Purpose: This study aimed to compare: the performance of K-TIRADS, EU-TIRADS and ACR TIRADS when used by observers with different levels of experience compared with the gold standard of cytology, and to evaluate the diagnostic performance of CAD (computer-aided design) compared with TI-RADS systems. Methods and Materials: In total, 323 thyroid nodules were evaluated in patients who were candidates for needle aspiration. Three observers with different levels of experience evaluated the diagnostic accuracy of three risk stratification systems (ACR TI-RADS, EU-TIRADS and K-TIRADS) and CAD software (S-Detect, made by Samsung) in characterizing the nodules. The results were compared with cytology examination. All nodules were characterized in terms of shape, margins, composition, calcifications, size, echogenicity and microcalcifications, and by stratifying individual nodules by using the three TIRADS systems; then S-detect software was applied and the data were compared with each other and with the gold standard. Results: Through cytology, 308 benign and 33 malignant nodules were identified. ACR-TIRADS showed a sensitivity of 100%, a specificity of 86%, a positive predictive value of 43% and a negative predictive value of 100%. EU-TIRADS showed a sensitivity of 100%, a specificity of 79%, a positive predictive value of 33% and a negative predictive value of 100%. K-TIRADS showed a sensitivity of 100%, a specificity of 89%, a positive predictive value of 50% and a negative predictive value of 100%. S-Detect combined with EU-TIRADS showed a high agreement (>95%) with the gold standard. Conclusions: K-TIRADS’s positive predictive power was slightly better than the other TIRADS, suggesting greater accuracy in correctly diagnosing positive cases. S-DETECT combined with EU-TIRADS has similar results to S-Detect with ACR- and K-TIRADS in terms of sensitivity, specificity and negative predictive power. However, it has a slightly better positive predictive power, suggesting greater accuracy in correctly diagnosing positive cases than the ACR- and K-TIRADS classification systems. In general, S-Detect cannot yet be considered a substitute for the human observer but only as an important support for human evaluation and an excellent and fast help to provide a comprehensive and complete report. Clinical Relevance/Application: S-Detect is a valuable tool for characterizing thyroid nodules when integrated with radiologist evaluation. It is also an important support tool for less experienced observers. Particularly interesting is the approach of use in integrated combination of the K-TIRADS by the human observer with S-Detect using EU-TIRADS, which could increase the overall diagnostic efficiency of the systems.

Background/Objectives: Thyroid nodules are a very common finding, mostly benign but sometimes malignant, and thus require accurate diagnosis. Ultrasound and fine needle biopsy are the most widely used and reliable diagnostic methods to date, but they are sometimes limited in addressing benign from malignant nodules, mainly with regard to ultrasound, by the operator’s experience. Radiomics, quantitative feature extraction from medical images and machine learning offer promising avenues to improve diagnosis. The aim of this work was to develop a machine learning model based on thyroid ultrasound images to classify nodules into benign and malignant classes. Methods: For this purpose, images of ultrasonography from 142 subjects were collected. Among these subjects, 40 patients (28.2%) belonged to the class “malignant” and 102 patients (71.8%) belonged to the class “benign”, according to histological diagnosis from fine-needle aspiration. This image set was used for the training, cross-validation and internal testing of three different machine learning models. A robust radiomic approach was applied, under the hypothesis that the radiomic feature could capture the disease heterogeneity among the two groups. Three models consisting of four ensembles of machine learning classifiers (random forests, support vector machines and k-nearest neighbor classifiers) were developed for the binary classification task of interest. The best performing model was then externally tested on a cohort of 21 new patients. Results: The best model (ensemble of random forest) showed Receiver Operating Characteristic-Area Under the Curve (ROC-AUC) (%) of 85 (majority vote), 83.7 ** (mean) [80.2–87.2], accuracy (%) of 83, 81.2 ** [77.1–85.2], sensitivity (%) of 70, 67.5 ** [64.3–70.7], specificity (%) of 88, 86.5 ** [82–91], positive predictive value (PPV) (%) of 70, 66.5 ** [57.9–75.1] and negative predictive value (NPV) (%) of 88, 87.1 ** [85.5–88.8] (* p < 0.05, ** p < 0.005) in the internal test cohort. It achieved an accuracy of 90.5%, a sensitivity of 100%, a specificity of 86.7%, a PPV of 75% and an NPV of 100% in the external testing cohort. Conclusions: The model constituted of four ensembles of random forest classifiers could identify all the malignant nodes and the consistent majority of benign in the external testing cohort.

Background: Renal transplant (RT) remains the optimal treatment for end-stage renal disease and early complications might be detected in the postoperative period to improve long-term outcomes. To this regard, contrast enhanced ultrasound (CEUS) could be utilized to evaluate RT functional recovery and potentially detect acute rejection (AR) and/or renal ischemia signs. Materials and methods: Observational study of 107 consecutive patients waitlisted for RT. Participants underwent conventional ultrasound (CUS) and color-doppler-ultrasound (CDUS) to evaluate resistive index of segmental and interlobar arteries and quantitative CEUS techniques recording the following perfusion parameters: peak intensity (PI-c), rising time (RT-c), area under (AUC-c) the time intensity curve (TIC), time to peak (TTP-c) and mean transit time (MTT-c). Results: CEUS parameters sensibility and specificity to predict AR in the early post-operative period resulted in: 90% and 69% for PI-c, 95% and 64% for RT-c, 85% and 65% for AUC-c. The overall diagnostic performance of these three CEUS parameters in comparison to the same in CUS and CDUS resulted in a sensitivity and specificity of 95% and 49%, versus 85% and 46%, respectively, therefore CEUS examination with the analysis of PI-c, RT-c and AUC-c values increases the diagnostic sensitivity in predicting AR by approximately 15–20% compared to CDUS and by 30–40% compared to CUS. Conclusion: CEUS could be routinely included in RT follow-up, as it shows to be a non-invasive helpful diagnostic tool for early detection of renal graft complications, selecting patients eventually in need of further confirmation.