Effect of ultrasound parameters of benign thyroid nodules on radiofrequency ablation efficacy.

BACKGROUND AND AIM: Ultrasound-guided radiofrequency ablation (RFA) is a minimally invasive therapy for thyroid nodules (TNs). Understanding the determinants of RFA efficacy can improve treatment and prognosis. This study aims to investigate the relationship between ultrasound parameters of benign TNs and the efficacy of RFA. METHODS: A pretest-posttest interventional study was conducted in 2021 on 250 randomly sampled patients with benign TNs, receiving RFA. For this purpose, the volume reduction (VR) and the VR ratio (VRR) of the nodules were measured at the 1-, 3-, 6-, and 12-month follow-up periods after the RFA completion. The technical success rate (TSR) of this procedure was then categorized into four states, including low (VRR < 25%), moderate (VRR = 25-49%), high (VRR = 50-74%), and very high (VRR ≥ 75). Ordered logistic regression (OLR) was further utilized to investigate the effect of the ultrasound parameters of TNs on the TSR. The analyses were notably performed using Stata 14.2. RESULTS: The VRR at the 1-, 3-, 6-, and 12-month follow-up periods were 38.7%, 53.6%, 59.3%, and 59.9%, respectively. The mean VR was also statistically significant at all follow-ups (p < 0.001). At the 1-, 3-, and 6-month follow-up periods, the VR of over 50% was observed in 28.2%, 52.1%, and 65.2% of the nodules, respectively. The odds ratios (ORs) of the RFA success were found to be 4.3 times higher for the nodules in the left lobe compared to the right lobe (OR: 4.31, p = 0.002), 6.3 times greater for isoechoic nodules compared to hyperechoic nodules (OR: 6.39, p < 0.001), 6.2 times higher for hyper-vascular nodules compared to hypo-vascular nodules (OR: 6.25, p = 0.005), and 2.3 times greater for mixed nodules compared to solid ones (OR: 2.37, p = 0.049). CONCLUSION: The ultrasound parameters of TNs had a statistically significant effect on the efficacy of RFA. Small-sized, isoechoic, and hyper-vascular nodules, as well as those with mixed tissue, were observed to respond better to RFA, leading to a better prognosis in terms of VR after treatment.

Generalizability assessment of COVID-19 3D CT data for deep learning-based disease detection.

BACKGROUND: Artificial intelligence technologies in classification/detection of COVID-19 positive cases suffer from generalizability. Moreover, accessing and preparing another large dataset is not always feasible and time-consuming. Several studies have combined smaller COVID-19 CT datasets into "supersets" to maximize the number of training samples. This study aims to assess generalizability by splitting datasets into different portions based on 3D CT images using deep learning. METHOD: Two large datasets, including 1110 3D CT images, were split into five segments of 20% each. Each dataset's first 20% segment was separated as a holdout test set. 3D-CNN training was performed with the remaining 80% from each dataset. Two small external datasets were also used to independently evaluate the trained models. RESULTS: The total combination of 80% of each dataset has an accuracy of 91% on Iranmehr and 83% on Moscow holdout test datasets. Results indicated that 80% of the primary datasets are adequate for fully training a model. The additional fine-tuning using 40% of a secondary dataset helps the model generalize to a third, unseen dataset. The highest accuracy achieved through transfer learning was 85% on LDCT dataset and 83% on Iranmehr holdout test sets when retrained on 80% of Iranmehr dataset. CONCLUSION: While the total combination of both datasets produced the best results, different combinations and transfer learning still produced generalizable results. Adopting the proposed methodology may help to obtain satisfactory results in the case of limited external datasets.

Radiofrequency ablation of parathyroid adenomas causing primary hyperparathyroidism: A report of 27 patients.

PURPOSE: To study treatment outcome of parathyroid adenomas using ultrasound-guided radiofrequency ablation. METHODS: Twenty-seven patients with a single adenoma of the parathyroid gland were included in the study. Using color Doppler ultrasonography, the lesion and its characteristics were determined, and dextrose was injected to dissect the gland from the surrounding structures. The ablation process was done with 6-12 watts of power. RESULTS: No complications were seen in any of the subjects. A significant reduction was seen in serum parathyroid hormone (PTH) and calcium levels after treatment. PTH levels showed a median decrease of 13.8%, and a median decrease of 8.2% was seen in serum calcium levels (p < 0.001). Phosphorus levels did not change significantly after treatment. In 1-month follow-up of patients, the lesion size had decreased considerably. In long-term follow-up, 11 of 20 patients having subsequent imaging had indistinguishable lesions. CONCLUSION: Our results showed that RFA of parathyroid adenomas caused a significant reduction in biomedical indicators of disease and resulted in a significant reduction or disappearance of the lesion in the majority of the patients while having no considerable complications.

Ultrasound-Guided Radiofrequency Ablation of Locally Recurrent Thyroid Carcinoma.

INTRODUCTION: We aimed to evaluate the effectiveness and safety of radiofrequency ablation (RFA) for non-surgical treatment of locally recurrent thyroid cancers, in both well-differentiated and medullary thyroid carcinomas (DTC and MTC) that are not amenable to traditional treatments. METHODS: We conducted a retrospective review of 48 patients with 103 recurrent tumors (81 DTC, 22 MTC) who underwent ultrasound-guided RFA. Patients were followed for 12-37 months to observe the outcomes and complications. RESULTS: 64 tumors (62.1%) completely disappeared at the last follow-up visit with 61 (59.2%) being resolved within 12 months. Technical success (volume reduction ratio (VRR) > 50%) was 96% (n = 99) in all tumors. The mean largest diameter of treated tumors decreased from 11.2 ± 5.3 to 2.4 ± 3.4 mm (p value < 0.001), and the mean volume decreased from 501.0 ± 807.0 to 41.6 ± 97.1 mm3 at the last follow-up (mean VRR = 91%). Our patients had a 77.1% recurrence-free survival rate (11 recurrences, 7 DTC, 4 MTC), with an overall mean recurrence-free survival time of 34.6 months (95% confidence interval, 30.0-39.1). We observed 3 cases with complications (voice changes in DTC patients) that completely resolved during follow-ups. CONCLUSION: RFA is a safe and effective alternative to repetitive surgeries in recurrent loco-regional DTCs as well as MTCs. LEVEL OF EVIDENCE: Level 3, Non-randomized controlled cohort/follow-up study.

Factors determining generalization in deep learning models for scoring COVID-CT images.

The COVID-19 pandemic has inspired unprecedented data collection and computer vision modelling efforts worldwide, focused on the diagnosis of COVID-19 from medical images. However, these models have found limited, if any, clinical application due in part to unproven generalization to data sets beyond their source training corpus. This study investigates the generalizability of deep learning models using publicly available COVID-19 Computed Tomography data through cross dataset validation. The predictive ability of these models for COVID-19 severity is assessed using an independent dataset that is stratified for COVID-19 lung involvement. Each inter-dataset study is performed using histogram equalization, and contrast limited adaptive histogram equalization with and without a learning Gabor filter. We show that under certain conditions, deep learning models can generalize well to an external dataset with F1 scores up to 86%. The best performing model shows predictive accuracy of between 75% and 96% for lung involvement scoring against an external expertly stratified dataset. From these results we identify key factors promoting deep learning generalization, being primarily the uniform acquisition of training images, and secondly diversity in CT slice position.

Hydrodissection: A Novel Approach for Safe Core Needle Biopsy of Small High-Risk Subcapsular Thyroid Nodules.

PURPOSE: To evaluate the safety and effectiveness of core needle biopsy (CNB) under the assistance of hydrodissection (HDS). MATERIALS AND METHODS: Of 2325 patients requiring biopsy of thyroid lesions, 21 high-risk patients with subcapsular nodules smaller than 10 mm were recruited into this study. All patients underwent HDS with 0.9% saline solution followed by ultrasound (US)-guided CNB with an 18-gauge semi-automated biopsy needle. The separation success rate (SSR) of the HDS, technical success rate (TSR) of CNB, histopathologic success rate (HSR), and complications were assessed. RESULTS: Both the SSR of HDS and TSR of CNB were 100% (21/21). The HSR of the thyroid nodules was 85.7% (18/21). No major complications were recorded. CONCLUSION: HDS before CNB can successfully lead to safe biopsy of small subcapsular thyroid nodules. LEVEL OF EVIDENCE: Level 4, Case Series.