US Risk Stratification System for Follicular Thyroid Neoplasms.

Background Preoperative assessment of follicular thyroid neoplasms is challenging using the current US risk stratification systems (RSSs) that are applicable to papillary thyroid neoplasms. Purpose To develop a US feature-based RSS for differentiating between follicular thyroid adenoma (FTA) and follicular thyroid carcinoma (FTC) in biopsy-proven follicular neoplasm and compare it with existing RSSs. Materials and Methods This retrospective multicenter study included consecutive adult patients who underwent conventional US and received a final diagnosis of follicular thyroid neoplasm from seven centers between January 2018 and December 2022. US images from a pretraining data set were used to improve readers' understanding of the US characteristics of the FTC and FTA. Univariable and multivariable logistic regression analyses were used to assess the association of qualitative US features with FTC in a training data set. Features with P < .05 were used to construct a prediction model (follicular tumor model, referred to as F model) and RSS for follicular neoplasms using the Thyroid Imaging Reporting and Data System (TI-RADS). Area under the receiver operating characteristic curve (AUC) was compared between follicular TI-RADS (hereafter, F-TI-RADS) and existing RSS (American College of Radiology [ACR] TI-RADS, Korean Society of Thyroid Radiology and Korean Society of Radiology TI-RADS [hereafter, referred to as K-TI-RADS], and Chinese TI-RADS [hereafter, referred to as C-TI-RADS]) in a validation data set. Results The pretraining, training, and validation data sets included 30 (mean age, 47.6 years ± 16.0 [SD]; 16 male patients; FTCs, 30 of 60 [50.0%]), 703 (mean age, 47.9 years ± 14.5; 530 female patients; FTCs, 188 of 703 [26.7%]), and 155 (mean age, 49.9 years ± 13.3 [SD]; 155 female patients; FTCs, 43 of 155 [27.7%]) patients. In the validation data set, the F-TI-RADS showed improved performance for differentiating between FTA and FTC (AUC, 0.81; 95% CI: 0.71, 0.86) compared with ACR TI-RADS (AUC, 0.74; 95% CI: 0.66, 0.80; P = .02), K-TI-RADS (AUC, 0.69; 95% CI: 0.61, 0.76; P = .002), and C-TI-RADS (AUC, 0.68; 95% CI: 0.60, 0.75; P = .002). Conclusion F-TI-RADS outperformed existing RSSs for differentiating between FTC and FTA. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Baumgarten in this issue.

Automatic deep learning-based colorectal adenoma detection system and its similarities with pathologists.

OBJECTIVES: The microscopic evaluation of slides has been gradually moving towards all digital in recent years, leading to the possibility for computer-aided diagnosis. It is worthwhile to know the similarities between deep learning models and pathologists before we put them into practical scenarios. The simple criteria of colorectal adenoma diagnosis make it to be a perfect testbed for this study. DESIGN: The deep learning model was trained by 177 accurately labelled training slides (156 with adenoma). The detailed labelling was performed on a self-developed annotation system based on iPad. We built the model based on DeepLab v2 with ResNet-34. The model performance was tested on 194 test slides and compared with five pathologists. Furthermore, the generalisation ability of the learning model was tested by extra 168 slides (111 with adenoma) collected from two other hospitals. RESULTS: The deep learning model achieved an area under the curve of 0.92 and obtained a slide-level accuracy of over 90% on slides from two other hospitals. The performance was on par with the performance of experienced pathologists, exceeding the average pathologist. By investigating the feature maps and cases misdiagnosed by the model, we found the concordance of thinking process in diagnosis between the deep learning model and pathologists. CONCLUSIONS: The deep learning model for colorectal adenoma diagnosis is quite similar to pathologists. It is on-par with pathologists' performance, makes similar mistakes and learns rational reasoning logics. Meanwhile, it obtains high accuracy on slides collected from different hospitals with significant staining configuration variations.

Clinically applicable histopathological diagnosis system for gastric cancer detection using deep learning.

The early detection and accurate histopathological diagnosis of gastric cancer increase the chances of successful treatment. The worldwide shortage of pathologists offers a unique opportunity for the use of artificial intelligence assistance systems to alleviate the workload and increase diagnostic accuracy. Here, we report a clinically applicable system developed at the Chinese PLA General Hospital, China, using a deep convolutional neural network trained with 2,123 pixel-level annotated H&E-stained whole slide images. The model achieves a sensitivity near 100% and an average specificity of 80.6% on a real-world test dataset with 3,212 whole slide images digitalized by three scanners. We show that the system could aid pathologists in improving diagnostic accuracy and preventing misdiagnoses. Moreover, we demonstrate that our system performs robustly with 1,582 whole slide images from two other medical centres. Our study suggests the feasibility and benefits of using histopathological artificial intelligence assistance systems in routine practice scenarios.

miR-34a regulates HDAC1 expression to affect the proliferation and apoptosis of hepatocellular carcinoma.

miR-34a is an important molecule that can inhibit the tumor growth. This study aimed to investigate the functional role of miR-34a in hepatocellular carcinoma (HCC) and explore the interaction between miR-34a and histone deacetylase 1 (HDAC1). RT-qPCR was employed to detect the mRNA expression of miR-34a and HDAC1 in 60 HCC tissues. Results showed miR-34a expression in HCC tissues was significantly lower than in normal tissues (P<0.05), but HDAC1 expression in HCC tissues was markedly higher than in normal tissues (P<0.05). In addition, miR-34a expression was negatively related to HDAC1 expression. miR-34a mimic was transfected into HCC cell lines (HepB3 and HepG2). CCK8 assay, colony formation assay and flow cytometry showed miR-34a over-expression could inhibit the proliferation of HCC cells and induce their apoptosis. Western blotting indicated miR-34a over-expression down-regulated the expression of Bcl-2, procaspase-3, procaspase-9 and c-Myc, but up-regulate p21 expression. Bioinformatics analysis indicated HDAC1 was a target gene of miR-34a. Dual Luciferase Reporter Gene Assay and retrieval assay showed miR-34a could act at the 3'UTR of HDAC1 gene to regulate its expression. Thus, miR-34a may inhibit the proliferation of HCC cells and induce their apoptosis via regulating HDAC1 expression. Our findings provide evidence for the diagnosis and therapeutic target of HCC.