Giá trị của siêu âm sử dụng bảng phân loại ACR - TIRADS 2017 trong chẩn đoán nhân giáp

TÓM TẮT Đặt vấn đề: Bệnh lý nhân giáp là một bệnh lý phổ biến, đặc biệt là ở phụ nữ và người lớn tuổi. Siêu âm tuyến giáp, được xem như là một phương tiện đầu tay, là phương pháp chẩn đoán hình ảnh có những khả năng vượt trội như tương đối đơn giản, rẻ tiền, không xâm lấn, có thể lặp lại nhiều lần để chẩn đoán bệnh, và có khả năng phát hiện được những tổn thương rất nhỏ. Nghiên cứu này nhằm xác định xác giá trị của siêu âm sử dụng bảng phân loại ACR-TIRADS 2017 trong chẩn đoán nhân giáp. Phương pháp: Thiết kế nghiên cứu mô tả cắt ngang, với cỡ mẫu 169 bệnh nhân được phẫu thuật nhân giáp. Trước phẫu thuật, bệnh nhân được siêu âm tuyến giáp bằng máy GE (LOGIQ S7 Pro, LOGIQ E9 …) với đầu dò linear tần số 7,5 - 12 MHz. Kết quả siêu âm bảng phân loại TI-RADS theo ACR 2017 so sánh với tiêu chuẩn vàng là kết quả giải phẫu bệnh. Kết quả: Siêu âm áp dụng bảng phân loại ACR-TIRADS 2017 trong phân biệt nhân giáp lành tính và ác tính: Độ nhạy 97,9%, độ đặc hiệu 82,6%, giá trị tiên đoán dương 95,8%, giá trị tiên đoán âm 90,5%, và độ chính xác 94,9%. Diện tích dưới đường cong ROC (AUC) của phân loại ACR-TIRADS trong chẩn đoán nhân giáp ác tính là bằng 0,953 (p < 0,001). Điểm cắt (cut - off) được chọn là TIRADS 4. Diện tích dưới đường cong ROC (AUC) của điểm số của hạt giáp theo phân loại ACR- là 0,967 (p < 0,001). Điểm cắt (cut - off) được chọn là 5 điểm. Kết luận: Siêu âm áp dụng bảng phân loại ACR-TIRADS 2017 có giá trị trong chẩn đoán phân biệt nhân giáp lành tính và ác tính với độ nhạy và độ đặc hiệu cao. ABSTRACT THE USE OF THYROIDULTRASOUND WITH ACR - TIRADS 2017 CLASSIFICATION IN THE DIAGNOSIS OF THYROID NODULES Backgrounds: Thyroid disease is very common, especially in women and the elderly. Thyroid ultrasound, as a first - line tool, is an imaging modality with outstanding capabilities such as being relatively simple, inexpensive, non - invasive, and repeatable for diagnosis of thyroid diseases, and can detect very small lesions. This study aims to determine the use of thyroid ultrasound with ACR-TIRADS 2017 classification in the diagnosis of thyroid nodules. Methods: A cross - sectional descriptive study was conducted in 169 patients undergoing thyroidectomy. All patients had been preoperatively performed thyroid ultrasound using a GE machine (LOGIQ S7 Pro, LOGIQ E9 ...) with a linear transducer frequency of 7.5 - 12 MHz. The ultrasound results using the 2017 ACR-TIRADS classification compared with pathological findings as the gold standard diagnostics. Results: Thyroid ultrasound using the 2017 ACR-TIRADS classification could distinguish benign and malignant thyroid nodules with the sensitivity of 97.9%, specificity 82.6%, positive predictive value 95.8%, negative predictive value 90.5%, and accuracy of 94.9%. The area under the ROC curve (AUC) of the ACRTIRADS classification in the diagnosis of malignant thyroid nodules was 0.953 (p < 0.001). The cut - off point was selected as TIRADS 4. The area under the ROC curve (AUC) of the ACR - classification score of the armor particles was 0.967 (p < 0.001). The cut - off point is selected as 5 points. Conclusion: Thyroid ultrasound using the 2017 ACR-TIRADS classification is valuable in the differential diagnosis of benign and malignant thyroid nodules with high sensitivity and specificity. Keywords: Ultrasound, thyroid nodules, ACR-TIRADS 2017, benign, malignant.

Ambiguous and Incomplete: Natural Language Processing Reveals Problematic Reporting Styles in Thyroid Ultrasound Reports

Objective Natural language processing (NLP) systems convert unstructured text into analyzable data. Here, we describe the performance measures of NLP to capture granular details on nodules from thyroid ultrasound (US) reports and reveal critical issues with reporting language. Methods We iteratively developed NLP tools using clinical Text Analysis and Knowledge Extraction System (cTAKES) and thyroid US reports from 2007 to 2013. We incorporated nine nodule features for NLP extraction. Next, we evaluated the precision, recall, and accuracy of our NLP tools using a separate set of US reports from an academic medical center (A) and a regional health care system (B) during the same period. Two physicians manually annotated each test-set report. A third physician then adjudicated discrepancies. The adjudicated “gold standard” was then used to evaluate NLP performance on the test-set. Results A total of 243 thyroid US reports contained 6,405 data elements. Inter-annotator agreement for all elements was 91.3%. Compared with the gold standard, overall recall of the NLP tool was 90%. NLP recall for thyroid lobe or isthmus characteristics was: laterality 96% and size 95%. NLP accuracy for nodule characteristics was: laterality 92%, size 92%, calcifications 76%, vascularity 65%, echogenicity 62%, contents 76%, and borders 40%. NLP recall for presence or absence of lymphadenopathy was 61%. Reporting style accounted for 18% errors. For example, the word “heterogeneous” interchangeably referred to nodule contents or echogenicity. While nodule dimensions and laterality were often described, US reports only described contents, echogenicity, vascularity, calcifications, borders, and lymphadenopathy, 46, 41, 17, 15, 9, and 41% of the time, respectively. Most nodule characteristics were equally likely to be described at hospital A compared with hospital B. Conclusions NLP can automate extraction of critical information from thyroid US reports. However, ambiguous and incomplete reporting language hinders performance of NLP systems regardless of institutional setting. Standardized or synoptic thyroid US reports could improve NLP performance.

Evaluation of a Clinical Pathway for Thyroid Nodular Disease: Timings and Delays in the Diagnosis and Treatment of Thyroid Cancer

Background: Due to the high prevalence of nodular thyroid disease in the general population and the need to rule out malignant tumours, a clinical pathway for nodular thyroid disease was created at our tertiary-level hospital. Our study aimed to quantify timings and delays in diagnosis and treatment in this clinical pathway, specifically for patients who were diagnosed with thyroid cancer. Methods: A retrospective review was conducted of patients who were newly diagnosed with thyroid cancer and who had been previously evaluated in the clinical pathway for nodular thyroid disease at our institution during 2015–2017. Patient demographics, previous diagnostic studies, cytological results, tumour details and key dates were analysed to identify wait times in diagnosis and treatment. Results: Forty patients with thyroid cancer were included. The diagnostic delay had a median time of 60 days, and the treatment delay was dependent on cytopathological results. The main cause for delay in the diagnostic phase was the timing of the thyroid ultrasound performed by the radiology department. In the treatment phase, patients with a cytological result of Bethesda III, V or VI underwent surgery at the suggested time, while those in the Bethesda II or IV category did not. Conclusions: The major delay found in the diagnostic phase was the timing of the thyroid ultrasound performed by the radiology department. We are not suggesting that this step must be eliminated, though the implementation of routine ultrasonography in a thyroid clinic can help identify patients who need more urgent evaluation for fine needle aspiration cytology. In our hospital, decision for surgery is based mainly on the cytopathological report. Imaging studies and/or molecular testing could be considered to reduce treatment delays.

Age-specific correlation between thyroid hormone concentrations and ultrasound thyroid volume for diagnosing thyroid dysfunction in preterm infants: a single-centre, prospective, observational study protocol

IntroductionThyroid disorders are commonly concomitant with premature birth; however, indications to start therapy remain unclear due to lack of gestational age-specific reference ranges and thyroid ultrasound nomograms. We aim to evaluate the age-specific correlation between circulating free thyroxine (FT4) and thyrotropin stimulating hormone (TSH) levels and ultrasound thyroid volume to assist identify infants requiring thyroid hormone replacement therapy.Methods and analysisThis is an observational, prospective, single-centre study that will include 200 preterm infants born between 24 and 32 weeks of gestational age, without any congenital diseases or malformation that may affect thyroid function. Venous blood will be obtained in infants at 14–21 days of life, and at 32 and 36 weeks of postconceptional age (PCA) to measure FT4 and TSH concentrations. Thyroid ultrasound will be performed at 32 and 36 weeks of PCA. Relevant outcomes will include determination of FT4 and TSH values and ultrasound thyroid volume for preterm infants born at 24–28 weeks of gestation and 29–32 weeks of gestation. Correlations among circulating hormone concentrations and thyroid volumes with the head circumference and body mass will also be determined.Ethics and disseminationThe Ethics Committee of the Medical University of Warsaw has approved the study protocol prior to recruitment (KB44/2019). Informed consent will be obtained from caretakers of preterm infants at the time of enrolment. Consent for participation in the study can be withdrawn at any time, without consequences and without obligation to justify the decision. All data will be stored in a secure, password-protected Excel file that is only accessible to researchers involved in the study. Findings will be published in a peer-reviewed journal and disseminated at relevant national and international conferences.Trial registration numberNCT04208503.

Diagnostic accuracy of B-mode ultrasound, ultrasound elastography and diffusion weighted MRI in differentiation of thyroid nodules (prospective study)

Background The incidence of the thyroid nodules and its detection is increasing rapidly. The most precise method for diagnosis of the nodules of the thyroid is FNAC. But, about 10–20% of specimens of FNAC are indeterminate and non-diagnostic. Therefore, there is a demand for another diagnostic method for evaluating thyroid nodules. Thyroid ultrasound elastography may improve the ability to differentiate malignant from benign thyroid nodules. Few articles were published about the results of DW MRI in thyroid nodules, with its results confirmed that malignant nodules have lower mean ADC values than benign nodules. This study aims to investigate and compare the accuracy of B-mode ultrasound, ultrasound elastography and diffusion-weighted MRI in characterization of the nodules of the thyroid. Results The study included 56 patients with thyroid nodules (36 benign and 20 malignant). Thyroid ultrasound, ultrasound elastography and DWI were done for all patients. Ultrasound-guided FNA Cytological examination (as the gold standard) was done for 48 patients and surgical histopathology was done to 8 patients with non-diagnostic FNAC. The results showed: TIRADS score had sensitivity 90%, specificity 77.8% and accuracy of 82.14%. The elastography score had sensitivity 80%, specificity 88.9% and accuracy 85.7%. The use of the strain ratio had 80% sensitivity, 94.4% specificity and 89.3% accuracy. DWI and ADC value had 100% sensitivity and 94.4% specificity and the accuracy was 96.4% for differentiating malignant from benign thyroid nodules. Multi-parametric analysis by TIRADS and ADC had 100% accuracy. Conclusion Ultrasound elastography add valuable data over ultrasound TIRADS. But, diffusion weighted MRI and ADC value has more accuracy in differentiating malignant from benign thyroid nodules. The best performance was achieved by the combination of ACR-TIRADS and ADC value.

Validation of Four Thyroid Ultrasound Risk Stratification Systems in Patients with Hashimoto’s Thyroiditis; Impact of Changes in the Threshold for Nodule’s Shape Criterion

The aim of the study was to validate thyroid US malignancy features, especially the nodule’s shape, and selected Thyroid Imaging Reporting and Data Systems (EU-TIRADS; K-TIRADS; ACR-TIRADS, ATA guidelines) in patients with or without Hashimoto’s thyroiditis (HT and non-HT groups). The study included 1188 nodules (HT: 358, non-HT: 830) with known final diagnoses. We found that the strongest indications of nodule’s malignancy were microcalcifications (OR: 22.7) in HT group and irregular margins (OR:13.8) in non-HT group. Solid echostructure and macrocalcifications were ineffective in patients with HT. The highest accuracy of nodule’s shape criterion was noted on transverse section, with the cut-off value of anteroposterior to transverse dimension ratio (AP/T) close to 1.15 in both groups. When round nodules were regarded as suspicious in patients with HT (the cut-off value of AP/T set to ≥1), it led to a three-fold increase in sensitivity of this feature, with a disproportionally lower decrease in specificity and similar accuracy. Such a modification was effective also for cancers other than PTC. The diagnostic effectiveness of analyzed TIRADS in patients with HT and without HT was similar. Changes in the threshold for AP/T ratio influenced the number of nodules classified into the category of the highest risk, especially in the case of EU-TIRADS.

A New Classification Method in Ultrasound Images of Benign and Malignant Thyroid Nodules Based on Transfer Learning and Deep Convolutional Neural Network

Thyroid nodule is a clinical disorder with a high incidence rate, with large number of cases being detected every year globally. Early analysis of a benign or malignant thyroid nodule using ultrasound imaging is of great importance in the diagnosis of thyroid cancer. Although the b-mode ultrasound can be used to find the presence of a nodule in the thyroid, there is no existing method for an accurate and automatic diagnosis of the ultrasound image. In this pursuit, the present study envisaged the development of an ultrasound diagnosis method for the accurate and efficient identification of thyroid nodules, based on transfer learning and deep convolutional neural network. Initially, the Total Variation- (TV-) based self-adaptive image restoration method was adopted to preprocess the thyroid ultrasound image and remove the boarder and marks. With data augmentation as a training set, transfer learning with the trained GoogLeNet convolutional neural network was performed to extract image features. Finally, joint training and secondary transfer learning were performed to improve the classification accuracy, based on the thyroid images from open source data sets and the thyroid images collected from local hospitals. The GoogLeNet model was established for the experiments on thyroid ultrasound image data sets. Compared with the network established with LeNet5, VGG16, GoogLeNet, and GoogLeNet (Improved), the results showed that using GoogLeNet (Improved) model enhanced the accuracy for the nodule classification. The joint training of different data sets and the secondary transfer learning further improved its accuracy. The results of experiments on the medical image data sets of various types of diseased and normal thyroids showed that the accuracy rate of classification and diagnosis of this method was 96.04%, with a significant clinical application value.