CT-based radiomics signature for differentiation between cardiac tumors and a thrombi: a retrospective, multicenter study

A Novel Data Augmentation Method for Radiomics Analysis Using Image Perturbations

Radiomics extracts hundreds of features from medical images to quantitively characterize a region of interest (ROI). When applying radiomics, imbalanced or small dataset issues are commonly addressed using under or over-sampling, the latter being applied directly to the extracted features. Aim of this study is to propose a novel balancing and data augmentation technique by applying perturbations (erosion, dilation, contour randomization) to the ROI in cardiac computed tomography images. From the perturbed ROIs, radiomic features are extracted, thus creating additional samples. This approach was tested addressing the clinical problem of distinguishing cardiac amyloidosis (CA) from aortic stenosis (AS) and hypertrophic cardiomyopathy (HCM). Twenty-one CA, thirty-two AS and twenty-one HCM patients were included in the study. From each original and perturbed ROI, 107 radiomic features were extracted. The CA-AS dataset was balanced using the perturbation-based method along with random over-sampling, adaptive synthetic (ADASYN) and the synthetic minority oversampling technique (SMOTE). The same methods were tested to perform data augmentation dealing with CA and HCM. Features were submitted to robustness, redundancy, and relevance analysis testing five feature selection methods (p-value, least absolute shrinkage and selection operator (LASSO), semi-supervised LASSO, principal component analysis (PCA), semi-supervised PCA). Support vector machine performed the classification tasks, and its performance were evaluated by means of a 10-fold cross-validation. The perturbation-based approach provided the best performances in terms of f1 score and balanced accuracy in both CA-AS (f1 score: 80%, AUC: 0.91) and CA-HCM (f1 score: 86%, AUC: 0.92) classifications. These results suggest that ROI perturbations represent a powerful approach to address both data balancing and augmentation issues.

The usefulness of contrast echocardiography in the evaluation of cardiac masses: a multicenter study

BACKGROUND

Cardiac masses can encompass a variety of conditions, such as tumors, thrombi, vegetations, calcific lesions, and other rare diseases. Treatment and management of these types of cardiac masses differ considerably. Thus, accurately distinguishing among thrombi, benign tumors, and malignant tumors in the heart is of great importance. Contrast echocardiography (CE) has emerged as a promising technology. Although published guidelines suggest that CE can enhance image quality and assist in differentiating between benign and malignant lesions, most studies on CE diagnosis of cardiac masses are limited to case reports or retrospective/small-sample-sized prospective cohorts. This study aims to evaluate the diagnostic accuracy of CE in patients with suspected cardiac masses and address the insufficient evidence for differential diagnosis using CE.

METHODS

Between April 2018 and July 2022, a prospective multicenter study was conducted, which included 145 consecutive patients suspected to have cardiac masses based on transthoracic echocardiography. All patients underwent CE examinations. The echocardiographic diagnosis relied on qualitative factors such as echogenicity, boundary, morphology of the base, mass perfusion, pericardial effusion, and motility as well as quantitative factors such as the area of the masses and the peak intensity ratio of the masses to adjacent myocardium (A1/A2).

RESULTS

The final confirmed diagnoses were as follows: 2 patients had no cardiac mass, 4 patients had pseudomass, 43 patients had thrombus, 66 patients had benign tumors, and 30 patients had malignant tumors. The receiver operating characteristic (ROC) analysis indicated that an optimal A1/A2 cutoff value of 0.499 distinguished a cardiac tumor from a thrombus, with AUC, sensitivity, specificity, PPV, and NPV of 0.977, 97.9%, 90.7%, 95.9%, and 95.1%, respectively. The optimal A1/A2 cutoff value of 1.583 distinguished a cardiac tumor from a thrombus, with AUC, sensitivity, specificity, PPV, and NPV of 0.950, 93.3%, 93.9%, 87.5%, and 96.9%, respectively.

CONCLUSIONS

Combined with qualitative and quantitative analyses, CE has the potential to accurately differentiate among different types of cardiac masses.

Cardiac Masses and Pseudomasses: An Overview about Diagnostic Imaging and Clinical Background

A cardiac lesion detected at ultrasonography might turn out to be a normal structure, a benign tumor or rarely a malignancy, and lesion characterization is very important to appropriately manage the lesion itself. The exact relationship of the mass with coronary arteries and the knowledge of possible concomitant coronary artery disease are necessary preoperative information. Moreover, the increasingly performed coronary CT angiography to evaluate non-invasively coronary artery disease leads to a rising number of incidental findings. Therefore, CT and MRI are frequently performed imaging modalities when echocardiography is deemed insufficient to evaluate a lesion. A brief comprehensive overview about diagnostic radiological imaging and the clinical background of cardiac masses and pseudomasses is reported.

Clinical characteristics of primary atrial tumor and their diagnostic value: A retrospective study of 10 years

Background

Primary atrial tumors are relatively rare and predominantly benign. However, some atrial tumors may be malignant and are associated with poor outcome. Currently, it is hard to determine the malignance of atrial tumors by preoperative clinical presentation or by echocardiography. We aimed to report the difference in the clinical characteristics of patients with benign and malignant atrial tumor.

Methods

This was a single-center retrospective study. A total of 194 patients with primary atrial tumor admitted to our center between 2012 and 2021 were included. The clinical characteristics of patients with benign and malignant tumor were compared.

Results

Benign and malignant tumor accounted for 93% (n = 180) and 7% (n = 14) of the total patients, respectively. Malignant atrial tumor tended to occur in younger patients (P < 0.05), was more likely to be located at the right atrium (P < 0.05), and tended to attach to the atrial wall or valve instead of the atrial septum. Fever symptoms were more common in patients with malignant tumors than in patients with benign tumors (P < 0.05). Compared to benign tumor, patients with malignant atrial tumor also demonstrated higher rates of fever, lower rates of increasing fibrinogen, increased blood glucose (P < 0.05), significantly longer prothrombin time, and lower prothrombin activity (P < 0.05). Patients with malignant primary atrial tumor had higher mortality rate, tumor metastasis rate, and tumor recurrence rate than patients with benign primary atrial tumor (P < 0.05).

Conclusion

We compared the clinical characteristics of patients with benign and malignant atrial tumor. These findings provide valuable information to preoperatively determine the malignance of atrial tumor and thus guide surgical treatment.

The diagnostic accuracy of contrast echocardiography in patients with suspected cardiac masses: A preliminary multicenter, cross-sectional study

Background

To evaluate the diagnostic accuracy of contrast echocardiography (CE) in patients with suspected cardiac masses.

Methods

A multicenter, prospective study involving 108 consecutive patients with suspected cardiac masses based on transthoracic echocardiography performed between November 2019 and December 2020 was carried out. CE examinations were performed in all patients. The echocardiographic diagnosis was established according to the qualitative (echogenicity, boundary, morphology of the base, mass perfusion, pericardial effusion, and motility) and quantitative (area of the masses and peak intensity ratio of the masses and adjacent myocardium A1/A2) evaluations.

Results

Final confirmed diagnoses were as follows: no cardiac mass (n = 3), pseudomass (n = 3), thrombus (n = 36), benign tumor (n = 30), and malignant tumor (n = 36). ROC analysis revealed the optimal A1/A2 with cutoff value of 0.295 for a cardiac tumor from a thrombus, with AUC, sensitivity, specificity, PPV, and NPV of 0.958 (95% confidence interval (CI): 0.899–0.988), 100, 91.7, 95.7, and 100%, respectively. CE was able to distinguish malignant from benign tumors with an AUC of 0.953 (95% CI: 0.870–0.990). Multivariate logistic regression analysis revealed that tumor area, base, and A1/A2 were associated with the risk of malignant tumor (OR = 1.003, 95% CI: 1.00003–1.005; OR = 22.64, 95% CI: 1.30–395.21; OR = 165.39, 95% CI: 4.68–5,850.94, respectively). When using A1/A2 > 1.28 as the only diagnostic criterion to identify the malignant tumor, AUC, sensitivity, specificity, PPV, and NPV were 0.886 (95% CI: 0.784–0.951), 80.6, 96.7, 96.7, and 80.7%, respectively.

Conclusion

CE has the potential to accurately differentiate cardiac masses by combining qualitative and quantitative analyses. However, more studies with a large sample size should be conducted to further confirm these findings.

Clinical trial registration

http://www.chictr.org.cn/, identifier: ChiCTR1900026809.