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.

Alkaline phosphatase combines with CT factors for differentiating small (≤ 4 cm) fat-poor angiomyolipoma from renal cell carcinoma: a multiple quantitative tool

PURPOSE

This study aimed to evaluate the diagnostic value of serum and CT factors to establish a convenient diagnostic method for differentiating small (≤ 4 cm) fat-poor angiomyolipoma (AML) from renal cell carcinoma (RCC).

MATERIALS AND METHODS

This study analyzed the preoperative serum laboratory data and CT data of 32 fat-poor AML patients and 133 RCC patients. The CT attenuation value of tumor (AVT), relative enhancement ratio (RER), and heterogeneous degree of tumor were detected using region of interest on precontrast phase (PCP) and the corticomedullary phase. Multivariate regression was performed to filter the main factors. The main factors were selected to establish the prediction models. The area under the curve (AUC) was measured to evaluate the diagnostic efficacy.

RESULTS

Fat-poor AML was more common found in younger (47.91 ± 2.09 years vs 53.63 ± 1.17 years, P = 0.02) and female (70.68 vs 28.13%, P < 0.001) patients. Alkaline phosphatase (ALP) was higher in RCC patients (81.80 ± 1.75 vs 63.25 ± 2.95 U/L, P < 0.01). For CT factors, fat-poor AML was higher in PCP_AVT (40.30 ± 1.49 vs 32.98 ± 0.69Hu, P < 0.01) but lower in RER (67.17 ± 3.17 vs 84.64 ± 2.73, P < 0.01). Gender, ALP, PCP_AVT and RER was found valuable for the differentiation. When compared with laboratory-based or CT-based diagnostic models, the combination model integrating gender, ALP, PCP_AVT and RER shows the best diagnostic performance (AUC = 0.922).

CONCLUSION

ALP was found higher in RCC patients. Female patients with ALP < 70.50U/L, PCP_AVT > 35.97Hu and RER < 82.66 are more likely to be diagnose as fat-poor AML.

Differential diagnosis and prognosis of small renal masses: association with collateral vessels detected using contrast-enhanced computed tomography

Background

Active surveillance (AS) is one of the treatment methods for patients with small renal masses (SRMs; < 4 cm), including renal cell carcinomas (RCCs). However, some small RCCs may exhibit aggressive neoplastic behaviors and metastasize. Little is known about imaging biomarkers capable of identifying potentially aggressive small RCCs. Contrast-enhanced computed tomography (CECT) often detects collateral vessels arising from neoplastic angiogenesis in RCCs. Therefore, this study aimed to evaluate the association between SRM differential diagnoses and prognoses, and the detection of collateral vessels using CECT.

Methods

A total of 130 consecutive patients with pathologically confirmed non-metastatic SRMs (fat-poor angiomyolipomas [fpAMLs; n = 7] and RCCs [n = 123]) were retrospectively enrolled. Between 2011 and 2019, SRM diagnoses in these patients were confirmed after biopsy or surgical resection. All RCCs were surgically resected. Regardless of diameter, a collateral vessel (CV) was defined as any blood vessel connecting the tumor from around the kidney using CECT. First, we analyzed the role of CV-detection in differentiating between fpAML and RCC. Then, we evaluated the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of RCC diagnosis based on CV-detection using CECT. We also assessed the prognostic value of CV-detection using the Fisher exact test, and Kaplan-Meier method and the log-rank test.

Results

The sensitivity, specificity, PPV, NPV, and accuracy of CV-detection for the diagnosis of small RCCs was 48.5, 45.5, 100, 100, and 9.5% respectively. Five of 123 (4.1%) patients with RCC experienced recurrence. CV-detection using CECT was the only significant factor associated with recurrence (p = 0.0177). Recurrence-free survival (RFS) was significantly lower in patients with CV compared with in those without CV (5-year RFS 92.4% versus 100%, respectively; p = 0.005). In addition, critical review of the CT images revealed the CVs to be continuous with the venous vessels around the kidney.

Conclusions

The detection of CVs using CECT is useful for differentiating between small fpAMLs and RCCs. CV-detection may also be applied as a predictive parameter for small RCCs prone to recurrence after surgical resection. Moreover, AS could be suitable for small RCCs without CVs.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-022-09971-w.

Image Interpretation: Practical Triage of Benign from Malignant Renal Masses

Indeterminate renal masses remain a diagnostic challenge for lesions not initially characterized as angiomyolipoma or Bosniak I/II cysts. Differential for indeterminate renal masses include oncocytoma, fat-poor angiomyolipoma, and clear cell, papillary, and chromophobe renal cell carcinoma. Qualitative and quantitative techniques using data derived from multiphase contrast-enhanced imaging have provided methods for specific differentiation and subtyping of indeterminate renal masses, with emerging applications such as radiocytogenetics. Early and accurate characterization of indeterminate renal masses by multiphase contrast-enhanced imaging will optimize triage of these lesions into surgical, ablative, and active surveillance treatment plans.

Differentiation between fat-poor angiomyolipoma and clear cell renal cell carcinoma: qualitative and quantitative analysis using arterial spin labeling MR imaging

PURPOSE

To assess the diagnostic effectiveness of arterial spin labeling (ASL) MR imaging in differentiating fat-poor AML from clear cell renal cell carcinoma (ccRCC).

METHODS

In this prospective study, 29 ccRCC patients and 9 fat-poor AML patients underwent routine anatomical MRI and ASL at 3T before surgery after signing written informed consent form. For each tumor, tumor blood flow (TBF) was measured in a region of interest (ROI) which was positioned to outline the edge of the target lesions on ASL perfusion maps. Additionally, the mean TBF values were obtained by standardizing the TBF using a blood flow measurement in the reference ROI. Moreover, a cluster containing more than 10 voxels was chosen from the renal cortex and medulla area in normal contralateral kidney as a reference ROI to calculate tumor-to-cortex ratio and tumor-to-medulla ratio. Independent sample t test was used to examine the alteration among the groups of fat-poor AML and ccRCC. ASL images were together analyzed by two radiologists to assess the following characteristics of the renal mass: predominant SI in the tumor on ASL images was lower than, as same as, or higher than SI of the cortex. For qualitative variables, Fisher's exact test was employed to compare the proportions of these two groups. The sensitivity, specificity ,and accuracy required for discrimination of fat-poor AML from ccRCC were quantified using receiver operating characteristic (ROC) curve. The corresponding optimal cutoff value was obtained for each parameter as well.

RESULTS

The TBF value was significantly higher in ccRCC group than that in fat-poor AML (270.49 ± 78.88 ml/100 g/min vs. 146.68 ± 47.21 ml/100 g/min; P < 0.01). Both tumor-to-cortex and tumor-to-medulla ratios were notably higher in ccRCC group compared with those in fat-poor AML group (1.22 ± 0.26 vs. 0.74 ± 0.14, 3.13 ± 0.94 vs. 1.77 ± 0.55; P < 0.05). The values of area under the ROC curve (AUC) for TBF, tumor-to-cortex ratio, and tumor-to-medulla ratio were 0.931, 0.964, and 0.900, respectively. No significant difference in AUC values among these three measurements was observed. For qualitative variables, the SI of fat-poor AML was equal to or slightly lower than that of renal medulla and the SI of ccRCC was found to be higher than renal cortex in ASL.

CONCLUSION

ASL MRI performs well in differentiating fat-poor AML from ccRCC in both qualitative and quantitative analyses.

Differentiation between papillary renal cell carcinoma and fat-poor angiomyolipoma: a preliminary study assessing detection of intratumoral hemorrhage with chemical shift MRI and T2*-weighted gradient echo

Background Recent literature suggests that intratumoral hemorrhage detection may be helpful in differentiating papillary renal cell carcinoma (pRCC) from fat-poor angiomyolipoma (fpAML). Purpose To determine whether intratumoral hemorrhage detected using chemical shift magnetic resonance imaging (MRI) and T2*-weighted (T2*W) gradient echo (GRE) can be used to differentiate pRCC from fpAML. Material and Methods This retrospective study included 42 patients with pRCC (n = 28) and fpAML (n = 14) who underwent MRI followed by surgery. Two blinded radiologists independently assessed the presence of intratumoral hemorrhage using chemical shift MRI (decrease in signal intensity from opposed- to in-phase) and T2*W GRE ("blooming"). Consensus reading was determined for discrepant cases. MRI findings were compared using Chi-square test. Inter-observer agreement was assessed using kappa statistics. Results Inter-observer agreement was substantial for both sequences ( k = 0.622 and 0.793, P < 0.001). For chemical shift MRI, the prevalence of intratumoral hemorrhage was significantly greater in pRCC than in fpAML (71.4% versus 28.6%, P = 0.019 for reader 1; 64.3% versus 14.3%, P = 0.003 for reader 2; and 75% versus 21.4%, P = 0.002 for the consensus). T2*W GRE showed a similar tendency (46.4% versus 14.3%, P = 0.049 for both readers; and 50% versus 14.3%, P = 0.042 for the consensus). Using the consensus reading, sensitivity and specificity of determining pRCC were 75% and 78.6% for chemical shift MRI and 50% and 85.7% for T2*W GRE. Conclusion The prevalence of intratumoral hemorrhage identified from chemical shift MRI or T2*W GRE was significantly different between pRCC and fpAML. These hemorrhage-sensitive MRI sequences may be used as an adjunctive tool for discriminating between the two entities.

Tumoral vascular pattern in renal cell carcinoma and fat-poor renal angiomyolipoma as a novel helpful differentiating factor on contrast-enhanced CT scan

Our objective was to evaluate the differences between tumoral vascular pattern of renal cell carcinoma and fat-poor angiomyolipoma by contrast-enhanced computed tomography. All included patients had a definitive pathological diagnosis of either angiomyolipoma or renal cell carcinoma, and then the contrast-enhanced computed tomography images of these patients were evaluated. The patients who had visible prominent vessels in cross-sectional imaging were selected. The tumor vascular pattern (prominent (>2 mm) intratumoral and peritumoral vessels), density, and diameter of the vessels in renal cell carcinoma and fat-poor angiomyolipoma were evaluated. All cases (n = 12) with fat-poor angiomyolipoma were found to have intratumoral vessels and all cases (n = 36) with clear cell renal cell carcinoma were found to have peritumoral vessels. There was no significant correlation detected between the diameter of tumor and the density as well as diameter of the vessels. In conclusion, the evaluation of the vascular pattern using contrast enhancement contrast-enhanced computed tomography may provide important information that is useful in helping accurate differential diagnosis of fat-poor angiomyolipoma or renal cell carcinoma preoperatively.