Lesion Size and Iodine Quantification to Distinguish Low-Grade From High-Grade Clear Cell Renal Cell Carcinoma Using Dual-Energy Spectral Computed Tomography. J Comput Assist Tomogr. 2016;40:673-677. [PMID: 27224223 DOI: 10.1097/rct.0000000000000441]

Dual-energy CT-based radiomics for predicting pathological grading of invasive lung adenocarcinoma

AIMS

The purpose of the study was to build a radiomics model using Dual-energy CT (DECT) to predict pathological grading of invasive lung adenocarcinoma.

MATERIALS AND METHODS

The retrospective study enrolled 107 patients (80 low-grade and 27 high-grade) with invasive lung adenocarcinoma before surgery. Clinical features, radiographic characteristics, and quantitative parameters were measured. Virtual monoenergetic images at 50kev and 150kev were reconstructed for extracting DECT radiomics features. To select features for constructing models, Pearson's correlation analysis, intraclass correlation coefficients, and least absolute shrinkage and selection operator penalized logistic regression were performed. Four models, including the DECT radiomics model, the clinical-DECT model, the conventional CT radiomics model, and the mixed model, were established. Area under the curve (AUC) and decision curve analysis were used to measure the performance and the clinical value of the models.

RESULTS

The radiomics model based on DECT exhibited outstanding performance in predicting tumor differentiation, with an AUC of 0.997 and 0.743 in the training and testing sets, respectively. Incorporating tumor density, lobulation, and effective atomic number at AP, the clinical-DECT model showed a comparable performance with an AUC of 0.836 in both the training and testing sets. In comparison to the conventional CT radiomics model (AUC of 0.998 in the training and 0.529 in the testing set) and the mixed model (AUC of 0.988 in the training and 0.707 in the testing set), the DECT radiomics model demonstrated a greater AUC value and provided patients with a more significant net benefit in the testing set.

CONCLUSIONS

In contrast to the conventional CT radiomics model, the DECT radiomics model produced greater predictive performance in pathological grading of invasive lung adenocarcinoma.

CT-based radiomics model using stability selection for predicting the World Health Organization/International Society of Urological Pathology grade of clear cell renal cell carcinoma

OBJECTIVES

This study aimed to develop a model to predict World Health Organization/International Society of Urological Pathology (WHO/ISUP) low-grade or high-grade clear cell renal cell carcinoma (ccRCC) using 3D multiphase enhanced CT radiomics features (RFs).

METHODS

CT data of 138 low-grade and 60 high-grade ccRCC cases were included. RFs were extracted from four CT phases: non-contrast phase (NCP), corticomedullary phase, nephrographic phase, and excretory phase (EP). Models were developed using various combinations of RFs and subjected to cross-validation.

RESULTS

There were 107 RFs extracted from each phase of the CT images. The NCP-EP model had the best overall predictive value (AUC = 0.78), but did not significantly differ from that of the NCP model (AUC = 0.76). By considering the predictive ability of the model, the level of radiation exposure, and model simplicity, the overall best model was the Conventional image and clinical features (CICFs)-NCP model (AUC = 0.77; sensitivity 0.75, specificity 0.69, positive predictive value 0.85, negative predictive value 0.54, accuracy 0.73). The second-best model was the NCP model (AUC = 0.76).

CONCLUSIONS

Combining clinical features with unenhanced CT images of the kidneys seems to be optimal for prediction of WHO/ISUP grade of ccRCC. This noninvasive method may assist in guiding more accurate treatment decisions for ccRCC.

ADVANCES IN KNOWLEDGE

This study innovatively employed stability selection for RFs, enhancing model reliability. The CICFs-NCP model's simplicity and efficacy mark a significant advancement, offering a practical tool for clinical decision-making in ccRCC management.

Nomogram combining pre-operative clinical characteristics and spectral CT parameters for predicting the WHO/ISUP pathological grading in clear cell renal cell carcinoma

PURPOSE

To develop a novel clinical-spectral-computed tomography (CT) nomogram incorporating clinical characteristics and spectral CT parameters for the preoperative prediction of the WHO/ISUP pathological grade in clear cell renal cell carcinoma (ccRCC).

METHODS

Seventy-three ccRCC patients who underwent spectral CT were included in this retrospective analysis from December 2020 to June 2023. The subjects were pathologically divided into low- and high-grade groups (WHO/ISUP 1/2, n = 52 and WHO/ISUP 3/4, n = 21, respectively). Information on clinical characteristics, conventional CT imaging features, and spectral CT parameters was collected. Multivariate logistic regression analyses were conducted to create a nomogram combing clinical data and image data for preoperatively predicting the pathological grade of ccRCC, and the area under the curve (AUC) was utilized to assess the predictive performance of the model.

RESULTS

Multivariate logistic regression analyses revealed that age, systemic immune-inflammation index (SII), and the slope of the spectrum curve in the cortex phase (CP-K) were independent predictors for predicting high-grade ccRCC. The clinical-spectral-CT model exhibited high evaluation efficacy, with an AUC of 0.933 (95% confidence interval [CI]: 0.878-0.998; sensitivity: 0.810; specificity: 0.923). The calibration curve revealed that the predicted probability of the clinical-spectral-CT nomogram could better fit the actual probability, with high calibration. The Hosmer-Lemeshow test showed that the model had a good fitness (χ2 = 5.574, p = 0.695).

CONCLUSION

The clinical-spectral-CT nomogram has the potential to predict WHO/ISUP grading of ccRCC preoperatively.

Spectral CT imaging versus conventional CT post-processing technique in differentiating malignant and benign renal tumors

OBJECTIVE

Quantitative comparison of the diagnostic value of spectral CT imaging and conventional CT post-processing technique in differentiating malignant and benign renal tumors.

METHODS

A total of 209 patients with renal tumors who had undergone CT enhancement were assigned to three groups-clear cell renal cell carcinoma (ccRCC, n = 106), non-ccRCC (n = 60), and benign renal tumor (n = 43) groups. Parametric CT enhancement of each tumor based on spectral CT and conventional CT was performed using in-house software, and the iodine concentration, water content, slope, and density values among the three groups were compared. The receiver operating characteristic (ROC) curve analysis was performed to determine the optimum diagnostic thresholds, the area under the ROC curve (AUC), sensitivity, specificity, and accuracy of the above parameters.

RESULTS

The iodine concentration, slope, and density values were higher in the ccRCCs group compared to the non-ccRCCs and benign renal tumor groups (p < 0.05). Moreover, the iodine concentration, slope, and density values were higher in benign renal tumors compared to non-ccRCCs (p < 0.05). According to the ROC curve analysis, iodine concentration presented the highest diagnostic efficacy in differentiating ccRCCs/non-ccRCCs from benign renal tumors. The pairwise comparisons of the ROC curves and the diagnostic efficacies revealed that ROI-based CT enhancement was worse than the spectral CT imaging analysis in terms of density (p < 0.05).

CONCLUSION

Iodine concentration presented the highest diagnostic efficacy in differentiating ccRCCs/non-ccRCCs from benign renal tumors.

ADVANCES IN KNOWLEDGE

1. The iodine concentration, slope, and density values were higher for the ccRCCs compared to non-ccRCCs and benign renal tumors.2. Iodine concentration presented the highest diagnostic efficacy in differentiating ccRCCs/non-ccRCCs from benign renal tumors.3. Spectral CT imaging analysis performed better than conventional CT in differentiating malignant and benign renal tumors.

Accuracy of iodine quantification using dual-energy computed tomography with focus on low concentrations

BACKGROUND

Iodine quantification using dual-energy computed tomography (DECT) is helpful in characterizing, and follow-up after treatment of tumors. Some malignant masses, for instance papillary renal cell carcinomas (p-RCC), are hard to differentiate from benign lesions because of very low contrast enhancement. In these cases, iodine concentrations might be very low, and it is therefore important that iodine quantification is reliable even at low concentrations if this technique is used.

PURPOSE

To examine the accuracy of iodine quantification and to determine whether it is also accurate for low iodine concentrations.

MATERIAL AND METHODS

Twenty-six syringes with different iodine concentrations (0-30 mg I/mL) were scanned in a phantom model using a DECT scanner with two different kilovoltage and image reconstruction settings. Iodine concentrations were measured and compared to known concentration. Absolute and relative errors were calculated.

RESULTS

For concentrations of 1 mg I/mL or higher, there was an excellent correlation between true and measured iodine concentrations for all settings (R = 0.999-1.000; P < 0.001). For concentrations <1.0 mg I/mL, the relative error was greater. Absolute and relative errors were smaller using tube voltages of 80/Sn140 kV than 100/Sn140 kV (P < 0.01). Reconstructions using a 3.0-mm slice thickness had less variance between repeated acquisitions versus 0.6 mm (P < 0.001).

CONCLUSION

Iodine quantification using DECT was in general very accurate, but for concentrations < 1.0 mg I/mL the technique was less reliable. Using a tube voltage with larger spectral separation was more accurate and the result was more reproducible using thicker image reconstructions.

Advances in the application of artificial intelligence in solid tumor imaging

Early diagnosis and timely treatment are crucial in reducing cancer-related mortality. Artificial intelligence (AI) has greatly relieved clinical workloads and changed the current medical workflows. We searched for recent studies, reports and reviews referring to AI and solid tumors; many reviews have summarized AI applications in the diagnosis and treatment of a single tumor type. We herein systematically review the advances of AI application in multiple solid tumors including esophagus, stomach, intestine, breast, thyroid, prostate, lung, liver, cervix, pancreas and kidney with a specific focus on the continual improvement on model performance in imaging practice.

Spectral CT in Lung Cancer: Usefulness of Iodine Concentration for Evaluation of Tumor Angiogenesis and Prognosis

OBJECTIVE. The purpose of this study was to investigate the correlation between iodine concentration (IC) derived from spectral CT and angiogenesis and the relationships between IC and clinical-pathologic features associated with lung cancer prognosis. SUBJECTS AND METHODS. Sixty patients with lung cancer were enrolled and underwent spectral CT. The IC, IC difference (ICD), and normalized IC (NIC) of tumors were measured in the arterial phase, venous phase (VP), and delayed phase. The microvessel densities (MVDs) of CD34-stained specimens were evaluated. Correlation analysis was performed for IC and MVD. The relationships between the IC index showing the best correlations with MVD and clinical-pathologic findings of pathologic types, histologic differentiation, tumor size, lymph node status, pathologic TNM stage, and intratumoral necrosis were investigated. RESULTS. The mean (± IQR) MVD of all tumors was 42.00 ± 27.50 vessels per field at ×400 magnification, with two MVD distribution types. The MVD of lung cancer correlated positively with the IC, ICD, and NIC on three-phase contrast-enhanced scanning (r range, 0.581-0.800; all p < 0.001), and the IC in the VP showed the strongest correlation with MVD (r = 0.800; p < 0.001). The correlations between IC and MVD, ICD and MVD, and NIC and MVD varied depending on whether the same scanning phase or same IC index was used. The IC in the VP showed statistically significant differences in the pathologic types of adenocarcinoma and squamous cell carcinoma, histologic differentiation, tumor size, and status of intratumoral necrosis of lung cancer (p < 0.05), but was not associated with nodal metastasis and pathologic TNM stages (p > 0.05). CONCLUSION. IC indexes derived from spectral CT, especially the IC in the VP, were useful indicators for evaluating tumor angiogenesis and prognosis.

A CT-based deep learning model for predicting the nuclear grade of clear cell renal cell carcinoma

PURPOSE

To investigate the effects of different methodologies on the performance of deep learning (DL) model for differentiating high- from low-grade clear cell renal cell carcinoma (ccRCC).

METHOD

Patients with pathologically proven ccRCC diagnosed between October 2009 and March 2019 were assigned to training or internal test dataset, and external test dataset was acquired from The Cancer Genome Atlas-Kidney Renal Clear Cell Carcinoma (TCGA-KIRC) database. The effects of different methodologies on the performance of DL-model, including image cropping (IC), setting the attention level, selecting model complexity (MC), and applying transfer learning (TL), were compared using repeated measures analysis of variance (ANOVA) and receiver operating characteristic (ROC) curve analysis. The performance of DL-model was evaluated through accuracy and ROC analyses with internal and external tests.

RESULTS

In this retrospective study, patients (n = 390) from one hospital were randomly assigned to training (n = 370) or internal test dataset (n = 20), and the other 20 patients from TCGA-KIRC database were assigned to external test dataset. IC, the attention level, MC, and TL had major effects on the performance of the DL-model. The DL-model based on the cropping of an image less than three times the tumor diameter, without attention, a simple model and the application of TL achieved the best performance in internal (ACC = 73.7 ± 11.6%, AUC = 0.82 ± 0.11) and external (ACC = 77.9 ± 6.2%, AUC = 0.81 ± 0.04) tests.

CONCLUSIONS

CT-based DL model can be conveniently applied for grading ccRCC with simple IC in routine clinical practice.

Stepwise analysis of potential accuracy-influencing factors of iodine quantification on a fast kVp-switching second-generation dual-energy CT: from 3D-printed phantom to a simple solution in clinical routine use

Background

Measurement of iodine concentration from dual-energy or spectral computed tomography (CT) provides useful diagnostic information especially in patients suffering from malignant tumors of various origins.

Purpose

The purpose of this study was to systematically investigate the accuracy of the measurement of iodine concentration, focusing on potential influencing factors and assessing its suitability for routine clinical use.

Material and Methods

First, a 3D-printed cylindrical phantom was used to assess reliability of dual-energy CT-based iodine concentration measurement. Second, a semi-anthropomorphic phantom was used to evaluate the potential impact of positional variation of the target volume as typically seen in clinical scans. Finally, a reference vial was placed on the body surface of 38 patients undergoing abdominal dual-energy CT to analyze correlations between applied doses and patient diameters.

Results

The position of the target volume within the cylindrical phantom and the applied dose level significantly influenced the magnitude of measured iodine concentrations ( P < 0.001). We also found a significant difference in accuracy depending on target volume position in the semi-anthropomorphic phantom ( P = 0.028). In patient scans, we observed an error of 19.6 ± 5.6% in iodine concentration measurements of a reference and significant, moderate to strong, negative correlations between measured iodine concentration, maximum patient diameter, and applied dose (maximum sagittal diameter: r = −0.455, P = 0.004; maximum coronal diameter: r=−0.517, P = 0.001; CTDIvol: r = −0.385, P = 0.017)

Conclusion

Dual-energy CT-based iodine concentration measurement should be interpreted with caution. In clinical examinations, placement of a reference vial could be a potential solution to relativize errors.

Predicting the ISUP grade of clear cell renal cell carcinoma with multiparametric MR and multiphase CT radiomics

OBJECTIVE

To investigate externally validated magnetic resonance (MR)-based and computed tomography (CT)-based machine learning (ML) models for grading clear cell renal cell carcinoma (ccRCC).

MATERIALS AND METHODS

Patients with pathologically proven ccRCC in 2009-2018 were retrospectively included for model development and internal validation; patients from another independent institution and The Cancer Imaging Archive dataset were included for external validation. Features were extracted from T1-weighted, T2-weighted, corticomedullary-phase (CMP), and nephrographic-phase (NP) MR as well as precontrast-phase (PCP), CMP, and NP CT. CatBoost was used for ML-model investigation. The reproducibility of texture features was assessed using intraclass correlation coefficient (ICC). Accuracy (ACC) was used for ML-model performance evaluation.

RESULTS

Twenty external and 440 internal cases were included. Among 368 and 276 texture features from MR and CT, 322 and 250 features with good to excellent reproducibility (ICC ≥ 0.75) were included for ML-model development. The best MR- and CT-based ML models satisfactorily distinguished high- from low-grade ccRCCs in internal (MR-ACC = 73% and CT-ACC = 79%) and external (MR-ACC = 74% and CT-ACC = 69%) validation. Compared to single-sequence or single-phase images, the classifiers based on all-sequence MR (71% to 73% in internal and 64% to 74% in external validation) and all-phase CT (77% to 79% in internal and 61% to 69% in external validation) images had significant increases in ACC.

CONCLUSIONS

MR- and CT-based ML models are valuable noninvasive techniques for discriminating high- from low-grade ccRCCs, and multiparameter MR- and multiphase CT-based classifiers are potentially superior to those based on single-sequence or single-phase imaging.

KEY POINTS

• Both the MR- and CT-based machine learning models are reliable predictors for differentiating high- from low-grade ccRCCs. • ML models based on multiparameter MR sequences and multiphase CT images potentially outperform those based on single-sequence or single-phase images in ccRCC grading.

FORCE dual-energy CT in pathological grading of clear cell renal cell carcinoma

The aim of the present study was to examine the value of FORCE dual-energy CT in grading the clear cell renal cell carcinoma (ccRCC). A total of 35 cases of ccRCC were included. Hematoxylin and eosin staining was performed, and the cases were divided into low- (Fuhrman I-II) and high-grade (Fuhrman III-IV) groups. FORCE dual-energy CT parameters, including virtual network computing CT value (VNCV), iodine overlay value (IOV), mixed energy CT value (MEV), iodine concentration (IC), normalized iodine concentration (NIC), NIC based on aorta (NICA), NIC based on cortex (NICC) and NIC based on medulla (NICM), were analyzed and compared. Receiver operating characteristic analysis was also performed. There were significant differences in the arterial phase IOV, MEV and IC, and the venous phase IOV and IC between the low- and high-grade groups. No significant differences were observed in VNCV and MEV between the low -and high-grade groups in the venous phase. Significant differences were observed in the NICA and NICC between these two groups, however no difference was observed in NICM. There were significant differences in the tumor CT values for the arterial phase at the 40, 60, 80 and 100 kiloelectron volt (keV) between the low- and high-grade groups, while no significant differences were observed at the 120-140 keV levels. The k-slope for the low-grade group was significantly higher than the high-grade group. In addition, the area under curve for the arterial phase IOV, arterial phase MEV, arterial phase IC, aortic NIC, cortical NIC, venous phase IOV, venous phase IC and curve slope K of mono-energy CT value suggested high value in diagnosis of low- and high-grade ccRCC cases.

CT imaging of solid renal masses: pitfalls and solutions

Computed tomography (CT) remains the first-line imaging test for the characterisation of renal masses; however, CT has inherent limitations, which if unrecognised, may result in errors. The purpose of this manuscript is to present 10 pitfalls in the CT evaluation of solid renal masses. Thin section non-contrast enhanced CT (NECT) is required to confirm the presence of macroscopic fat and diagnosis of angiomyolipoma (AML). Renal cell carcinoma (RCC) can mimic renal cysts at NECT when measuring <20 HU, but are usually heterogeneous with irregular margins. Haemorrhagic cysts (HC) may simulate solid lesions at NECT; however, a homogeneous lesion measuring >70 HU is essentially diagnostic of HC. Homogeneous lesions measuring 20-70 HU at NECT or >20 HU at contrast-enhanced (CE) CT, are indeterminate, requiring further evaluation. Dual-energy CT (DECT) can accurately characterise these lesions at baseline through virtual NECT, iodine overlay images, or quantitative iodine concentration analysis without recalling the patient. A minority of hypo-enhancing renal masses (most commonly papillary RCC) show indeterminate or absent enhancement at multiphase CT. Follow-up, CE ultrasound or magnetic resonance imaging (MRI) is required to further characterise these lesions. Small (<3 cm) endophytic cysts commonly show pseudo-enhancement, which may simulate RCC; this can be overcome with DECT or MRI. In small (<4 cm) solid renal masses, 20% of lesions are benign, chiefly AML without visible fat or oncocytoma. Low-dose techniques may simulate lesion heterogeneity due to increased image noise, which can be ameliorated through the appropriate use of iterative reconstruction algorithms.