial intelligence is rapidly expanding into nearly all facets of life, particularly within the field of medicine. The diagnosis, characterization, management, and treatment of kidney cancer is ripe with areas for improvement that may be met with the promises of artificial intelligence. Here, we explore the impact of current research work in artificial intelligence for clinicians caring for patients with renal cancer, with a focus on the perspectives of radiologists, pathologists, and urologists. Promising preliminary results indicate that artificial intelligence may assist in the diagnosis and risk stratification of newly discovered renal masses and help guide the clinical treatment of patients with kidney cancer. However, much of the work in this field is still in its early stages, limited in its broader applicability, and hampered by small datasets, the varied appearance and presentation of kidney cancers, and the intrinsic limitations of the rigidly structured tasks artificial intelligence algorithms are trained to complete. Nonetheless, the continued exploration of artificial intelligence holds promise toward improving the clinical care of patients with kidney cancer.

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.

Radiomics signature from [18F]FDG PET images for prognosis predication of primary gastrointestinal diffuse large B cell lymphoma

OBJECTIVES

To investigate the prognostic value of PET radiomics feature in the prognosis of patients with primary gastrointestinal diffuse large B cell lymphoma (PGI-DLBCL) treated with R-CHOP-like regimen.

METHODS

A total of 140 PGI-DLBCL patients who underwent pre-therapy [¹⁸F] FDG PET/CT were enrolled in this retrospective analysis. PET radiomics features obtained from patients in the training cohort were subjected to three machine learning methods and Pearson's correlation test for feature selection. Support vector machine (SVM) was used to build a radiomics signature classifier associated with progression-free survival (PFS) and overall survival (OS). A multivariate Cox proportional hazards regression model was established to predict survival outcomes.

RESULTS

A total of 1421 PET radiomics features were extracted and reduced to 5 features to build a radiomics signature which was significantly associated with PFS and OS (p < 0.05). The combined model incorporating radiomics signatures, metabolic metrics, and clinical risk factors showed high C-indices in both the training (PFS: 0.825, OS: 0.834) and validation sets (PFS: 0.831, OS: 0.877). Decision curve analysis (DCA) demonstrated that the combined models achieved the most net benefit across a wider reasonable range of threshold probabilities for predicting PFS and OS.

CONCLUSION

The newly developed radiomics signatures obtained by the ensemble strategy were independent predictors of PFS and OS for PGI-DLBCL patients. Moreover, the combined model with clinical and metabolic factors was able to predict patient prognosis and may enable personalized treatment decision-making.

KEY POINTS

• Radiomics signatures generated from the optimal radiomics feature set from the [¹⁸F]FDG PET images can predict the survival of PGI-DLBCL patients. • The optimal radiomics feature set is constructed by integrating the feature selection outputs of LASSO, RF, Xgboost, and PC methods. • Combined models incorporating radiomics signatures from¹⁸F-FDG PET images, metabolic parameters, and clinical factors outperformed clinical models, and NCCN-IPI.

Radiology Imaging Scans for Early Diagnosis of Kidney Tumors: A Review of Data Analytics-Based Machine Learning and Deep Learning Approaches

Plenty of disease types exist in world communities that can be explained by humans’ lifestyles or the economic, social, genetic, and other factors of the country of residence. Recently, most research has focused on studying common diseases in the population to reduce death risks, take the best procedure for treatment, and enhance the healthcare level of the communities. Kidney Disease is one of the common diseases that have affected our societies. Sectionicularly Kidney Tumors (KT) are the 10th most prevalent tumor for men and women worldwide. Overall, the lifetime likelihood of developing a kidney tumor for males is about 1 in 466 (2.02 percent) and it is around 1 in 80 (1.03 percent) for females. Still, more research is needed on new diagnostic, early, and innovative methods regarding finding an appropriate treatment method for KT. Compared to the tedious and time-consuming traditional diagnosis, automatic detection algorithms of machine learning can save diagnosis time, improve test accuracy, and reduce costs. Previous studies have shown that deep learning can play a role in dealing with complex tasks, diagnosis and segmentation, and classification of Kidney Tumors, one of the most malignant tumors. The goals of this review article on deep learning in radiology imaging are to summarize what has already been accomplished, determine the techniques used by the researchers in previous years in diagnosing Kidney Tumors through medical imaging, and identify some promising future avenues, whether in terms of applications or technological developments, as well as identifying common problems, describing ways to expand the data set, summarizing the knowledge and best practices, and determining remaining challenges and future directions.

An Ultrasonic-Based Radiomics Nomogram for Distinguishing Between Benign and Malignant Solid Renal Masses

Objectives

This study was conducted in order to develop and validate an ultrasonic-based radiomics nomogram for diagnosing solid renal masses.

Methods

Six hundred renal solid masses with benign renal lesions (n = 204) and malignant renal tumors (n = 396) were divided into a training set (n = 480) and a validation set (n = 120). Radiomics features were extracted from ultrasound (US) images preoperatively and then a radiomics score (RadScore) was calculated. By integrating the RadScore and independent clinical factors, a radiomics nomogram was constructed. The diagnostic performance of junior physician, senior physician, RadScore, and radiomics nomogram in identifying benign from malignant solid renal masses was evaluated based on the area under the receiver operating characteristic curve (ROC) in both the training and validation sets. The clinical usefulness of the nomogram was assessed using decision curve analysis (DCA).

Results

The radiomics signature model showed satisfactory discrimination in the training set [area under the ROC (AUC), 0.887; 95% confidence interval (CI), 0.860–0.915] and the validation set (AUC, 0.874; 95% CI, 0.816–0.932). The radiomics nomogram also demonstrated good calibration and discrimination in the training set (AUC, 0.911; 95% CI, 0.886–0.936) and the validation set (AUC, 0.861; 95% CI, 0.802–0.921). In addition, the radiomics nomogram model showed higher accuracy in discriminating benign and malignant renal masses compared with the evaluations by junior physician (DeLong p = 0.004), and the model also showed significantly higher specificity than the senior and junior physicians (0.93 vs. 0.57 vs. 0.46).

Conclusions

The ultrasonic-based radiomics nomogram shows favorable predictive efficacy in differentiating solid renal masses.

Pulmonary MRI Radiomics and Machine Learning: Effect of Intralesional Heterogeneity on Classification of Lesion

RATIONALE AND OBJECTIVES

To investigate the effect of intralesional heterogeneity on differentiating benign and malignant pulmonary lesions, quantitative magnetic resonance imaging (MRI) radiomics, and machine learning methods were adopted.

MATERIALS AND METHODS

A total of 176 patients with multiparametric MRI were involved in this exploratory study. To investigate the effect of intralesional heterogeneity on lesion classification, a radiomics model called tumor heterogeneity model was developed and compared to the conventional radiomics model based on the entire tumor. In tumor heterogeneity model, each lesion was divided into five sublesions depending on the spatial location through clustering algorithm. From the five sublesions in multi-parametric MRI sequences, 1100 radiomics features were extracted. The recursive feature elimination method was employed to select features and support vector machine classifier was used to distinguish benign and malignant lesion. The performance of classification was evaluated with the receiver operating characteristic curve and the area under the curve (AUC) was the figure of merit. The 3-fold cross-validation (CV) with and without nesting was used to validate the model, respectively.

RESULTS

The tumor heterogeneity model (AUC = 0.74 ± 0.04 and 0.90 ± 0.03, CV with and without nesting, respectively) outperforms conventional model (AUC = 0.68 ± 0.04 and 0.87 ± 0.03). The difference between the two models is statistically significant (p = 0.03) for lesions greater than 18.80 cm³.

CONCLUSION

Intralesional heterogeneity influences the classification of pulmonary lesions. The tumor heterogeneity model tends to perform better than conventional radiomics model.

A combined postoperative nomogram for survival prediction in clear cell renal carcinoma

PURPOSE

To investigate and validate the prognostic value of nomogram models for predicting disease-free survival (DFS) and overall survival (OS) in patients with clear cell renal cell carcinoma (ccRCC).

METHODS

In this retrospective study, 223 patients (age 54.38 ± 10.93 years) with pathologically confirmed ccRCC who underwent resection and lymph node dissection between March 2010 and September 2018 were investigated. All patients were randomly divided into training (n = 155) and validation (n = 68) cohorts. Radiomics features were extracted from computed tomography (CT) images in the unenhanced, corticomedullary, and nephrographic phases. Radiomic score was calculated and combined with clinicopathological factors for model construction and nomogram development. Clinicopathological factors and imaging features were collected at initial diagnosis. Univariate and multivariate Cox proportional hazards regression analyses were used to evaluate the relationship between the radiomics signature and prognosis outcomes.

RESULTS

There were four prognostic factors for predicting DFS and five factors for predicting OS in our nomogram model (P < 0.05). The radiomics signature correlated independently with DFS (hazard ratio = 27; P < 0.001) and OS (hazard ratio = 25; P < 0.001). The nomogram showed excellent performance (C-index = 0.825) for predicting DFS. The combined nomogram also showed the highest C-index for OS (C-index = 0.943), which was verified in the validation dataset.

CONCLUSION

The combined nomogram model based on radiomics, clinicopathological factors, and preoperative CT features can accurately perform prognosis and survival analysis and can potentially be used for preoperative non-invasive survival prediction in ccRCC patients.

Exploring value of CT coronary imaging combined with machine-learning methods to predict myocardial ischemia

PURPOSE

To establish a machine-learning (ML) model based on coronary computed tomography angiography (CTA) images for evaluating myocardial ischemia in patients diagnosed with coronary atherosclerosis.

METHODS

This retrospective analysis includes CTA images acquired from 110 patients. Among them, 58 have myocardial ischemia and 52 have normal myocardial blood supply. The patients are divided into training and test datasets with a ratio 7 : 3. Deep learning model-based CQK software is used to automatically segment myocardium on CTA images and extract texture features. Then, seven ML models are constructed to classify between myocardial ischemia and normal myocardial blood supply cases. Predictive performance and stability of the classifiers are determined by receiver operating characteristic curve with cross validation. The optimal ML model is then validated using an independent test dataset.

RESULTS

Accuracy and areas under ROC curves (AUC) obtained from the support vector machine with extreme gradient boosting linear method are 0.821 and 0.777, respectively, while accuracy and AUC achieved by the neural network (NN) method are 0.818 and 0.757, respectively. The naive Bayes model yields the highest sensitivity (0.942), and the random forest model yields the highest specificity (0.85). The k-nearest neighbors model yields the lowest accuracy (0.74). Additionally, NN model demonstrates the lowest relative standard deviations (0.16 for accuracy and 0.08 for AUC) indicating the high stability of this model, and its AUC applying to the independent test dataset is 0.72.

CONCLUSION

The NN model demonstrates the best performance in predicting myocardial ischemia using radiomics features computed from CTA images, which suggests that this ML model has promising potential in guiding clinical decision-making.

A Multiparametric MR-Based RadioFusionOmics Model with Robust Capabilities of Differentiating Glioblastoma Multiforme from Solitary Brain Metastasis

Simple Summary

Glioblastoma multiforme (GBM) and solitary brain metastasis (SBM) are common brain tumors in adults. The two tumors often pose a diagnostic dilemma owing to their similar features on conventional magnetic resonance imaging (MRI). Ability to discriminate the two tumors is critical as it informs clinical treatment strategies. This pilot study attempts to employ the machine learning technique to identify GBM and SBM by fusing radiomics features of multiple MRI sequences and multiple models. A multiparametric MR-based RadioFusionOmics (RFO) model was developed and has demonstrated promising prediction accuracy for the identifications of GBM and SBM.

Abstract

This study aimed to evaluate the diagnostic potential of a novel RFO model in differentiating GBM and SBM with multiparametric MR sequences collected from 244 (131 GBM and 113 SBM) patients. Three basic volume of interests (VOIs) were delineated on the conventional axial MR images (T₁WI, T₂WI, T₂_FLAIR, and CE_T₁WI), including volumetric non-enhanced tumor (nET), enhanced tumor (ET), and peritumoral edema (pTE). Using the RFO model, radiomics features extracted from different multiparametric MRI sequence(s) and VOI(s) were fused and the best sequence and VOI, or possible combinations, were determined. A multi-disciplinary team (MDT)-like fusion was performed to integrate predictions from the high-performing models for the final discrimination of GBM vs. SBM. Image features extracted from the volumetric ET (VOI_ET) had dominant predictive performances over features from other VOI combinations. Fusion of VOI_ET features from the T₁WI and T₂_FLAIR sequences via the RFO model achieved a discrimination accuracy of AUC = 0.925, accuracy = 0.855, sensitivity = 0.856, and specificity = 0.853, on the independent testing cohort 1, and AUC = 0.859, accuracy = 0.836, sensitivity = 0.708, and specificity = 0.919 on the independent testing cohort 2, which significantly outperformed three experienced radiologists (p = 0.03, 0.01, 0.02, and 0.01, and p = 0.02, 0.01, 0.45, and 0.02, respectively) and the MDT-decision result of three experienced experts (p = 0.03, 0.02, 0.03, and 0.02, and p = 0.03, 0.02, 0.44, and 0.03, respectively).

Development and Validation of Multivariate Models Integrating Preoperative Clinicopathological Parameters and Radiographic Findings Based on Late Arterial Phase CT Images for Predicting Lymph Node Metastasis in Gastric Cancer

RATIONALE AND OBJECTIVES

To develop and validate multivariate models integrating endoscopic biopsy, tumor markers, computed tomography (CT) morphological characteristics based on late arterial phase (LAP), and CT value-related and texture parameters to predict lymph node (LN) metastasis in gastric cancers (GCs).

MATERIALS AND METHODS

The preoperative differentiation degree based on biopsy, 6 tumor markers, 8 CT morphological characteristics based on LAP, 18 CT value-related parameters, and 35 CT texture parameters of 163 patients (111 men and 52 women) with GC were analyzed retrospectively. The differences in parameters between N (-) and N (+) GCs were analyzed by the Mann-Whitney U test. Diagnostic performance was obtained by receiver operating characteristic (ROC) curve analysis. Multivariate models based on regression analysis and machine learning algorithms were performed to improve diagnostic efficacy.

RESULTS

The differentiation degree, carbohydrate antigen (CA) 199 and CA242, 5 CT morphological characteristics, and 22 CT texture parameters showed significant differences between N (-) and N (+) GCs in the primary cohort (all p < 0.05). The multivariate model integrating clinicopathological parameters and radiographic findings based on regression analysis achieved areas under the ROC curve (AUCs) of 0.936 and 0.912 in the primary and validation cohorts, respectively. The model generated by the support vector machine algorithm achieved AUCs of 0.914 and 0.948, respectively.

CONCLUSION

We developed and validated multivariate models integrating endoscopic biopsy, tumor markers, CT morphological characteristics based on LAP, and CT texture parameters to predict LN metastasis in GCs and achieved satisfactory performance.

A Reliable Prediction Model for Renal Cell Carcinoma Subtype Based on Radiomic Features from 3D Multiphase Enhanced CT Images

Background

This study aimed to develop a prediction model to distinguish renal cell carcinoma (RCC) subtypes.

Methods

The radiomic features (RFs) from 5 different computed tomography (CT) phases were used in the prediction models: noncontrast phase (NCP), corticomedullary phase (CMP), nephrographic phase (NP), excretory phase (EP), and all-phase (ALL-P).

Results

For the ALL-P model, all of the RFs obtained from the 4 single-phase images were combined to 420 RFs. The ALL-P model performed the best of all models, with an accuracy of 0.80; the sensitivity and specificity for clear cell RCC (ccRCC) were 0.85 and 0.83; those for papillary RCC (pRCC) were 0.60 and 0.91; those for chromophobe RCC (cRCC) were 0.66 and 0.91, respectively. Binary classification experiments showed for distinguishing ccRCC vs. not-ccRCC that the area under the receiver operating characteristic curve (AUC) of the ALL-P and CMP models was 0.89, but the overall sensitivity/specificity/accuracy of the ALL-P model was better. For cRCC vs. non-cRCC, the ALL-P model had the best performance.

Conclusions

A reliable prediction model for RCC subtypes was constructed. The performance of the ALL-P prediction model was the best as compared to individual single-phase models and the traditional prediction model.

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.

Development and Validation of a Radiomic Nomogram for Predicting the Prognosis of Kidney Renal Clear Cell Carcinoma

Purpose

The present study aims to comprehensively investigate the prognostic value of a radiomic nomogram that integrates contrast-enhanced computed tomography (CECT) radiomic signature and clinicopathological parameters in kidney renal clear cell carcinoma (KIRC).

Methods

A total of 136 and 78 KIRC patients from the training and validation cohorts were included in the retrospective study. The intraclass correlation coefficient (ICC) was used to assess reproducibility of radiomic feature extraction. Univariate Cox analysis and least absolute shrinkage and selection operator (LASSO) as well as multivariate Cox analysis were utilized to construct radiomic signature and clinical signature in the training cohort. A prognostic nomogram was established containing a radiomic signature and clinicopathological parameters by using a multivariate Cox analysis. The predictive ability of the nomogram [relative operating characteristic curve (ROC), concordance index (C-index), Hosmer–Lemeshow test, and calibration curve] was evaluated in the training cohort and validated in the validation cohort. Patients were split into high- and low-risk groups, and the Kaplan–Meier (KM) method was conducted to identify the forecasting ability of the established models. In addition, genes related with the radiomic risk score were determined by weighted correlation network analysis (WGCNA) and were used to conduct functional analysis.

Results

A total of 2,944 radiomic features were acquired from the tumor volumes of interest (VOIs) of CECT images. The radiomic signature, including ten selected features, and the clinical signature, including three selected clinical variables, showed good performance in the training and validation cohorts [area under the curve (AUC), 0.897 and 0.712 for the radiomic signature; 0.827 and 0.822 for the clinical signature, respectively]. The radiomic prognostic nomogram showed favorable performance and calibration in the training cohort (AUC, 0.896, C-index, 0.846), which was verified in the validation cohort (AUC, 0.768). KM curves indicated that the progression-free interval (PFI) time was dramatically shorter in the high-risk group than in the low-risk group. The functional analysis indicated that radiomic signature was significantly associated with T cell activation.

Conclusions

The nomogram combined with CECT radiomic and clinicopathological signatures exhibits excellent power in predicting the PFI of KIRC patients, which may aid in clinical management and prognostic evaluation of cancer patients.

Kidney cancer management 3.0: can artificial intelligence make us better?

PURPOSE OF REVIEW

Artificial intelligence holds tremendous potential for disrupting clinical medicine. Here we review the current role of artificial intelligence in the kidney cancer space.

RECENT FINDINGS

Machine learning and deep learning algorithms have been developed using information extracted from radiomic, histopathologic, and genomic datasets of patients with renal masses.

SUMMARY

Although artificial intelligence applications in medicine are still in their infancy, they already hold immediate promise to improve accuracy of renal mass characterization, grade, and prognostication. As algorithms become more robust and generalizable, artificial intelligence is poised to significantly disrupt kidney cancer care.

Uncontrolled Confounders May Lead to False or Overvalued Radiomics Signature: A Proof of Concept Using Survival Analysis in a Multicenter Cohort of Kidney Cancer

Purpose

We aimed to explore potential confounders of prognostic radiomics signature predicting survival outcomes in clear cell renal cell carcinoma (ccRCC) patients and demonstrate how to control for them.

Materials and Methods

Preoperative contrast enhanced abdominal CT scan of ccRCC patients along with pathological grade/stage, gene mutation status, and survival outcomes were retrieved from The Cancer Imaging Archive (TCIA)/The Cancer Genome Atlas-Kidney Renal Clear Cell Carcinoma (TCGA-KIRC) database, a publicly available dataset. A semi-automatic segmentation method was applied to segment ccRCC tumors, and 1,160 radiomics features were extracted from each segmented tumor on the CT images. Non-parametric principal component decomposition (PCD) and unsupervised hierarchical clustering were applied to build the radiomics signature models. The factors confounding the radiomics signature were investigated and controlled sequentially. Kaplan-Meier curves and Cox regression analyses were performed to test the association between radiomics signatures and survival outcomes.

Results

183 patients of TCGA-KIRC cohort with available imaging, pathological, and clinical outcomes were included in this study. All 1,160 radiomics features were included in the first radiomics signature. Three additional radiomics signatures were then modelled in successive steps removing redundant radiomics features first, removing radiomics features biased by CT slice thickness second, and removing radiomics features dependent on tumor size third. The final radiomics signature model was the most parsimonious, unbiased by CT slice thickness, and independent of tumor size. This final radiomics signature stratified the cohort into radiomics phenotypes that are different by cancer-specific and recurrence-free survival; HR (95% CI) = 3.0 (1.5-5.7), p <0.05 and HR (95% CI) = 6.6 (3.1-14.1), p <0.05, respectively.

Conclusion

Radiomics signature can be confounded by multiple factors, including feature redundancy, image acquisition parameters like slice thickness, and tumor size. Attention to and proper control for these potential confounders are necessary for a reliable and clinically valuable radiomics signature.

CT-based radiomics for differentiating renal tumours: a systematic review

PURPOSE

Differentiating renal tumours into grades and tumour subtype from medical imaging is important for patient management; however, there is an element of subjectivity when performed qualitatively. Quantitative analysis such as radiomics may provide a more objective approach. The purpose of this article is to systematically review the literature on computed tomography (CT) radiomics for grading and differentiating renal tumour subtypes. An educational perspective will also be provided.

METHODS

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses checklist was followed. PubMed, Scopus and Web of Science were searched for relevant articles. The quality of each study was assessed using the Radiomic Quality Score (RQS).

RESULTS

13 studies were found. The main outcomes were prediction of pathological grade and differentiating between renal tumour types, measured as area under the curve (AUC) for either the receiver operator curve or precision recall curve. Features extracted to predict pathological grade or tumour subtype included shape, intensity, texture and wavelet (a type of higher order feature). Four studies differentiated between low-grade and high-grade clear cell renal cell cancer (RCC) with good performance (AUC = 0.82-0.978). One other study differentiated low- and high-grade chromophobe with AUC = 0.84. Finally, eight studies used radiomics to differentiate between tumour types such as clear cell RCC, fat-poor angiomyolipoma, papillary RCC, chromophobe RCC and renal oncocytoma with high levels of performance (AUC 0.82-0.96).

CONCLUSION

Renal tumours can be pathologically classified using CT-based radiomics with good performance. The main radiomic feature used for tumour differentiation was texture. Fuhrman was the most common pathologic grading system used in the reviewed studies. Renal tumour grading studies should be extended beyond clear cell RCC and chromophobe RCC. Further research with larger prospective studies, performed in the clinical setting, across multiple institutions would help with clinical translation to the radiologist's workstation.

Prediction of Microvascular Invasion in Hepatocellular Carcinoma With a Multi-Disciplinary Team-Like Radiomics Fusion Model on Dynamic Contrast-Enhanced Computed Tomography

Objective

To investigate microvascular invasion (MVI) of HCC through a noninvasive multi-disciplinary team (MDT)-like radiomics fusion model on dynamic contrast enhanced (DCE) computed tomography (CT).

Methods

This retrospective study included 111 patients with pathologically proven hepatocellular carcinoma, which comprised 57 MVI-positive and 54 MVI-negative patients. Target volume of interest (VOI) was delineated on four DCE CT phases. The volume of tumor core (V_tc) and seven peripheral tumor regions (V_pt, with varying distances of 2, 4, 6, 8, 10, 12, and 14 mm to tumor margin) were obtained. Radiomics features extracted from different combinations of phase(s) and VOI(s) were cross-validated by 150 classification models. The best phase and VOI (or combinations) were determined. The top predictive models were ranked and screened by cross-validation on the training/validation set. The model fusion, a procedure analogous to multidisciplinary consultation, was performed on the top-3 models to generate a final model, which was validated on an independent testing set.

Results

Image features extracted from V_tc+V_pt(12mm) in the portal venous phase (PVP) showed dominant predictive performances. The top ranked features from V_tc+V_pt(12mm) in PVP included one gray level size zone matrix (GLSZM)-based feature and four first-order based features. Model fusion outperformed a single model in MVI prediction. The weighted fusion method achieved the best predictive performance with an AUC of 0.81, accuracy of 78.3%, sensitivity of 81.8%, and specificity of 75% on the independent testing set.

Conclusion

Image features extracted from the PVP with V_tc+V_pt(12mm) are the most reliable features indicative of MVI. The MDT-like radiomics fusion model is a promising tool to generate accurate and reproducible results in MVI status prediction in HCC.

The Prognostic Value of Radiomics Features Extracted From Computed Tomography in Patients With Localized Clear Cell Renal Cell Carcinoma After Nephrectomy

Background and purpose

Radiomics is an emerging field of quantitative imaging. The prognostic value of radiomics analysis in patients with localized clear cell renal cell carcinoma (ccRCC) after nephrectomy remains unknown.

Methods

Computed tomography images of 167 eligible cases were obtained from the Cancer Imaging Archive database. Radiomics features were extracted from the region of interest contoured manually for each patient. Hierarchical clustering was performed to divide patients into distinct groups. Prognostic assessments were performed by Kaplan–Meier curves, COX regression, and least absolute shrinkage and selection operator COX regression. Besides, transcriptome mRNA data were also included in the prognostic analyses. Endpoints were overall survival (OS) and disease-free survival (DFS). Concordance index (C-index), decision curve analysis and calibration curves with 1,000 bootstrapping replications were used for model’s validation.

Results

Hierarchical clustering groups from nephrographic features and mRNA can divide patients into different prognostic groups while clustering groups from corticomedullary or unenhanced phase couldn’t distinguish patients’ prognosis. In multivariate analyses, 11 OS-predicting and eight DFS-predicting features were identified in nephrographic phase. Similarly, seven OS-predictors and seven DFS-predictors were confirmed in mRNA data. In contrast, limited prognostic features were found in corticomedullary (two OS-predictor and two DFS-predictors) and unenhanced phase (one OS-predictors and two DFS-predictors). Prognostic models combining both nephrographic features and mRNA showed improved C-index than any model alone (C-index: 0.927 and 0.879 for OS- and DFS-predicting, respectively). In addition, decision curves and calibration curves also revealed the great performance of the novel models.

Conclusion

We firstly investigated the prognostic significance of preoperative radiomics signatures in ccRCC patients. Radiomics features obtained from nephrographic phase had stronger predictive ability than features from corticomedullary or unenhanced phase. Multi-omics models combining radiomics and transcriptome data could further increase the predictive accuracy.

A CT-Based Tumoral and Mini-Peritumoral Radiomics Approach: Differentiate Fat-Poor Angiomyolipoma from Clear Cell Renal Cell Carcinoma

Objective

This study aimed to evaluate the role of tumor and mini-peritumor in the context of CT-based radiomics analysis to differentiate fat-poor angiomyolipoma (fp-AML) from clear cell renal cell carcinoma (ccRCC).

Methods

A total of 58 fp-AMLs and 172 ccRCCs were enrolled. The volume of interest (VOI) was manually delineated in the standardized CT images and radiomics features were automatically calculated with software. After methods of feature selection, the CT-based logistic models including tumoral model (Ra-tumor), mini-peritumoral model (Ra-peritumor), perirenal model (Ra-Pr), perifat model (Ra-Pf), and tumoral+perirenal model (Ra-tumor+Pr) were constructed. The area under curves (AUCs) were calculated by DeLong test to evaluate the efficiency of logistic models.

Results

The AUCs of Ra-peritumor of nephrographic phase (NP) were slightly higher than those of corticomedullary phase (CMP). Furthermore, the Ra-Pr showed significant higher efficiency than the Ra-Pf, and relative more optimal radiomics features were selected in the Ra-Pr than Ra-Pf. The Ra-tumor+Pr combined tumoral and perirenal radiomics analysis was of most significant in distinction compared with Ra-tumor and Ra-peritumor.

Conclusion

The validity of NP to differentiate fp-AML from ccRCC was slightly higher than that of CMP. To the NP analysis, the Ra-Pr was superior to the Ra-Pf in distinction, and the lesions invaded to the perirenal tissue more severely than to the perifat tissue. It is important to the individual therapeutic surgeries according to the different lesion location. The pooled tumoral and perirenal radiomics analysis was the most promising approach in distinguishing fp-AML and ccRCC.

Use of radiomics based on 18F-FDG PET/CT and machine learning methods to aid clinical decision-making in the classification of solitary pulmonary lesions: an innovative approach

PURPOSE

This study was designed and performed to assess the ability of ¹⁸F-fluorodeoxyglucose (FDG) positron emission tomography (PET) and computed tomography (CT) radiomics features combined with machine learning methods to differentiate between primary and metastatic lung lesions and to classify histological subtypes. Moreover, we identified the optimal machine learning method.

METHODS

A total of 769 patients pathologically diagnosed with primary or metastatic lung cancers were enrolled. We used the LIFEx package to extract radiological features from semiautomatically segmented PET and CT images within the same volume of interest. Patients were randomly distributed in training and validation sets. Through the evaluation of five feature selection methods and nine classification methods, discriminant models were established. The robustness of the procedure was controlled by tenfold cross-validation. The model's performance was evaluated using the area under the receiver operating characteristic curve (AUC).

RESULTS

Based on the radiomics features extracted from PET and CT images, forty-five discriminative models were established. Combined with appropriate feature selection methods, most classifiers showed excellent discriminative ability with AUCs greater than 0.75. In the differentiation between primary and metastatic lung lesions, the feature selection method gradient boosting decision tree (GBDT) combined with the classifier GBDT achieved the highest classification AUC of 0.983 in the PET dataset. In contrast, the feature selection method eXtreme gradient boosting combined with the classifier random forest (RF) achieved the highest AUC of 0.828 in the CT dataset. In the discrimination between squamous cell carcinoma and adenocarcinoma, the combination of GBDT feature selection method with GBDT classification had the highest AUC of 0.897 in the PET dataset. In contrast, the combination of the GBDT feature selection method with the RF classification had the highest AUC of 0.839 in the CT dataset. Most of the decision tree (DT)-based models were overfitted, suggesting that the classification method was not appropriate for practical application.

CONCLUSION

¹⁸F-FDG PET/CT radiomics features combined with machine learning methods can distinguish between primary and metastatic lung lesions and identify histological subtypes in lung cancer. GBDT and RF were considered optimal classification methods for the PET and CT datasets, respectively, and GBDT was considered the optimal feature selection method in our analysis.

Enhanced computed tomography radiomics-based machine-learning methods for predicting the Fuhrman grades of renal clear cell carcinoma

OBJECTIVE

To develop and test an optimal machine learning model based on the enhanced computed tomography (CT) to preoperatively predict pathological grade of clear cell renal cell carcinoma (ccRCC).

METHODS

A retrospective analysis of 53 pathologically confirmed cases of ccRCC was performed and 25 consecutive ccRCC cases were selected as a prospective testing set. All patients underwent routine preoperative abdominal CT plain and enhanced scans. Renal tumor lesions were segmented on arterial phase images and 396 radiomics features were extracted. In the training set, seven discrimination classifiers for high- and low-grade ccRCCs were constructed based on seven different machine learning models, respectively, and their performance and stability for predicting ccRCC grades were evaluated through receiver operating characteristic (ROC) analysis and cross-validation. Prediction accuracy and area under ROC curve were used as evaluation indices. Finally, the diagnostic efficacy of the optimal model was verified in the testing set.

RESULTS

The accuracies and AUC values achieved by support vector machine with radial basis function kernel (svmRadial), random forest and naïve Bayesian models were 0.860±0.158 and 0.919±0.118, 0.840±0.160 and 0.915±0.138, 0.839±0.147 and 0.921±0.133, respectively, which showed high predictive performance, whereas K-nearest neighborhood model yielded lower accuracy of 0.720±0.188 and lower AUC value of 0.810±0.150. Additionally, svmRadial had smallest relative standard deviation (RSD, 0.13 for AUC, 0.17 for accuracy), which indicates higher stability.

CONCLUSION

svmRadial performs best in predicting pathological grades of ccRCC using radiomics features computed from the preoperative CT images, and thus may have high clinical potential in guiding preoperative decision.

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.

Prediction of Benign and Malignant Solid Renal Masses: Machine Learning-Based CT Texture Analysis

RATIONALE AND OBJECTIVES

This study aimed to investigate whether benign and malignant renal solid masses could be distinguished through machine learning (ML)-based computed tomography (CT) texture analysis.

MATERIALS AND METHODS

Seventy-nine patients with 84 solid renal masses (21 benign; 63 malignant) from a single center were included in this retrospective study. Malignant masses included common renal cell carcinoma (RCC) subtypes: clear cell RCC, papillary cell RCC, and chromophobe RCC. Benign masses are represented by oncocytomas and fat-poor angiomyolipomas. Following preprocessing steps, a total of 271 texture features were extracted from unenhanced and contrast-enhanced CT images. Dimension reduction was done with a reliability analysis and then with a feature selection algorithm. A nested-approach was used for feature selection, model optimization, and validation. Eight ML algorithms were used for the classifications: decision tree, locally weighted learning, k-nearest neighbors, naive Bayes, logistic regression, support vector machine, neural network, and random forest.

RESULTS

The number of features with good reproducibility was 198 for unenhanced CT and 244 for contrast-enhanced CT. Random forest algorithm demonstrated the best predictive performance using five selected contrast-enhanced CT texture features. The accuracy and area under the curve metrics were 90.5% and 0.915, respectively. Having eliminated the highly collinear features from the analysis, the accuracy and area under the curve values slightly increased to 91.7% and 0.916, respectively.

CONCLUSION

ML-based contrast-enhanced CT texture analysis might be a potential method for distinguishing benign and malignant solid renal masses with satisfactory performance.

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.

Importance of phase enhancement for machine learning classification of solid renal masses using texture analysis features at multi-phasic CT

OBJECTIVE

To compare machine learning (ML) of texture analysis (TA) features for classification of solid renal masses on non-contrast-enhanced CT (NCCT), corticomedullary (CM) and nephrographic (NG) phase contrast-enhanced (CE) CT.

MATERIALS AND METHODS

With IRB approval, we retrospectively identified 177 consecutive solid renal masses (116 renal cell carcinoma [RCC]; 51 clear cell [cc], 40 papillary, 25 chromophobe and 61 benign tumors; 49 oncocytomas and 12 fat-poor angiomyolipomas) with renal protocol CT between 2012 and 2017. Tumors were independently segmented by two blinded radiologists. Twenty-five 2-dimensional TA features were extracted from each phase. Diagnostic accuracy for 1) RCC versus benign tumor and 2) cc-RCC versus other tumor was assessed using XGBoost.

RESULTS

ML of texture analysis features on different phases achieved mean area under the ROC curve (AUC [SD]), sensitivity/specificity for 1) RCC vs benign = 0.70(0.19), 96%/32% on CM-CECT and 0.71(0.14), 83%/58% on NG-CECT and; 2) cc-RCC vs other = 0.77(0.12), 49%/90% on CM-CECT and 0.71(0.16), 22%/94% on NG-CECT. There was no difference in AUC comparing CECT to NCCT (p = 0.058-0.54) and no improvement when combining data across all three phases compared single-phase assessment (p = 0.39-0.68) for either outcome. AUCs decreased when ML models were trained with one phase and tested on a different phase for both outcomes (RCC;p = 0.045-0.106, cc-RCC; < 0.001).

CONCLUSION

Accuracy of machine learning classification of renal masses using texture analysis features did not depend on phase; however, models trained using one phase performed worse when tested on another phase particularly when associating NCCT and CECT. These findings have implications for large registries which use varying CT protocols to study renal masses.