Abstract
Purpose
To compare published CT-based systems for small solid renal mass (SoRM) assessment, propose modifications that may increase specificity and interreader agreement, and validate the revised system.
Materials and Methods
Our retrospective study included patients with histologically confirmed SoRMs measuring ≤4 cm who underwent CT imaging (single-institution internal dataset, n = 194; external dataset from The Cancer Imaging Archive, n = 55). Two blinded radiologists (readers 1 [R1] and 2 [R2]) compared four CT systems (CT score, modified CT score, abbreviated CT score, and UCLA CT score) for diagnostic accuracy in clear cell renal cell carcinoma (ccRCC) and papillary RCC (pRCC) and for interreader agreement (Gwet agreement coefficient [AC1]). We also evaluated the addition of two decision rules to the best-performing algorithm (noncontrast CT [NCCT] attenuation ≤ 20 HU and corticomedullary phase–NCCT attenuation at two thresholds, ≤20 HU and ≤30 HU) to create a modified algorithm (CT-Score version 2.0).
Results
The abbreviated CT score had the best combination of accuracy for ccRCC (R1: 85% [95% CI: 79, 89], R2: 72% [95% CI: 65, 78]) and pRCC (R1: 86% [95% CI: 80, 91], R2: 86% [95% CI: 80, 91]) and interreader agreement (Gwet AC1 = 0.53). CT-Score version 2.0 (derived by adding decision rules to the abbreviated CT score) demonstrated substantial agreement (Gwet AC1 = 0.63). Specificity of CT-Score version 2.0 was higher for ccRCC (R1: 99% [95% CI: 94, 100], R2: 99% [95% CI: 94, 100] vs R1: 92% [95% CI: 84, 96], R2: 81% [95% CI: 72, 89]; P = .02, P < .001) and pRCC (R1: 100% [95% CI: 98, 100], R2: 100% [95% CI: 98, 100] vs R1: 93% [95% CI: 87, 96], R2: 93% [95% CI: 87, 96]; P = .003, P = .003) when compared with the abbreviated CT score. Validation in the external dataset showed similar results: Gwet AC1 = 0.53; specificity for ccRCC (R1: 100% [95% CI: 83, 100], R2: 100% [95% CI: 83, 100]); and specificity for pRCC (R1: 100% [95% CI: 82, 100], R2: 100% [95% CI: 92, 100]).
Conclusion
Application of CT-Score version 2.0 resulted in modest improvements in interreader agreement and high specificity for ccRCC and pRCC diagnosis.
Keywords: CT, Kidney, Urinary, Oncology, Renal Mass, Algorithm, Clear Cell RCC, Papillary RCC
Supplemental material is available for this article.
© RSNA, 2026
An earlier incorrect version of this article appeared online. This article was corrected on March 4, 2026.
Keywords: CT, Kidney, Urinary, Oncology, Renal Mass, Algorithm, Clear Cell RCC, Papillary RCC
Summary
CT-Score version 2.0, derived by adding decision rules to the abbreviated CT score, was developed and classified clear cell and papillary renal cell carcinoma with high specificity in internal and external datasets.
Key Points
■ Among existing CT algorithms, the abbreviated CT score, based on the contrast-enhanced corticomedullary (CM) phase mass to cortex (MC) ratio, had the best combination of reproducibility (Gwet agreement coefficient [AC1] = 0.53) and accuracy for classifying clear cell renal cell carcinoma (ccRCC) (reader 1 [R1]: 85%, reader 2 [R2]: 72%) and papillary RCC (pRCC) (R1: 86%, R2: 86%).
■ Two additional CT variables (attenuation at noncontrast CT [NCCT] ≤ 20 HU and at CM-NCCT ≤ 20 HU) were specific and associated with the diagnosis of ccRCC (R1: odds ratio = 8.32, P = .045; R2: odds ratio = 13.00, P = .014) and pRCC (R1: odds ratio = 179.1, P < .001; R2: odds ratio = 227.18, P < .001), respectively.
■ A modified five-tiered CT algorithm (CT-Score version 2.0), with the MC ratio with NCCT attenuation ≤ 20 HU and CM-NCCT at two thresholds, ≤ 20 HU and ≤ 30 HU, added as decision rules, was reproducible and highly specific for ccRCC and pRCC diagnosis, with satisfactory overall diagnostic accuracy in internal and external datasets (internal dataset: Gwet AC1 = 0.63, accuracy for ccRCC = R1: 81%, R2: 74%; accuracy for pRCC = R1: 94%, R2: 94%; external dataset: Gwet AC1 = 0.53, accuracy for ccRCC = R1: 71%, R2: 69%; accuracy for pRCC = R1: 95%, R2: 89%).
Introduction
The characterization of solid renal masses (SoRMs) using CT and MRI is desirable, as renal masses are common incidental imaging findings (1,2). While most SoRMs (defined as the presence of >25% enhancing tissue [3]) are malignant, small (≤4 cm) solid masses can be benign in up to 20% of cases (4,5). Moreover, cT1a (≤4 cm) renal cell carcinoma (RCC) typically follows an indolent course, with only rare (~3%) instances of local recurrence or metastasis (6). With evidence supporting active surveillance of small SoRMs in selected patients (7), determining the nature of an SoRM to help guide therapeutic decisions is becoming increasingly important. The American College of Radiology has approved the Kidney Imaging Reporting and Data System (KI-RADS) as a work in progress (8). The aim of KI-RADS is to determine the probability of malignancy and the likelihood of aggressive disease within malignant masses to help guide therapy (9). As the proposed KI-RADS is an algorithm that intends to predict the likelihood of aggressive malignancy within a renal mass, standardized assessment strategies are required. Cystic renal masses are characterized with the Bosniak classification (10); however, there is no standardized characterization approach for SoRMs.
The clear cell likelihood score (ccLS) system is a five-tiered Likert scale that estimates the likelihood that an SoRM is a clear cell RCC (ccRCC) (11). The ccLS is generated through the interpretation of multiparametric MRI by using a standardized diagnostic algorithm (12). The ccLS system has been shown in multicenter studies to diagnose ccRCC with moderate accuracy, high negative predictive value, and moderate interreader agreement (13–15). Although not perfect, the ccLS system is the best progenitor system for evaluating SoRMs via multiparametric MRI. No similar standardized reporting system has been established for the evaluation of SoRMs with CT. At present, there are four published competing algorithms for CT assessment of SoRMs: the algorithm proposed by Al Nasibi et al (16) (hereafter, “CT score”), the algorithm by Eldehimi et al (17,18) (hereafter, “modified CT score”), the algorithm by Eldehimi et al (18) and Lemieux et al (19) (hereafter, “abbreviated CT score”), and the algorithm by Chung and Raman (20) (hereafter, “UCLA CT score”). Results published to date indicate moderate accuracy for the diagnosis of ccRCC, high accuracy for the diagnosis of papillary RCC (pRCC), and modest interrater agreement (16–19).
The purpose of our study was to compare the performance and interreader agreement of the four published CT-based algorithms for the diagnosis of RCC among small (≤4 cm) SoRMs. The secondary aims were to determine whether the addition of decision rules, as included in the MRI-based ccLS system, to the best-performing CT algorithm could increase specificity for the diagnosis of RCC and to validate this revised system in an external dataset.
Materials and Methods
Patients
We received institutional review board approval for our Health Insurance Portability and Accountability Act–compliant retrospective study and a waiver of the requirement to obtain written informed patient consent.
Internal dataset.
A fellowship-trained genitourinary radiologist (S.K.) with 12 years of posttraining experience searched an institutional renal mass database for consecutive SoRMs in adult patients (age ≥ 18 years) who underwent imaging evaluation via CT before surgical resection and received a surgical histopathology diagnosis from January 2016 to December 2021. A total of 467 patients were identified in this search. We excluded masses that were not ≤4 cm (n = 164) and masses for which surgery was performed more than 12 months after CT (n = 34). The remaining patients had their CT examinations reviewed by the genitourinary radiologist. On the basis of this review, we excluded patients for the following reasons: a complete renal mass CT protocol was not used (n = 51), the mass was cystic (<25% enhancing internal elements) (n = 24), and patients had a known or suspected hereditary kidney cancer syndrome (16 masses in eight patients).
These exclusions resulted in 178 small SoRMs being evaluated via a multiparametric renal mass CT protocol and having a subsequent histopathology diagnosis following surgical resection. Surgical histopathology was chosen as the reference standard owing to the limitations of percutaneous biopsy sample analysis in subtyping RCCs, especially oncocytic renal neoplasms (21). However, the use of surgical histopathology as the reference standard generally precludes the inclusion of fat-poor angiomyolipomas owing to frequent minimally invasive determination with imaging and biopsy. As a definitive diagnosis of angiomyolipoma can be established via renal mass biopsy (22), we queried another institutional biopsy database of all consecutive renal mass biopsies that were performed during the same time period, with a histopathology diagnosis of angiomyolipoma. From a possible 30 consecutive masses, nine patients were excluded because a complete renal mass CT protocol was not performed, and five patients were excluded because biopsy was performed more than 12 months after CT. Thus, the final study sample for the internal dataset included a total of 194 histologically confirmed small SoRMs (including 16 fat-poor angiomyolipomas diagnosed via biopsy and 178 masses diagnosed via surgical histopathology) evaluated preoperatively with a multiparametric renal mass CT protocol. A flowchart of patient selection is presented in Figure 1A.
Figure 1:
Flowcharts show patient selection for (A) internal dataset and (B) external dataset.
All CT examinations were performed for the characterization or staging of renal masses. The mean interval between CT and surgery was 4.8 months ± 3 (SD) (range, 1–12 months). No patient underwent intervening therapy between the CT examination and surgical resection. Institutional fellowship-trained genitourinary pathologists established the histologic diagnosis for each mass as part of clinical care via pathology evaluation of the specimens according to the World Health Organization criteria for renal tumors (23). Histologic diagnoses were retrieved from the pathology reports.
External dataset.
We used the C4KC-KiTS dataset (version 3, updated June 18, 2020) from The Cancer Imaging Archive (TCIA) for external validation of the tested algorithms (24). This dataset contains the CT scans of renal masses along with the final diagnosis after nephrectomy from 210 patients from the training set of the 2019 Kidney and Kidney Tumor Segmentation Challenge (ie, KiTS19) (25). The fellowship-trained genitourinary radiologist reviewed the dataset to exclude masses larger than 4 cm (n = 88) and masses for which a complete renal mass CT protocol was not conducted (n = 67). Thus, the final study sample for the external dataset included 55 histologically confirmed small SoRMs evaluated via multiparametric CT. A flowchart of patient selection is presented in Figure 1B.
CT Protocol
In the internal dataset, 163 of the 194 patients had their CT examinations performed at the Joint Department of Medical Imaging, Toronto, Ontario, Canada, a tertiary care academic center. In the remaining 31 patients, the CT examinations were performed at institutions within the regional health network. All examinations were reviewed on the same picture archiving and communication system (Coral Workstation version 3.8; Joint Department of Medical Imaging). The renal mass CT protocol examinations were compliant with recommendations from the Society of Abdominal Radiology Disease-Focused Panel on RCC (3). Details are provided in Appendix S1.
Comparison of Existing CT Algorithms
We compared four existing CT algorithms: (a) CT score (16), a five-tiered system that includes the renal mass to cortex corticomedullary (CM) phase attenuation ratio (MC ratio) and a subjective assessment of mass CM phase heterogeneity; (b) modified CT score (17), a five-tiered system that includes all of the variables of the CT score as well as mass hyperattenuation, segmental enhancement inversion, and the arterial to delayed enhancement ratio; (c) abbreviated CT score (18,19), which adapts the CT score by eliminating subjective heterogeneity and is thus a three-tier system that is based on the MC ratio only; and (d) UCLA CT score (20), a numerical system that assigns points on the basis of various qualitative and quantitative features. Studies were independently assessed by two blinded readers with expertise in abdominal imaging (M.K. and B.N.) with 5 and 7 years of experience, respectively, with all necessary variables for each scoring system assessed at the same time. A detailed summary of how each reader applied each CT algorithm in our study is provided in Appendix S2.
For the CT score, modified CT score, and UCLA CT Score, a score of 4 or 5 is considered positive for ccRCC and a score of 1 or 2 is considered positive for pRCC. Score 3 masses are considered indeterminate (16). For the abbreviated CT score, a score of 3 is considered positive for ccRCC, a score of 1 is considered positive for pRCC, and masses with a score of 2 are considered indeterminate.
Evaluation of Two Additional Discriminatory Variables
We evaluated two additional variables that may add specificity to the proposed algorithms. First, noncontrast CT (NCCT) attenuation ≤ 20 HU within solid masses is a specific feature of RCC and, more specifically, ccRCC. In several studies of SoRMs, attenuation at NCCT ≤ 20 HU (in solid enhancing masses that are heterogeneous) has been strongly associated with RCC and, most commonly, ccRCC (26). This low attenuation at NCCT is thought to be analogous to the microscopic fat detected in ccRCC at chemical shift MRI (similar to lipid-rich adrenal adenoma) (26,27). Thus, attenuation at NCCT ≤ 20 HU could be a useful feature to increase the specificity for the diagnosis of ccRCC, similar to how microscopic fat is used in the MRI-based ccLS system to increase the specificity for the detection of ccRCC (28).
Second, absolute enhancement in the CM phase (CM-NCCT) can be used to add specificity. While the MC ratio differentiates intense (<75%), moderate (40%–75%) and mild (<40%) enhancement, multiple studies have shown large interpatient variation in renal cortex enhancement during the CM phase that is due to various factors (eg, contrast media volume and concentration, timing of contrast bolus arrival, cardiovascular status) (29–31). Moreover, several studies have shown that pRCCs exhibit low-level enhancement during the CM phase; therefore, we evaluated CM-NCCT attenuation as a specific feature of pRCC (32–34) by using previously described thresholds: (a) CM-NCCT attenuation ≤ 20 HU (ie, enhancement ≤ 20 HU in the CM phase; highest specificity for pRCC) and (b) CM-NCCT attenuation ≤ 30 HU (ie, enhancement ≤ 30 HU in the CM phase; high sensitivity and moderate specificity for pRCC) (32).
We first evaluated these two variables in the internal dataset via univariable analysis to determine whether there was a significant association with RCC diagnosis. Then, we added these variables to the best-performing existing CT algorithm in the form of decision rules to determine whether there was any improvement in diagnostic accuracy for the classification of ccRCC or pRCC.
Statistical Analysis
We tabulated the number of ccRCCs, pRCCs, and renal masses with other histopathologic diagnoses. We calculated the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy for the CT score, modified CT score, abbreviated CT score, and UCLA CT score along with 95% CIs. We compared accuracy between systems by performing the McNemar test with Bonferroni correction for six pairwise comparisons, with a P value < .008 as the threshold for significance. For all other comparisons, a P value < .05 was the threshold for significance. We assessed interreader agreement by calculating the Gwet agreement coefficient (AC1) (35). We categorized interreader agreement for the Gwet AC1 as follows: less than 0.20, poor agreement; 0.20–0.39, fair agreement; 0.40–0.59, moderate agreement; 0.60–0.79, substantial agreement; and 0.80 or greater, near perfect agreement. We selected the algorithm that showed the highest overall accuracy and interreader agreement as the best-performing algorithm.
We determined the association between attenuation at NCCT ≤ 20 HU and the diagnosis of ccRCC, and the association between the CM-NCCT attenuation ≤ 20 HU and ≤ 30 HU and the diagnosis of pRCC, using univariable analysis and calculation of odds ratio and beta coefficients. We tabulated the number of masses (RCC and other) with NCCT attenuation ≤ 20 HU and CM-NCCT attenuation ≤ 20 HU and ≤ 30 HU for each reader.
We derived a CT algorithm by using the best-performing existing algorithm coupled with the two new CT features applied as decision rules (henceforth, “CT-Score version 2.0”). We evaluated the performance of the optimized algorithm on the internal dataset and external (TCIA) datasets by using the same methods previously described. We performed statistical analysis with R studio version 2024.12.0 (R Studio Team) and R version 4.4.2. (R Foundation for Statistical Computing; http://www.R-project.org/) with help from a statistician (Y.D.).
Results
Patient and Renal Mass Characteristics
The internal dataset comprised 194 renal masses (patient mean age, 56 years ± 11; 120 male, 74 female patients), and the external dataset comprised 55 renal masses (patient mean age, 59 years ± 13; 32 male, 23 female patients). A summary of patient demographic features and histopathology diagnoses for the internal and external (TCIA) datasets is provided in Table 1.
Table 1:
Patient Demographics and Histologic Diagnosis of Renal Masses in Internal and External Datasets
| Parameter | Internal Dataset (n = 194) | External Dataset (The Cancer Imaging Archive) (n = 55) |
|---|---|---|
| Age (y)* | 56 ± 11 | 59 ± 13 |
| Sex | ||
| Male | 120 | 32 |
| Female | 74 | 23 |
| Renal mass size (cm)* | 2.6 ± 0.9 | 2.8 ± 0.8 |
| Renal mass histologic diagnosis | ||
| Clear cell | 97 (50) | 35 (63) |
| Papillary | 43 (22) | 10 (18) |
| Chromophobe | 27 (14) | 5 (9) |
| Oncocytoma | 9 (5) | 3 (5) |
| Angiomyolipoma | 16 (8) | 1 (2) |
| Clear cell papillary | 1 (1) | … |
| Collecting duct carcinoma | 1 (1) | … |
| Renal cell carcinoma unclassified | … | 1 (2) |
Note.—Unless otherwise noted, values are numbers, with percentages in parentheses.
Values are means ± SDs.
Comparison of Existing CT Algorithms
Summaries of the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy for the diagnosis of ccRCC and pRCC by the two blinded study radiologists applying the existing CT algorithms are presented in Tables 2 and 3. For the diagnosis of ccRCC, both readers demonstrated higher accuracy when applying the simplest three-tiered abbreviated CT score versus the UCLA CT score (R1: 85% [95% CI: 79, 88], R2: 72% [95% CI: 65, 78] vs R1: 56% [95% CI: 49, 63], R2: 57% [95% CI: 49, 64]; P < .001 for both reader 1 [R1] and reader 2 [R2]). There was no evidence of a difference in accuracy for either reader when applying the abbreviated CT score versus the CT score (R1: 85% [95% CI: 79, 88], R2: 72% [95% CI: 65, 78] vs R1: 77% [95% CI: 70, 83], R2: 74% [95% CI: 67, 80]; P = .045 for R1, .61 for R2). Furthermore, there was no evidence of a difference in accuracy for either reader when applying the abbreviated CT score versus the modified CT score (R1: 85% [95% CI: 79, 88]), R2: 72% [95% CI: 65, 78] vs R1: 78% [95% CI: 72, 84], R2: 78% [95% CI: 71, 83]; P = .09 for R1, .15 for R2) (Table 2). For the diagnosis of pRCC, readers achieved higher accuracy when applying the simplest three-tiered abbreviated CT score versus the UCLA CT score (R1: 86% [95% CI: 80, 91], R2: 86% [95% CI: 80, 91] vs R1: 62% [95% CI: 55, 69], R2: 64% [95% CI: 57, 71]; P < .001 for both R1 and R2). There was no evidence of a difference in accuracy for either reader when applying the abbreviated CT score versus the CT score (R1: 86% [95% CI: 80, 91], R2: 86% [95% CI: 80, 91] vs R1: 79% [95% CI: 73, 85], R2: 78% [95% CI: 71, 83]; P = .02 for R1, .06 for R2). Moreover, there was no evidence of a difference in accuracy for either reader when applying the abbreviated CT score versus the modified CT score (R1: 86% [95% CI: 80, 91], R2: 86% [95% CI: 80, 91] vs R1: 82% [95% CI: 76, 87], R2: 81% [95% CI: 75, 86]; P = .26 for R1, .15 for R2) (Table 3) (Table S1).
Table 2:
Diagnostic Accuracy for Clear Cell Renal Cell Carcinoma for Readers Applying Previously Established CT Scoring Systems for Renal Mass CT Protocol for Internal Dataset
| Scoring System | Sensitivity (%) | Specificity (%) | PPV (%) | NPV (%) | Accuracy (%)* | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| R1 | R2 | R1 | R2 | R1 | R2 | R1 | R2 | R1 | R2 | |
| CT score ≥ 4 | 63 (52, 72) | 73 (63, 82) | 91 (83, 86) | 75 (65, 83) | 87 (77, 94) | 75 (65, 83) | 71 (62, 79) | 74 (64, 82) | 77 (70, 83) | 74 (67, 80) |
| Modified CT score ≥ 4 | 68 (58, 77) | 78 (69, 86) | 89 (81, 94) | 77 (68, 85) | 86 (76, 93) | 78 (68, 85) | 74 (65, 81) | 78 (69, 86) | 78 (72, 84) | 78 (71, 83) |
| UCLA CT score ≥ 4 | 15 (9, 24) | 13 (7, 22) | 97 (91, 99) | 100 (96, 100) | 83 (59, 96) | 100 (75, 100) | 53 (46, 61) | 54 (46, 61) | 56 (49, 63) | 57 (49, 64) |
| Abbreviated CT score = 3 | 77 (68, 85) | 62 (51, 72) | 92 (84, 96) | 81 (72, 89) | 90 (82, 96) | 77 (66, 86) | 80 (72, 87) | 68 (59, 76) | 85 (79, 89) | 72 (65, 78) |
Note.—Internal dataset comprised 194 solid renal masses measuring ≤4 cm. Values in parentheses are 95% CIs. Comparisons were performed using McNemar test. NPV = negative predictive value, PPV = positive predictive value, R1 = reader 1, R2 = reader 2.
P values for comparisons of accuracy between competing systems for R1 were as follows: CT score vs modified CT score, P = .68; CT score vs UCLA CT score, P < .001; CT score vs abbreviated CT score, P = .045; modified CT score vs UCLA CT score, P < .001; modified CT score vs abbreviated CT score, P = .09; UCLA CT score vs abbreviated CT score, P < .001. For R2, P values were as follows: CT score vs modified CT score, P = .27; CT score vs UCLA CT score, P < .001; CT score vs abbreviated CT score, P = .60; modified CT score vs UCLA CT score, P < .001; modified CT score vs abbreviated CT score, P = .15; UCLA CT score vs abbreviated CT score, P < .001.
Table 3:
Diagnostic Accuracy for Papillary Renal Cell Carcinoma for Readers Applying Previously Established CT Scoring Systems for Renal Mass CT Protocol for Internal Dataset
| Scoring System | Sensitivity (%) | Specificity (%) | PPV (%) | NPV (%) | Accuracy (%)* | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| R1 | R2 | R1 | R2 | R1 | R2 | R1 | R2 | R1 | R2 | |
| CT score ≤ 2 | 74 (59, 86) | 53 (38, 69) | 81 (74, 87) | 85 (78, 90) | 52 (39, 65) | 50 (35, 65) | 92 (86, 96) | 86 (80, 92) | 79 (73, 85) | 78 (71, 83) |
| Modified CT score ≤ 2 | 72 (56, 85) | 53 (38, 69) | 85 (78, 90) | 89 (83, 93) | 57 (43, 70) | 58 (41, 73) | 91 (86, 95) | 87 (81, 92) | 82 (76, 87) | 81 (75, 86) |
| UCLA CT score ≤ 2 | 95 (84, 99) | 98 (88, 100) | 53 (45, 61) | 54 (46, 62) | 37 (28, 46) | 38 (29, 48) | 98 (91, 100) | 99 (93, 100) | 62 (55, 69) | 64 (57, 71) |
| Abbreviated CT score = 1 | 63 (47, 77) | 63 (47, 77) | 93 (87, 96) | 93 (87, 96) | 71 (54, 85) | 71 (54, 85) | 90 (84, 94) | 90 (84, 94) | 86 (80, 91) | 86 (80, 91) |
Note.—Internal dataset comprised 194 solid renal masses measuring ≤4 cm. Values in parentheses are 95% CIs. Comparisons were performed using McNemar test. NPV = negative predictive value, PPV = positive predictive value, R1 = reader 1, R2 = reader 2.
P values for comparisons of accuracy between competing systems for R1 were as follows: CT score vs modified CT score, P = .13; CT score vs UCLA CT score, P < .001; CT score vs abbreviated CT score, P = .06; modified CT score vs UCLA CT score, P < .001; modified CT score vs abbreviated CT score, P = .26; UCLA CT score vs abbreviated CT score, P < .001. For R2, P values were as follows: CT score vs modified CT score, P = .04; CT score vs UCLA CT score, P < .001; CT score vs abbreviated CT score, P = .02; modified CT score vs UCLA CT score, P < .001; modified CT score vs abbreviated CT score, P = .15; UCLA CT score vs abbreviated CT score, P < .001.
Values of Gwet AC1, indicative of interobserver agreement, for the CT score, modified CT score, abbreviated CT score, and UCLA CT score were as follows: 0.41, 0.44, 0.53, and 0.53, respectively. Interobserver agreement was lowest with the CT score, which was due to subjective and imprecise assessment of heterogeneity (Gwet AC1 for heterogeneity = 0.29 [fair]).
Evaluation of the Two Previously Described Discriminatory Variables
NCCT attenuation ≤ 20 HU in intensely and moderately enhancing masses.
Attenuation at NCCT ≤ 20 was associated with ccRCC diagnosis for both readers (R1: odds ratio = 8.32, β = 2.12, P = .045; R2: odds ratio = 13.00, β = 2.56, P = .014). When applying an NCCT attenuation ≤ 20 HU threshold for the diagnosis of ccRCC, R1 identified 12 masses (11 ccRCC, one pRCC) and R2 identified 16 masses (15 ccRCC, one pRCC).
CM-NCCT in mildly and moderately enhancing masses.
When applying the previously reported specific attenuation threshold of CM-NCCT ≤ 20 HU for the diagnosis of pRCC, R1 identified 23 masses (all pRCCs) and R2 identified 24 masses (all pRCCs). CM-NCCT attenuation ≤ 20 HU was associated with pRCC diagnosis for both readers (R1: odds ratio = 179.1, β = 5.19, P < .001; R2: odds ratio = 227.18, β = 5.43, P < .001).
When applying the previously reported, higher-sensitivity attenuation threshold of CM-NCCT ≤ 30 HU for the diagnosis of pRCC, R1 identified 44 masses (38 pRCCs, four chromophobe RCCs, one collecting duct carcinoma, and one angiomyolipoma) and R2 identified 37 masses (34 pRCCs, two chromophobe RCCs, and one ccRCC). CM-NCCT attenuation ≤ 30 HU was associated with pRCC diagnosis for both readers (R1: odds ratio = 135.11, β = 4.91, P < .001; R2: odds ratio = 137.89, β = 4.93, P < .001).
Creation of a revised CT algorithm for the diagnosis of ccRCC and pRCC among SoRMs.
We selected the abbreviated CT score as the core algorithm for subsequent modification because readers had overall higher or similar accuracy for the diagnosis of ccRCC and pRCC when applying this system, interobserver agreement was highest when readers applied this system, and this system was the simplest of the four evaluated. We implemented a priori identified decision rules by using the two new variables to develop a five-point classification system (CT-Score version 2.0). The feature of attenuation at NCCT ≤ 20 HU was applied to assign a score of 5 to moderate and intensely enhancing masses. The features of attenuation at CM-NCCT ≤ 20 HU and at CM-NCCT ≤ 30 HU were applied to assign scores of 1 and 2 to moderate and mildly enhancing masses, respectively. A summary of the resulting CT-Score version 2.0 is provided in Figure 2, and a calculator is provided at www.renalctscore.com.
Figure 2:
Algorithm for CT-Score version 2.0. AML = angiomyolipoma, CM = corticomedullary, MC ratio = mass to cortex CM phase attenuation ratio, NC = noncontrast.
Values for the diagnostic accuracy of CT-Score version 2.0 at various thresholds for ccRCC and pRCC diagnosis are presented in Table 4, and the results for the comparison with the abbreviated CT score are presented in Table S2. Compared with the abbreviated CT score, the accuracy of CT-Score version 2.0 was similar for the diagnosis of ccRCC for masses with scores ≥ 4 (R1: 81% [95% CI: 75, 87], R2: 74% [95% CI: 67, 80] vs R1: 85% [95% CI: 79, 89], R2: 72% [95% CI: 65, 78]; P =.26 for R1, P =.22 for R2), but the specificity was higher for masses with scores of 5 (R1: 99% [95% CI: 94, 100], R2: 99% [95% CI: 94, 100] vs R1: 92% [95% CI: 84, 96], R2: 81% [95% CI: 72, 89]; P =.02 for R1, P < .001 for R2). Score 5 masses were highly likely to represent ccRCC (Table S3, Fig 3; 92% [11 of 12] for R1, 94% [15 of 16] for R2) (Fig 4).
Table 4:
Diagnostic Accuracy of CT-Score Version 2.0 for Diagnosis of ccRCC and pRCC for Readers Applying Different Thresholds for Renal Mass CT Protocol for Internal Dataset
| Diagnosis and Score | Sensitivity (%) | Specificity (%) | PPV (%) | NPV (%) | Accuracy (%) | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| R1 | R2 | R1 | R2 | R1 | R2 | R1 | R2 | R1 | R2 | |
| ccRCC | ||||||||||
| 5 | 11 (6, 19) | 15 (9, 24) | 99 (94, 100) | 99 (94, 100) | 92 (62, 100) | 94 (70, 100) | 53 (45, 60) | 94 (70, 100) | 55 (48, 62) | 57 (50, 64) |
| ≥4 | 76 (67, 84) | 67 (57, 76) | 87 (78, 93) | 80 (71, 88) | 85 (76, 92) | 77 (67, 86) | 79 (70, 86) | 71 (61, 79) | 81 (75, 87) | 74 (67, 80) |
| ≥3 | 100 (96, 100) | 99 (94, 100) | 45 (35, 56) | 37 (28, 48) | 65 (56, 72) | 61 (53, 69) | 100 (92, 100) | 97 (86, 100) | 73 (66, 79) | 68 (61, 75) |
| pRCC | ||||||||||
| ≤2 | 88 (75, 96) | 79 (64, 90) | 96 (92, 99) | 98 (94, 100) | 96 (73, 85) | 92 (78, 98) | 97 (92, 99) | 94 (89, 97) | 94 (90, 97) | 94 (89, 97) |
| 1 | 53 (38, 69) | 56 (40, 71) | 100 (98, 100) | 100 (98, 100) | 100 (85, 100) | 100 (86, 100) | 88 (83, 93) | 89 (83, 93) | 90 (85, 94) | 90 (85, 94) |
Note.—Internal dataset comprised 194 solid renal masses measuring ≤4 cm. Values in parentheses are 95% CIs. Comparisons were performed using McNemar test. ccRCC = clear cell renal cell carcinoma, NPV = negative predictive value, PPV = positive predictive value, pRCC = papillary renal cell carcinoma, R1 = reader 1, R2 = reader 2.
Figure 3:
Two-dimensional box plots depict frequency distribution (in percentages) of clear cell renal cell carcinoma (ccRCC, blue), papillary RCC (pRCC, orange), and other renal masses (green) among solid renal masses ≤ 4 cm in each of the five categories of the proposed CT-Score version 2.0 algorithm for reader 1 (darker shade) and reader 2 (lighter shade) using (A) the internal dataset and (B) external (The Cancer Imaging Archive) dataset.
Figure 4:
Images in a 73-year-old male patient with a 3-cm right renal clear cell renal cell carcinoma. Axial (A) noncontrast, (B) corticomedullary, and (C) nephrographic phase CT images show renal mass with intense enhancement (mass to cortex ratio > 75%), with noncontrast attenuation less than 20 HU, with CT-Score version 2.0 score of 5.
CT-Score version 2.0 had higher accuracy for the diagnosis of pRCC when a threshold score ≤ 2 was used (R1: 94% [95% CI: 90, 97], R2: 94% [95% CI: 90, 97] vs R1: 86% [95% CI: 80, 91], R2: 86% [95% CI: 80, 91]; P =.002 for R1, P =.004 for R2). CT-Score version 2.0 also had higher specificity for score 1 masses compared with the abbreviated CT score (R1: 100% [95% CI: 98, 100], R2: 100% [95% CI: 98, 100] vs R1: 93% [95% CI: 87, 96], R2: 93% [95% CI: 87, 96]; P =.003 for R1, P =.003 for R2). Score 1 masses were highly likely to represent pRCC (Table S3, Fig 3; 100% [23 of 23] for R1, 100% [24 of 24] for R2) (Fig 5).
Figure 5:
Images in a 51-year-old man with a 2.5-cm left renal papillary renal cell carcinoma. Axial (A) noncontrast, (B) corticomedullary, and (C) nephrographic phase CT images show an enhancing renal mass with mild enhancement (mass to cortex ratio < 40%), with an attenuation difference between the corticomedullary and noncontrast phases less than 20 HU, with CT-Score version 2.0 score of 1.
The percentages of score 3 masses (ie, indeterminate) were 32% (63 of 194) for R1 and 38% (73 of 194) for R2. Details are presented in Table S3.
The interobserver agreement on CT-Score version 2.0 for the two readers was substantial (Gwet AC1 = 0.63).
Validation of CT-Score version 2.0 on the external dataset.
Metrics for the performance of CT-Score version 2.0 on the external dataset are presented in Table 5. The interobserver agreement was moderate (Gwet AC1 = 0.53). CT-Score version 2.0 was moderately accurate for the diagnosis of ccRCC when a threshold score of ≥4 was used (71% [95% CI: 57, 82] for R1, 69% [95% CI: 55, 81] for R2) but highly specific when a score of 5 was used (100% [95% CI: 83, 100] for both readers). CT-Score version 2.0 was highly accurate for the diagnosis of pRCC when a score ≤ 2 was used (95% [95% CI: 85, 99] for R1, 89% [95% CI: 78, 96] for R2) and highly specific when a score of 1 was used (100% [95% CI: 82, 100] for R1, 100% [95% CI: 92, 100] for R2). Details are presented in Table S3 and Figure 3.
Table 5:
Diagnostic Accuracy of CT-Score Version 2.0 for Diagnosis of pRCC and ccRCC for Readers Applying Different Thresholds for Renal Mass CT Protocol for External Dataset
| Diagnosis and Score | Sensitivity (%) | Specificity (%) | PPV (%) | NPV (%) | Accuracy (%) | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| R1 | R2 | R1 | R2 | R1 | R2 | R1 | R2 | R1 | R2 | |
| ccRCC | ||||||||||
| 5 | 34 (19, 52) | 34 (19, 52) | 100 (83, 100) | 100 (83, 100) | 100 (74, 100) | 100 (74, 100) | 47 (31, 62) | 47 (31, 62) | 58 (44, 71) | 58 (44, 71) |
| ≥4 | 71 (54, 85) | 63 (45, 79) | 70 (46, 88) | 80 (56, 94) | 81 (63, 93) | 85 (65, 96) | 58 (37, 78) | 55 (36, 74) | 71 (57, 82) | 69 (55, 81) |
| ≥3 | 97 (85, 100) | 94 (81, 99) | 40 (19, 64) | 40 (19, 64) | 74 (59, 86) | 73 (58, 85) | 89 (52, 100) | 80 (44, 97) | 76 (63, 87) | 75 (61, 85) |
| pRCC | ||||||||||
| ≤2 | 80 (44, 97) | 70 (35, 93) | 98 (88, 100) | 93 (82, 99) | 89 (52, 100) | 70 (35, 93) | 96 (85, 99) | 93 (92, 99) | 95 (85, 99) | 89 (78, 96) |
| 1 | 80 (44, 97) | 50 (19, 81) | 100 (82, 100) | 100 (92, 100) | 100 (63, 100) | 100 (48, 100) | 96 (85, 99) | 90 (78, 97) | 96 (87, 100) | 91 (80, 97) |
Note.—External dataset comprised 55 solid renal masses measuring ≤4 cm, from The Cancer Imaging Archive. Values in parentheses are 95% CIs. Comparisons were performed using McNemar test. ccRCC = clear cell renal cell carcinoma, NPV = negative predictive value, PPV = positive predictive value, pRCC = papillary renal cell carcinoma, R1 = reader 1, R2 = reader 2.
Discussion
In our study, we evaluated four competing multiparametric CT algorithms proposed to differentiate ccRCC and pRCC from other histopathology diagnoses among SoRMs ≤ 4 cm. The three-category abbreviated CT score was the best-overall-performing algorithm. We applied two decision rules to increase the specificity and stratify masses into five likelihood categories. The revised algorithm achieved very high specificity for ccRCCs at score 5, high sensitivity and moderate specificity for ccRCC at score 4, very high specificity for pRCC at score 1, high sensitivity and moderate specificity for pRCC at score 2, and moderate to substantial overall interobserver agreement. Score 3 masses were indeterminate and represented approximately one-third of the masses evaluated. We validated the performance of the resulting optimized system (ie, CT-Score version 2.0) in an external dataset from TCIA and observed similar results.
The existing multiparametric MRI ccLS algorithm includes the CM phase enhancement ratio (36). The use of CM phase attenuation or enhancement to differentiate ccRCC from other histopathology diagnoses with CT has been described (37,38). While the CT score and modified CT score both use the CM phase enhancement ratio as well as a subjective assessment of mass heterogeneity, the abbreviated CT score removes heterogeneity assessment to improve agreement. Heterogeneity of renal masses at CT has been extensively studied, for example, through subjective reader assessment and quantitative texture analysis (39,40). Reproducibility is a major limitation of heterogeneity assessment in clinical practice (41). Indeed, the findings of our analysis confirm the results of Lemieux et al (19) showing that the abbreviated CT score, which eliminates heterogeneity, performs at least as well as the original CT score for the diagnosis of ccRCC and pRCC, but with improved agreement.
To improve the modified CT score, we applied two evidence-based features that have purported high specificity for ccRCC and pRCC diagnosis. The first feature was attenuation at NCCT ≤ 20 HU, which, in heterogeneous and solid enhancing masses is a specific feature of RCC and, especially, ccRCC (26,27). In our study, NCCT attenuation ≤ 20 HU in moderately and intensely enhancing solid masses was 99% and 100% specific for ccRCC in both the internal and external datasets, respectively. This variable is akin to signal intensity loss on opposed-phase compared with in-phase chemical shift MRI and has been speculated to be due to the presence of microscopic fat (26,27,42). It is important to emphasize that this feature must be applied after a mass has been determined to be solid, not cystic, according to the CT-Score version 2.0 algorithm. This avoids the possibility of confusing a renal cyst containing simple fluid and measuring ≤ 20 HU with a solid enhancing mass measuring ≤ 20 HU containing microscopic fat. The second feature was absolute enhancement in the CM phase (calculation = CM attention – NCCT attenuation). CM-NCCT ≤ 20 HU in mildly and moderately enhancing solid masses was 100% specific for pRCC diagnosis in both our internal and external datasets. This is because pRCCs are known to enhance slowly, and enhancement in the CM phase is often less than 20 HU (32). This added variable overcomes limitations incurred when only the MC ratio is used to categorize masses; for example, because of differing patient physiology or CT technical factors, the renal cortex could be hypoenhancing (resulting in a higher MC ratio) or avidly enhancing (resulting in a lower MC ratio).
CT-Score version 2.0 achieves several goals for the characterization of SoRMs ≤ 4 cm with CT. First, the feature of attenuation at NCCT ≤ 20 HU creates a new score of 5, which is highly specific for ccRCC, similar to the MRI-based ccLS of 5. This specificity was lacking in the original CT score and other CT scoring systems, which had moderate to high sensitivity for ccRCC but low specificity (16–19). Second, the other feature (ie, CM-NCCT attenuation ≤ 20 HU) adds specificity to CT score 1 masses, which had a very high likelihood of being pRCCs. Thus, the added decision rules result in increased specificity of CT-Score version 2.0 for the classification of ccRCC and pRCC, overcoming a limitation of the initial system. A 2024 meta-analysis of available literature on the ccLS system and the CT score (combining the original CT score, modified CT score, and abbreviated CT score) revealed similar sensitivity (MRI ccLS: 82% [95% CI: 78, 85] vs CT: 80% [95% CI: 72, 86]) but lower specificity for CT (MRI ccLS: 76% [95% CI: 67, 83] vs CT: 59% [95% CI: 50, 67]) (43). The accuracy for the classification of ccRCC via CT-Score version 2.0 with threshold of ≥4 was comparable to the pooled results for the ccLS system (43). Given that this system has different thresholds to emphasize sensitivity (score 2 for pRCC, score 4 for ccRCC) and specificity (score 1 for pRCC, score 5 for ccRCC), the desired thresholds of this system can be applicable to a wide variety of management scenarios. Furthermore, depending on institutional preferences, either score 3 masses or both score 3 and 4 masses may be considered for renal mass biopsy. Last, the added modifying rules are straightforward and quantitative and remove subjective mass heterogeneity as a discriminating feature, resulting in acceptable interobserver agreement.
Our study had limitations. First, the sample size was moderate and consisted predominantly of masses with surgically confirmed histologic diagnoses (exception: 21 fat-poor angiomyolipomas diagnosed via biopsy). This reference standard biased the natural distribution of the analyzed masses, as a proportion of tumors ≤ 4 cm are expected to be managed via active surveillance. Second, the renal mass CT protocol was performed according to Society of Abdominal Radiology Disease-Focused Panel on RCC recommendations, although with different CT models. This variability in equipment could have negatively influenced the observed performance of the derived CT score. Third, the external dataset was obtained from the TCIA database instead of being a separate institutional dataset. Fourth, we assessed numerous CT features that have been described for predicting the histologic diagnosis of SoRMs. However, we did not evaluate all prior CT features and preferred to avoid subjective features that are not reproducible, such as shape or growth pattern (44) and tumor margin (37). Last, two radiologists from a single institution performed the evaluation task. More readers from more institutions would improve generalizability.
In conclusion, we performed validated optimization of the existing, best-performing CT scoring system for the diagnosis of ccRCC and pRCC among SoRMs ≤ 4 cm. The result is CT-Score version 2.0. By including two highly specific variables (NCCT attenuation ≤ 20 HU and CM-NCCT attenuation ≤ 20 and ≤ 30 HU), we derived a five-tiered scoring system that was specific for ccRCC and pRCC (in score 5 and score 1, respectively), sensitive with moderate specificity for ccRCC and pRCC (in score 4 and score 2, respectively), and had moderate to substantial interobserver agreement. Importantly, similar results were obtained in separate internal and external datasets. CT-Score version 2.0 has increased specificity compared with the existing abbreviated CT score and is reproducible, performing similarly in this sample to the ccLS system. The CT-Score version 2.0 algorithm (for solid masses at CT), combined with the ccLS system (for solid masses at MRI) and Bosniak version 2019 (for cystic renal masses at MRI and CT), should directly inform the backbone for the recently endorsed KI-RADS by the American College of Radiology. Further evaluation of our work is required. If validated, the next iterative step would be to improve prediction of aggressive cancer beyond stratifying among common RCC subtypes.
Supplemental Files
Funding: Authors declared no funding for this work.
Abbreviations:
- AC1
- agreement coefficient
- ccLS
- clear cell likelihood score
- ccRCC
- clear cell RCC
- CM
- corticomedullary
- KI-RADS
- Kidney Imaging Reporting and Data System
- MC ratio
- mass to cortex CM phase ratio
- NCCT
- noncontrast CT
- pRCC
- papillary RCC
- RCC
- renal cell carcinoma
- SoRM
- solid renal masses
- TCIA
- The Cancer Imaging Archive
Disclosures of conflicts of interest
N.S. is a senior editor for the American Journal of Roentgenology. M.S.D. is an associate editor for Radiology. Please see ICMJE form(s) for author conflicts of interest. These have been provided as supplemental materials.
References
- 1. Meyer HJ , Pfeil A , Schramm D , Bach AG , Surov A . Renal incidental findings on computed tomography: Frequency and distribution in a large non selected cohort . Medicine (Baltimore) 2017. ; 96 ( 26 ): e7039 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. O’Connor SD , Pickhardt PJ , Kim DH , Oliva MR , Silverman SG . Incidental finding of renal masses at unenhanced CT: prevalence and analysis of features for guiding management . AJR Am J Roentgenol 2011. ; 197 ( 1 ): 139 – 145 . [DOI] [PubMed] [Google Scholar]
- 3. Silverman SG , Pedrosa I , Ellis JH , et al . Bosniak Classification of Cystic Renal Masses, Version 2019: An Update Proposal and Needs Assessment . Radiology 2019. ; 292 ( 2 ): 475 – 488 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Remzi M , Ozsoy M , Klingler HC , et al . Are small renal tumors harmless? Analysis of histopathological features according to tumors 4 cm or less in diameter . J Urol 2006. ; 176 ( 3 ): 896 – 899 . [DOI] [PubMed] [Google Scholar]
- 5. Pahernik S , Ziegler S , Roos F , Melchior SW , Thüroff JW . Small renal tumors: correlation of clinical and pathological features with tumor size . J Urol 2007. ; 178 ( 2 ): 414 – 417 ; discussion 416–417 . [DOI] [PubMed] [Google Scholar]
- 6. Lee H , Lee JK , Kim K , et al . Risk of metastasis for T1a renal cell carcinoma . World J Urol 2016. ; 34 ( 4 ): 553 – 559 . [DOI] [PubMed] [Google Scholar]
- 7. Schieda N , Krishna S , Pedrosa I , Kaffenberger SD , Davenport MS , Silverman SG . Active Surveillance of Renal Masses: The Role of Radiology . Radiology 2022. ; 302 ( 1 ): 11 – 24 . [DOI] [PubMed] [Google Scholar]
- 8. Reporting and Data Systems (RADS) . https://www.acr.org/Clinical-Resources/Clinical-Tools-and-Reference/Reporting-and-Data-Systems. Accessed December 2024 .
- 9. Silverman SG , Pedrosa I , Schieda N , Margulis V , Kapur P , Davenport MS . In Pursuit of KI-RADS: Toward a Single, Evidence-based Imaging Classification of Renal Masses . Radiology 2025. ; 314 ( 3 ): e240308 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Silverman SG , Pedrosa I , Ellis JH , et al . Bosniak classification of cystic renal masses, version 2019: an update proposal and needs assessment . Radiology 2019. ; 292 ( 2 ): 475 – 488 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Kay FU , Canvasser NE , Xi Y , et al . Diagnostic Performance and Interreader Agreement of a Standardized MR Imaging Approach in the Prediction of Small Renal Mass Histology . Radiology 2018. ; 287 ( 2 ): 543 – 553 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Pedrosa I , Cadeddu JA . How I Do It: Managing the Indeterminate Renal Mass with the MRI Clear Cell Likelihood Score . Radiology 2022. ; 302 ( 2 ): 256 – 269 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Canvasser NE , Kay FU , Xi Y , et al . Diagnostic Accuracy of Multiparametric Magnetic Resonance Imaging to Identify Clear Cell Renal Cell Carcinoma in cT1a Renal Masses . J Urol 2017. ; 198 ( 4 ): 780 – 786 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Johnson BA , Kim S , Steinberg RL , de Leon AD , Pedrosa I , Cadeddu JA . Diagnostic performance of prospectively assigned clear cell Likelihood scores (ccLS) in small renal masses at multiparametric magnetic resonance imaging . Urol Oncol 2019. ; 37 ( 12 ): 941 – 946 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Schieda N , Davenport MS , Silverman SG , et al . Multicenter Evaluation of Multiparametric MRI Clear Cell Likelihood Scores in Solid Indeterminate Small Renal Masses . Radiology 2022. ; 303 ( 3 ): 590 – 599 . [Published correction appears in Radiology 2023;306(3):e239001.] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Al Nasibi K , Pickovsky JS , Eldehimi F , et al . Development of a Multiparametric Renal CT Algorithm for Diagnosis of Clear Cell Renal Cell Carcinoma Among Small (≤ 4 cm) Solid Renal Masses . AJR Am J Roentgenol 2022. ; 219 ( 5 ): 814 – 823 . [DOI] [PubMed] [Google Scholar]
- 17. Eldehimi F , Pickovsky JS , Al Nasibi K , Schieda N . Proposed Modifications to the Multiparametric CT Score for Diagnosis of ccRCC Using Hyperattenuation, Segmental Enhancement Inversion, and ADER: A Secondary Analysis . AJR Am J Roentgenol 2023. ; 221 ( 1 ): 144 – 146 . [DOI] [PubMed] [Google Scholar]
- 18. Eldihimi F , Walsh C , Hibbert RM , Nasibi KA , Pickovsky JS , Schieda N . Evaluation of a multiparametric renal CT algorithm for diagnosis of clear-cell renal cell carcinoma among small (≤ 4 cm) solid renal masses . Eur Radiol 2024. ; 34 ( 6 ): 3992 – 4000 . [DOI] [PubMed] [Google Scholar]
- 19. Lemieux S , Shen L , Liang T , et al . External Validation of a Five-Tiered CT Algorithm for the Diagnosis of Clear Cell Renal Cell Carcinoma: A Retrospective Five-Reader Study . AJR Am J Roentgenol 2023. ; 221 ( 3 ): 334 – 343 . [DOI] [PubMed] [Google Scholar]
- 20. Chung A , Raman SS . Radiologist’s Disease: Imaging for Renal Cancer . Urol Clin North Am 2023. ; 50 ( 2 ): 161 – 180 . [DOI] [PubMed] [Google Scholar]
- 21. Branger N , Bigot P , Pignot G , et al . Oncocytoma on renal mass biopsy: is it still the same histology when surgery is performed? Results from UroCCR-104 study . World J Urol 2023. ; 41 ( 2 ): 483 – 489 . [DOI] [PubMed] [Google Scholar]
- 22. Renshaw AA . The diagnosis of benign renal tumors on limited material . Cancer Cytopathol 2022. ; 130 ( 12 ): 927 – 929 . [DOI] [PubMed] [Google Scholar]
- 23. Moch H , Cubilla AL , Humphrey PA , Reuter VE , Ulbright TM . The 2016 WHO Classification of Tumours of the Urinary System and Male Genital Organs-Part A: Renal, Penile, and Testicular Tumours . Eur Urol 2016. ; 70 ( 1 ): 93 – 105 . [DOI] [PubMed] [Google Scholar]
- 24. Heller N , Sathianathen N , Kalapara A , et al . Data from C4KC-KiTS [Data set] . The Cancer Imaging Archive . 2019. . doi: 10.7937/TCIA.2019.IX49E8NX . [DOI]
- 25. Heller N , Isensee F , Maier-Hein KH , et al . The state of the art in kidney and kidney tumor segmentation in contrast-enhanced CT imaging: Results of the KiTS19 challenge . Med Image Anal 2021. ; 67 : 101821 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Schieda N , Vakili M , Dilauro M , Hodgdon T , Flood TA , Shabana WM . Solid renal cell carcinoma measuring water attenuation (−10 to 20 HU) on unenhanced CT . AJR Am J Roentgenol 2015. ; 205 ( 6 ): 1215 – 1221 . [Published correction appears in AJR Am J Roentgenol 2016;206(4):901.] [DOI] [PubMed] [Google Scholar]
- 27. McGahan JP , Sidhar K , Fananapazir G , et al . Renal cell carcinoma attenuation values on unenhanced CT: importance of multiple, small region-of-interest measurements . Abdom Radiol (NY) 2017. ; 42 ( 9 ): 2325 – 2333 . [DOI] [PubMed] [Google Scholar]
- 28. Pedrosa I , Cadeddu JA . How We Do It: Managing the Indeterminate Renal Mass with the MRI Clear Cell Likelihood Score . Radiology 2022. ; 302 ( 2 ): 256 – 269 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Yuh BI , Cohan RH . Different phases of renal enhancement: role in detecting and characterizing renal masses during helical CT . AJR Am J Roentgenol 1999. ; 173 ( 3 ): 747 – 755 . [DOI] [PubMed] [Google Scholar]
- 30. Birnbaum BA , Jacobs JE , Ramchandani P . Multiphasic renal CT: comparison of renal mass enhancement during the corticomedullary and nephrographic phases . Radiology 1996. ; 200 ( 3 ): 753 – 758 . [DOI] [PubMed] [Google Scholar]
- 31. Bata P , Gyebnar J , Tarnoki DL , et al . Clear cell renal cell carcinoma and papillary renal cell carcinoma: differentiation of distinct histological types with multiphase CT . Diagn Interv Radiol 2013. ; 19 ( 5 ): 387 – 392 . [DOI] [PubMed] [Google Scholar]
- 32. Pierorazio PM , Hyams ES , Tsai S , et al . Multiphasic enhancement patterns of small renal masses (≤4 cm) on preoperative computed tomography: utility for distinguishing subtypes of renal cell carcinoma, angiomyolipoma, and oncocytoma . Urology 2013. ; 81 ( 6 ): 1265 – 1271 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Dilauro M , Quon M , McInnes MDF , et al . Comparison of contrast-enhanced multiphase renal protocol CT versus MRI for diagnosis of papillary renal cell carcinoma . AJR Am J Roentgenol 2016. ; 206 ( 2 ): 319 – 325 . [DOI] [PubMed] [Google Scholar]
- 34. Krishna S , Sadoughi N , McInnes MDF , Chatelain R , MacDonald DB , Schieda N . Attenuation and degree of enhancement with conventional 120-kVp polychromatic CT and 70-keV monochromatic rapid kilovoltage-switching dual-energy CT in cystic and solid renal masses . AJR Am J Roentgenol 2018. ; 211 ( 4 ): 789 – 796 . [DOI] [PubMed] [Google Scholar]
- 35. Gwet KL . Computing inter-rater reliability and its variance in the presence of high agreement . Br J Math Stat Psychol 2008. ; 61 ( Pt 1 ): 29 – 48 . [DOI] [PubMed] [Google Scholar]
- 36. de Leon Diaz A , Davenport MS , Silverman SG , Schieda N , Cadeddu JA , Pedrosa I . Role of Virtual Biopsy in the Management of Renal Masses . AJR Am J Roentgenol 2019. ; 212 ( 6 ): 1234 – 1243 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Lee-Felker SA , Felker ER , Tan N , et al . Qualitative and quantitative MDCT features for differentiating clear cell renal cell carcinoma from other solid renal cortical masses . AJR Am J Roentgenol 2014. ; 203 ( 5 ): W516 – W524 . [DOI] [PubMed] [Google Scholar]
- 38. Young JR , Margolis D , Sauk S , Pantuck AJ , Sayre J , Raman SS . Clear cell renal cell carcinoma: discrimination from other renal cell carcinoma subtypes and oncocytoma at multiphasic multidetector CT . Radiology 2013. ; 267 ( 2 ): 444 – 453 . [DOI] [PubMed] [Google Scholar]
- 39. Hodgdon T , McInnes MDF , Schieda N , Flood TA , Lamb L , Thornhill RE . Can Quantitative CT Texture Analysis be Used to Differentiate Fat-poor Renal Angiomyolipoma from Renal Cell Carcinoma on Unenhanced CT Images? Radiology 2015. ; 276 ( 3 ): 787 – 796 . [DOI] [PubMed] [Google Scholar]
- 40. Schieda N , Thornhill RE , Al-Subhi M , et al . Diagnosis of Sarcomatoid Renal Cell Carcinoma With CT: Evaluation by Qualitative Imaging Features and Texture Analysis . AJR Am J Roentgenol 2015. ; 204 ( 5 ): 1013 – 1023 . [DOI] [PubMed] [Google Scholar]
- 41. Doshi AM , Tong A , Davenport MS , et al . Assessment of Renal Cell Carcinoma by Texture Analysis in Clinical Practice: A Six-Site, Six-Platform Analysis of Reliability . AJR Am J Roentgenol 2021. ; 217 ( 5 ): 1132 – 1140 . [DOI] [PubMed] [Google Scholar]
- 42. Schieda N , Siegelman ES . Update on CT and MRI of Adrenal Nodules . AJR Am J Roentgenol 2017. ; 208 ( 6 ): 1206 – 1217 . [DOI] [PubMed] [Google Scholar]
- 43. Zhong J , Hu Y , Xing Y , et al . Is there enough evidence supporting the clinical adoption of clear cell likelihood score (ccLS)? An updated systematic review and meta-analysis . Insights Imaging 2024. ; 15 ( 1 ): 242 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Lim RS , McInnes MDF , Siddaiah M , Flood TA , Lavallee LT , Schieda N . Are growth patterns on MRI in small (< 4 cm) solid renal masses useful for predicting benign histology? Eur Radiol 2018. ; 28 ( 8 ): 3115 – 3124 . [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.