Machine learning for differentiation of lipid-poor adrenal adenoma and subclinical pheochromocytoma based on multiphase CT imaging radiomics.

BACKGROUND: There is a paucity of research investigating the application of machine learning techniques for distinguishing between lipid-poor adrenal adenoma (LPA) and subclinical pheochromocytoma (sPHEO) based on radiomic features extracted from non-contrast and dynamic contrast-enhanced computed tomography (CT) scans of the abdomen. METHODS: We conducted a retrospective analysis of multiphase spiral CT scans, including non-contrast, arterial, venous, and delayed phases, as well as thin- and thick-thickness images from 134 patients with surgically and pathologically confirmed. A total of 52 patients with LPA and 44 patients with sPHEO were randomly assigned to training/testing sets in a 7:3 ratio. Additionally, a validation set was comprised of 22 LPA cases and 16 sPHEO cases from two other hospitals. We used 3D Slicer and PyRadiomics to segment tumors and extract radiomic features, respectively. We then applied T-test and least absolute shrinkage and selection operator (LASSO) to select features. Six binary classifiers, including K-nearest neighbor (KNN), logistic regression (LR), decision tree (DT), random forest (RF), support vector machine (SVM), and multi-layer perceptron (MLP), were employed to differentiate LPA from sPHEO. Receiver operating characteristic (ROC) curves and area under the curve (AUC) values were compared using DeLong's method. RESULTS: All six classifiers showed good diagnostic performance for each phase and slice thickness, as well as for the entire CT data, with AUC values ranging from 0.706 to 1. Non-contrast CT densities of LPA were significantly lower than those of sPHEO (P < 0.001). However, using the optimal threshold for non-contrast CT density, sensitivity was only 0.743, specificity 0.744, and AUC 0.828. Delayed phase CT density yielded a sensitivity of 0.971, specificity of 0.641, and AUC of 0.814. In radiomics, AUC values for the testing set using non-contrast CT images were: KNN 0.919, LR 0.979, DT 0.835, RF 0.967, SVM 0.979, and MLP 0.981. In the validation set, AUC values were: KNN 0.891, LR 0.974, DT 0.891, RF 0.964, SVM 0.949, and MLP 0.979. CONCLUSIONS: The machine learning model based on CT radiomics can accurately differentiate LPA from sPHEO, even using non-contrast CT data alone, making contrast-enhanced CT unnecessary for diagnosing LPA and sPHEO.

Deep learning approach for differentiating indeterminate adrenal masses using CT imaging.

PURPOSE: Distinguishing stage 1-2 adrenocortical carcinoma (ACC) and large, lipid poor adrenal adenoma (LPAA) via imaging is challenging due to overlapping imaging characteristics. This study investigated the ability of deep learning to distinguish ACC and LPAA on single time-point CT images. METHODS: Retrospective cohort study from 1994 to 2022. Imaging studies of patients with adrenal masses who had available adequate CT studies and histology as the reference standard by method of adrenal biopsy and/or adrenalectomy were included as well as four patients with LPAA determined by stability or regression on follow-up imaging. Forty-eight (48) subjects with pathology-proven, stage 1-2 ACC and 43 subjects with adrenal adenoma >3 cm in size demonstrating a mean non-contrast CT attenuation > 20 Hounsfield Units centrally were included. We used annotated single time-point contrast-enhanced CT images of these adrenal masses as input to a 3D Densenet121 model for classifying as ACC or LPAA with five-fold cross-validation. For each fold, two checkpoints were reported, highest accuracy with highest sensitivity (accuracy focused) and highest sensitivity with the highest accuracy (sensitivity focused). RESULTS: We trained a deep learning model (3D Densenet121) to predict ACC versus LPAA. The sensitivity-focused model achieved mean accuracy: 87.2% and mean sensitivity: 100%. The accuracy-focused model achieved mean accuracy: 91% and mean sensitivity: 96%. CONCLUSION: Deep learning demonstrates promising results distinguishing between ACC and large LPAA using single time-point CT images. Before being widely adopted in clinical practice, multicentric and external validation are needed.

Adrenal incidentaloma: machine learning-based quantitative texture analysis of unenhanced CT can effectively differentiate sPHEO from lipid-poor adrenal adenoma.

Objective: To evaluate the feasibility and accuracy of machine learning based texture analysis of unenhanced CT images in differentiating subclinical pheochromocytoma (sPHEO) from lipid-poor adenoma (LPA) in adrenal incidentaloma (AI). Methods: Seventy-nine patients with 80 LPA and 29 patients with 30 sPHEO were included in the study. Texture parameters were derived using imaging software (MaZda). Thirty texture features were selected and LPA was performed for the features selected. The number of positive features was used to predict results. Logistic multiple regression analysis was performed on the 30 texture features, and a predictive equation was created based on the coefficients obtained. Results: LPA yielded a misclassification rate of 19.39% in differentiating sPHEO from LPA. Our predictive model had an accuracy rate of 94.4% (102/108), with a sensitivity of 86.2% (25/29) and a specificity of 97.5% (77/79) for differentiation. When the number of positive features was greater than 8, the accuracy of prediction was 85.2% (92/108), with a sensitivity of 96.6% (28/29) and a specificity of 81% (64/79). Conclusions: Machine learning-based quantitative texture analysis of unenhanced CT may be a reliable quantitative method in differentiating sPHEO from LPA when AI is present.

The role of dynamic post-contrast T1-w MRI sequence to characterize lipid-rich and lipid-poor adrenal adenomas in comparison to non-adenoma lesions: preliminary results.

PURPOSE: The purpose of the article is to compare the features of wash-out (WO) parameters between lipid-rich and lipid-poor adrenal adenomas as well as with a group of non-adenoma adrenal lesions. METHODS: 46 patients (36 F and 10 M, median age 58 years) with unilateral adrenal lesions (35 adenomas, 7 pheochromocytomas, 1 carcinoma, and 3 metastases) were prospectively evaluated; adrenal lesions were divided into adenomas (Group 1) and non-adenomas (Group 2). MR imaging was performed with a 3-Tesla scanner using pre- and post-contrast dedicated sequences. On the basis of the evaluation of qualitative chemical-shift (CS) signal intensity (SI) loss, adrenal adenomas were, respectively, divided in Group 1A (n = 25) as lipid-rich and Group 1B (n = 10) as lipid-poor; non-adenoma adrenal lesions were grouped in Group 2 (n = 11). The following parameters were evaluated: size (mm), CS SI index (%), early (5 min), and delayed (10 min) Relative (R) and Absolute (A) WO values (%). RESULTS: The comparison of AWO and RWO showed significant (p ≤ 0.05) differences between Group 1A and Groups 1B and 2, both using 5- and 10-min images for calculation; conversely, no differences in these dynamic parameters were found between Group 1B and 2; AWO and RWO values were significantly lower in adrenal lesions of Groups 1B and 2 compared to Group 1A, both using 5- and 10-min images for calculation. CONCLUSIONS: The quantitative evaluation of WO parameters could not be used to characterize lipid-poor adrenal adenomas for which alternative imaging modalities are required.

Using quantitative CT texture analysis to differentiate adrenal tumors / 实用放射学杂志

Objective To investigate the feasibility of using CT texture analysis to differentiate among lipid-poor adrenal adenoma,pheochromocytoma and adrenal metastases.Methods 66 lipid-poor adrenal adenoma,98 pheochromocytoma and 101 adrenal metastases lesions were analyzed retrospectively.All the patients had abdominal non-enhanced CT and adrenal enhanced CT scans.We used TexRAD software to analyze the textural features of CT images and compared the differences in each texture parameter among three adrenal lesions.Results On non-enhanced CT images,there were significant differences in Mean and Kurtosis at all the texture scales(SSF 0-6) among the three types of adrenal lesions (P<0.05),as well as SD at fine and coarse texture scale (SSF 2,6)(P<0.05).Entropy (SSF 0-3, 5-6) and MPP (SSF 0-2, 4-6) were significantly lower in lipid-poor adrenal adenoma and adrenal metastases than that in pheochromocytoma (P<0.05).There were significant differences in Skewness (SSF 0-3) among three types of lesions, which was lowest in pheochromocytoma and highest in adrenal metastases.On enhanced CT images, Mean, SD, Entrophy and MPP showed significantly differences among the three types of adrenal lesions at all the texture scales (SSF 0-6) (P<0.05), which were all highest in pheochromocytoma and lowest in adrenal metastases.Skewness (SSF 0) and Kurtosis (SSF 0, 2) were significantly lower in adrenal metastases than that in lipid-poor adrenal adenoma and pheochromocytoma (P<0.05).Conclusion There are significant differences in CT texture analysis parameters among lipid-poor adrenal adenoma,pheochromocytoma and adrenal metastases.CT texture analysis has potential clinical application values in differentiating these three adrenal lesions.

Differentiation of lipid-poor adrenal adenomas from non-adenomas with magnetic resonance imaging: Utility of dynamic, contrast enhancement and single-shot T2-weighted sequences.

PURPOSE: To evaluate the utility of dynamic, contrast-enhanced magnetic resonance imaging (MRI) in combination with single-shot T2-weighted (ssT2) sequences in the differentiation of lipid-poor adrenal adenomas from non-adenomas. MATERIALS AND METHODS: This retrospective study was approved by the institutional review board and is HIPAA compliant. Between January 2007 and December 2010, 46 patients with MRI demonstrating a lipid-poor adrenal lesion who underwent either surgical resection or a minimum of 24 months of imaging follow-up were identified retrospectively. All images were retrospectively reviewed in blinded fashion by two radiologists. Each adrenal lesion was categorized by dynamic enhancement features and qualitative signal on ssT2 images and was categorized as an adenoma if it demonstrated homogenous enhancement in the arterial phase, washout with capsule enhancement in the delayed phase, and T2 signal isointense to normal adrenal tissue. Any lesion that did not fulfill all the criteria was classified as a non-adenoma. Sensitivity, specificity, positive predictive value, negative predictive value, and accuracy for characterization of adenoma were calculated for each reader with 95% confidence intervals. A κ test assessed level of agreement between readers. RESULTS: Application of our criteria lead to an MRI diagnosis of lipid-poor adrenal adenoma with a sensitivity of 84.2-89.5% (16/19-17/19), specificity of 96.3% (26/27), positive predictive value of 94.1-94.4% (16/17-17/18), negative predictive value of 89.7-92.9% (26/29-26/28), and accuracy of 91.3-93.5% (42/46-43/46). Agreement between the two readers showed substantial κ agreement for the differentiation of adenoma from non-adenoma. CONCLUSIONS: Dynamic, contrast-enhanced T1-weighted three-dimensional gradient echo sequences in combination with ssT2 images can accurately differentiate lipid-poor adrenal adenomas from non-adenomas.