Strategies for integrating artificial intelligence (AI) into thyroid nodule management require additional development and testing. We developed a deep-learning AI model (ThyNet) to differentiate between malignant tumours and benign thyroid nodules and aimed to investigate how ThyNet could help radiologists improve diagnostic performance and avoid unnecessary fine needle aspiration. ThyNet was developed and trained on 18 049 images of 8339 patients (training set) from two hospitals (the First Affiliated Hospital of Sun Yat-sen University, Guangzhou, China, and Sun Yat-sen University Cancer Center, Guangzhou, China) and tested on 4305 images of 2775 patients (total test set) from seven hospitals (the First Affiliated Hospital of Guangzhou University of Chinese Medicine, Guangzhou, China; the Sixth Affiliated Hospital of Sun Yat-sen University, Guangzhou, China; the Guangzhou Army General Hospital, Guangzhou, China; the Third Affiliated Hospital of Sun Yat-sen University, Guangzhou, China; the First Affiliated Hospital of Sun Yat-sen University; Sun Yat-sen University Cancer Center; and the First Affiliated Hospital of Guangxi Medical University, Nanning, China) in three stages. All nodules in the training and total test set were pathologically confirmed. The diagnostic performance of ThyNet was first compared with 12 radiologists (test set A); a ThyNet-assisted strategy, in which ThyNet assisted diagnoses made by radiologists, was developed to improve diagnostic performance of radiologists using images (test set B); the ThyNet assisted strategy was then tested in a real-world clinical setting (using images and videos; test set C). In a simulated scenario, the number of unnecessary fine needle aspirations avoided by ThyNet-assisted strategy was calculated. The area under the receiver operating characteristic curve (AUROC) for accurate diagnosis of ThyNet (0·922 [95% CI 0·910-0·934]) was significantly higher than that of the radiologists (0·839 [0·834-0·844]; p<0·0001). Furthermore, ThyNet-assisted strategy improved the pooled AUROC of the radiologists from 0·837 (0·832-0·842) when diagnosing without ThyNet to 0·875 (0·871-0·880; p<0·0001) with ThyNet for reviewing images, and from 0·862 (0·851-0·872) to 0·873 (0·863-0·883; p<0·0001) in the clinical test, which used images and videos. In the simulated scenario, the number of fine needle aspirations decreased from 61·9% to 35·2% using the ThyNet-assisted strategy, while missed malignancy decreased from 18·9% to 17·0%. The ThyNet-assisted strategy can significantly improve the diagnostic performance of radiologists and help reduce unnecessary fine needle aspirations for thyroid nodules. National Natural Science Foundation of China and Guangzhou Science and Technology Project.

Background and Purpose: Thioredoxin‐interacting protein (TXNIP), a regulator of cellular oxidative stress, has been associated with activation of NOD‐like receptor 3 (NLRP3) inflammasome, inflammation and lipid metabolism, suggesting it has a role in the pathogenesis of non‐alcoholic fatty liver disease (NAFLD) in diabetes. In this study we investigated whether TXNIP is involved in type 1 diabetes‐associated NAFLD and whether antioxidants, quercetin and allopurinol, alleviate NAFLD by targeting TXNIP.Experimental Approach: Diabetes was induced in male Sprague‐Dawley rats by a single i.p. injection of 55 mg·kg⁻¹ streptozotocin. Quercetin and allopurinol were given p.o. to diabetic rats for 7 weeks. Hepatic function, oxidative stress, inflammation and lipid levels were determined. Rat BRL‐3A and human HepG2 cells were exposed to high glucose (30 mM) in the presence and absence of antioxidants, TXNIP siRNA transfection or caspase‐1 inhibitor, Ac‐YVAD‐CMK.Key Results: Quercetin and allopurinol significantly inhibited the TXNIP overexpression, activation of NLRP3 inflammasome, down‐regulation of PPARα and up‐regulation of sterol regulatory element binding protein‐1c (SREBP‐1c), SREBP‐2, fatty acid synthase and liver X receptor α, as well as elevation of ROS and IL‐1β in diabetic rat liver. These effects were confirmed in hepatocytes in vitro and it was further shown that TXNIP down‐regulation contributed to the suppression of NLRP3 inflammasome activation, inflammation and changes in PPARα and SREBPs.Conclusions and Implications: Inhibition of hepatic TXNIP by quercetin and allopurinol contributes to the reduction in liver inflammation and lipid accumulation under hyperglycaemic conditions. The targeting of hepatic TXNIP by quercetin and allopurinol may have therapeutic implications for prevention of type 1 diabetes‐associated NAFLD.

To construct a prediction model based on peritumoral radiomics signatures from CT images and investigate its efficiency in predicting early recurrence (ER) of hepatocellular carcinoma (HCC) after curative treatment. In total, 156 patients with primary HCC were randomly divided into the training cohort (109 patients) and the validation cohort (47 patients). From the pretreatment CT images, we extracted 3-phase two-dimensional images from the largest cross-sectional area of the tumor. A region of interest (ROI) was manually delineated around the lesion for tumoral radiomics (T-RO) feature extraction, and another ROI was outlined with an additional 2 cm peritumoral area for peritumoral radiomics (PT-RO) feature extraction. The least absolute shrinkage and selection operator (LASSO) logistic regression model was applied for feature selection and model construction. The T-RO and PT-RO models were constructed. In the validation cohort, the prediction efficiencies of the two models and peritumoral enhancement (PT-E) were evaluated qualitatively by receiver operating characteristic (ROC) curves, calibration curves and decision curves and quantitatively by area under the curve (AUC), the category-free net reclassification index (cfNRI) and integrated discrimination improvement values (IDI). By comparing AUC values, the prediction accuracy in the validation cohort was good for the PT-RO model (0.80 vs. 0.79, P = 0.47) but poor for the T-RO model (0.82 vs. 0.62, P < 0.01), which was significantly overfitted. In the validation cohort, the ROC curves, calibration curves and decision curves indicated that the PT-RO model had better calibration efficiency and provided greater clinical benefits. CfNRI indicated that the PT-RO model correctly reclassified 47% of ER patients and 32% of non-ER patients compared to the T-RO model (P < 0.01); additionally, the PT-RO model correctly reclassified 24% of ER patients and 41% of non-ER patients compared to PT-E (P = 0.02). IDI indicated that the PT-RO model could improve prediction accuracy by 0.22 (P < 0.01) compared to the T-RO model and by 0.20 (P = 0.01) compared to PT-E. The CT-based PT-RO model can effectively predict the ER of HCC and is more efficient than the T-RO model and the conventional imaging feature PT-E.

Background Large language models (LLMs) hold substantial promise for medical imaging interpretation. However, there is a lack of studies on their feasibility in handling reasoning questions associated with medical diagnosis. Purpose To investigate the viability of leveraging three publicly available LLMs to enhance consistency and diagnostic accuracy in medical imaging based on standardized reporting, with pathology as the reference standard. Materials and Methods US images of thyroid nodules with pathologic results were retrospectively collected from a tertiary referral hospital between July 2022 and December 2022 and used to evaluate malignancy diagnoses generated by three LLMs-OpenAI's ChatGPT 3.5, ChatGPT 4.0, and Google's Bard. Inter- and intra-LLM agreement of diagnosis were evaluated. Then, diagnostic performance, including accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC), was evaluated and compared for the LLMs and three interactive approaches: human reader combined with LLMs, image-to-text model combined with LLMs, and an end-to-end convolutional neural network model. Results A total of 1161 US images of thyroid nodules (498 benign, 663 malignant) from 725 patients (mean age, 42.2 years ± 14.1 [SD]; 516 women) were evaluated. ChatGPT 4.0 and Bard displayed substantial to almost perfect intra-LLM agreement (κ range, 0.65-0.86 [95% CI: 0.64, 0.86]), while ChatGPT 3.5 showed fair to substantial agreement (κ range, 0.36-0.68 [95% CI: 0.36, 0.68]). ChatGPT 4.0 had an accuracy of 78%-86% (95% CI: 76%, 88%) and sensitivity of 86%-95% (95% CI: 83%, 96%), compared with 74%-86% (95% CI: 71%, 88%) and 74%-91% (95% CI: 71%, 93%), respectively, for Bard. Moreover, with ChatGPT 4.0, the image-to-text-LLM strategy exhibited an AUC (0.83 [95% CI: 0.80, 0.85]) and accuracy (84% [95% CI: 82%, 86%]) comparable to those of the human-LLM interaction strategy with two senior readers and one junior reader and exceeding those of the human-LLM interaction strategy with one junior reader. Conclusion LLMs, particularly integrated with image-to-text approaches, show potential in enhancing diagnostic medical imaging. ChatGPT 4.0 was optimal for consistency and diagnostic accuracy when compared with Bard and ChatGPT 3.5. © RSNA, 2024 Supplemental material is available for this article.

Radiomics reflects the texture and morphological features of tumours by quantitatively analysing the grey values of medical images. We aim to develop a nomogram incorporating radiomics and the Breast Imaging Reporting and Data System (BI-RADS) for predicting breast cancer in BI-RADS ultrasound (US) category 4 or 5 lesions. From January 2017 to August 2018, a total of 315 pathologically proven breast lesions were included. Patients from the study population were divided into a training group (n = 211) and a validation group (n = 104) according to a cut-off date of March 1^st, 2018. Each lesion was assigned a category (4A, 4B, 4C or 5) according to the second edition of the American College of Radiology (ACR) BI-RADS US. A radiomics score was generated from the US image. A nomogram was developed based on the results of multivariate regression analysis from the training group. Discrimination, calibration and clinical usefulness of the nomogram for predicting breast cancer were assessed in the validation group. The radiomics score included 9 selected radiomics features. The radiomics score and BI-RADS category were independently associated with breast malignancy. The nomogram incorporating the radiomics score and BI-RADS category showed better discrimination (area under the receiver operating characteristic curve [AUC]: 0.928; 95% confidence interval [CI]: 0.876, 0.980) between malignant and benign lesions than either the radiomics score (P = 0.029) or BI-RADS category (P = 0.011). The nomogram demonstrated good calibration and clinical usefulness. In conclusion, the nomogram combining the radiomics score and BI-RADS category is potentially useful for predicting breast malignancy in BI-RADS US category 4 or 5 lesions.