Semi-supervised Network for Thyroid Nodule Segmentation via Joint Consistency Learning and Co-Training

Background: The incidence rate of thyroid cancer in the US has increased significantly over the last few decades. However, besides radiation exposure and obesity, no modifiable risk factors have been identified for thyroid cancer. Although theories suggest exposure to per- and polyfluoroalkyl substances (PFAS) could contribute to the development of papillary thyroid cancer, existing studies on the effects of PFAS on thyroid cancer risk do not yield consensus results. Using data for the state of Florida, this ecological study aims to determine whether incidence rates of thyroid cancer vary with respect to residential proximity to industrial facilities handling PFAS materials as well as living in areas where drinking water containing detectable levels of PFAS. Methods: Florida cancer registry data from 2011 to 2020 was used to calculate the incidence rates of papillary thyroid cancer by race/ethnicity (Hispanic, non-Hispanic White (NHW), and non-Hispanic Black (NHB)) for all census tracts in Florida. Environmental Protection Agency data was used to identify the locations of industrial facilities handling PFAS as well as ZIP codes with public drinking water testing positive for PFAS above reportable levels. A geographic information system was used to calculate the number of PFAS facilities within 10KM of a census tract. Correlation, linear regression, and t-tests for sample mean comparison were used to assess the associations between PFAS exposure and thyroid cancer rates at the census tract level. Results: A total of 14,994 cases (9,213 in NHWs, 3,995 in Hispanics, and 989 in NHB) of thyroid cancers were diagnosed in Florida from 2011 to 2020. Correlation analyses show a positive correlation between residential proximity to PFAS facilities and greater Hispanic and NHB population compositions by higher percentages of (r=0.36, p<0.001) with lower income (r=-0.192, p<0.001)) and higher obesity rate (r=0.197, p<0.001)). Regression analyses show that proximity to PFAS facilities is a significant factor for thyroid cancer rates in Hispanic (p=0.006) and NHW (p<0.001). Hispanic and NHW living are in IPZIP code areas with reportable PFAS in drinking water also have higher average thyroid cancer rates than those in areas with no reportable PFAS (t-test p=0.036 for Hispanic and p=0.039 for NHW). With a lower average incidence rate than Hispanic and NHW, thyroid cancer rates in NHB show no statistically significant associations to PFAS exposure. Conclusions: ecological Our findings suggest potential effects of environmental PFAS exposure on thyroid cancer disparities. set up trials to validate these effects by sampling volunteers for blood-level PFAS measurement from neighborhoods with different levels of PFAS exposure (facility proximity and drinking water). Citation Format: Ming S. Lee, Laura Wild, Niko DiStefano, Garrett Forman, Elizabeth Franzmann, Natasha Solle, Alberto Caban-Martinez, Erin Kobetz. An ecological study on environmental exposure to PFAS and thyroid cancer incidence rates by races/ethnicities in Florida [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 2309.

Existing emotion-driven music generation models heavily rely on labeled data and lack interpretability and controllability of emotions. To address these limitations, a semi-supervised emotion-driven music generation model based on category-dispersed Gaussian mixture variational autoencoders is proposed. Initially, a controllable music generation model is introduced, which disentangles and manipulates rhythm and tonal features, enabling controlled music generation. Building on this, a semi-supervised model is developed, leveraging a category-dispersed Gaussian mixture variational autoencoder to infer emotions from the latent representations of rhythm and tonal features. Finally, the objective loss function is optimized to enhance the separation of distinct emotional clusters. Experimental results on real-world datasets demonstrate that the proposed method effectively separates music with different emotions in the latent space, thereby strengthening the association between music and emotions. Additionally, the model successfully disentangles and separates various musical features, facilitating more accurate emotion-driven music generation and emotion transitions through feature manipulation.

Thyroid cancer is the most common malignancy in Korea; in 2012, about 44,000 new cases (19.6% of all malignancies) were registered and the estimated age-standardized incidence rate of thyroid cancer was 73.6 per 100,000 (17.3 and 88.6 per 100,000 in men and women, respectively). Despite the steep increase in its incidence, the age-standardized mortality rate of thyroid cancer has remained stable and 10-year relative survival rate is 99.2%. Increased detection using high-resolution ultrasonography may have contributed to the increased incidence of thyroid cancer if not all. However, the effectiveness of thyroid cancer screening using ultrasonography has not been fully evaluated as to whether screening and early diagnosis could decrease the morbidity or mortality of thyroid cancer. A multidisciplinary expert committee for developing a guideline for thyroid cancer screening was organized and established a recommendation for thyroid cancer screening using ultrasonography in Korea based on scientific evidence for the first time. In conclusion, the current evidence is insufficient to assess the balance of benefits and harms of the thyroid cancer screening by ultrasonography and the recommendation is that thyroid ultrasonography is not routinely recommended for healthy subjects.

Nowadays, time series data is more and more likely to appear in various real-world systems, such as power plants, medical care, etc. In these systems, time series anomaly detection is necessary, which involves predictive maintenance, intrusion detection, anti-fraud, cloud platform monitoring and management, etc. Generally, the anomaly detection of time series is regarded as an unsupervised learning problem. However, in a real scenario, in addition to a large set of unlabeled data, there is usually a small set of available labeled data, such as normal or abnormal data sets labeled by experts. Only a few methods use labeled data, and the existing semi-supervised algorithms are not yet suitable for the field of time series anomaly detection. In this work, we propose a semi-supervised time series anomaly detection model based on LSTM autoencoder. We improve the loss function of the LSTM autoencoder so that it can be affected by unlabeled data and labeled data at the same time, and learn the distribution of unlabeled data and labeled data at the same time by minimizing the loss function. In a large number of experiments on the Yahoo! Webscope S5 and NAB data sets, we compared the performance of the unsupervised model and the semi-supervised model of the same network framework to prove that the performance of the semi-supervised model is improved compared to the unsupervised model.KeywordsTime seriesAnomaly detectionSemi-supervised learningAutoencoderLSTM

We live in the era of big data with dataset sizes growing steadily over the past decades. In addition, obtaining expert labels for all the instances is time-consuming and in many cases may not even be possible. This necessitates the development of advanced semi-supervised models that can learn from both labeled and unlabeled data points and also scale at worst linearly with the number of examples. In the context of kernel based semi-supervised models, constructing the training kernel matrix for the large training dataset is expensive and memory inefficient. This paper investigates the scalability of the recently proposed multi-class semi-supervised kernel spectral clustering model (MSSKSC) by means of random Fourier features. The proposed model maps the input data into an explicit low-dimensional feature space. Thanks to the explicit feature maps, one can then solve the MSSKSC optimization formation in the primal, making the complexity of the method linear in number of training data points. The performance of the proposed model is compared with that of recently introduced reduced kernel techniques and Nystrom based MSSKSC approaches. Experimental results demonstrate the scalability, efficiency and faster training computation times of the proposed model over conventional large scale semi-supervised models on large scale real-life datasets.

A Semi-Supervised Machine Learning Model for Risk Analysis of Urinary Tract Infections in People with Dementia

BackgroundUnderstanding the functional effects of non-coding variants is important as they are often associated with gene-expression alteration and disease development. Over the past few years, many computational tools have been developed to predict their functional impact. However, the intrinsic difficulty in dealing with the scarcity of data leads to the necessity to further improve the algorithms. In this work, we propose a novel method, employing a semi-supervised deep-learning model with pseudo labels, which takes advantage of learning from both experimentally annotated and unannotated data.ResultsWe prepared known functional non-coding variants with histone marks, DNA accessibility, and sequence context in GM12878, HepG2, and K562 cell lines. Applying our method to the dataset demonstrated its outstanding performance, compared with that of existing tools. Our results also indicated that the semi-supervised model with pseudo labels achieves higher predictive performance than the supervised model without pseudo labels. Interestingly, a model trained with the data in a certain cell line is unlikely to succeed in other cell lines, which implies the cell-type-specific nature of the non-coding variants. Remarkably, we found that DNA accessibility significantly contributes to the functional consequence of variants, which suggests the importance of open chromatin conformation prior to establishing the interaction of non-coding variants with gene regulation.ConclusionsThe semi-supervised deep learning model coupled with pseudo labeling has advantages in studying with limited datasets, which is not unusual in biology. Our study provides an effective approach in finding non-coding mutations potentially associated with various biological phenomena, including human diseases.

Thyroid cancer incidence has been increasing steadily in much of the developed world, but there is limited information on trends in the developing world. The overall rise in thyroid cancer incidence varies by gender, race/ethnicity, and geographic location, with the largest increases found among women and white non-Hispanics. Where histology is available, this increase has been accompanied by a change in the distribution of histological types, with a particular increase in papillary thyroid cancer. Sao Paulo, Brazil is the largest registry in Brazil and covers has one of the ten highest rates of thyroid cancer incidence worldwide, and has the highest rate among Brazilian registries. The Sao Paulo Registry covers about 15 million people in the region. Approximately 1000 new thyroid cancer cases are documented each year in Sao Paulo. This large number of cases presents the opportunity to evaluate trends in thyroid cancer incidence in a Latin American population. Using 12,495 cases reported between 1997 and 2005 to the Sao Paulo Cancer Registry, we analyzed thyroid cancer incidence patterns by gender and histology using Joinpoint analysis to calculate the annual percentage change (APC) in incidence, adjusting for age (standard world population). We found that females made up the vast majority of cases (87%) compared with males (13%). Papillary was the most common histology (74.7%), followed by follicular (17.3%), medullary (4%), and other (4%) among 77% of cases with histological confirmation. Overall incidence rates increased for both males and females, with a larger APC for females (+7.7%) compared with males (+6.6%). The greatest increases were seen for papillary and follicular histologies for both sexes. Rates increased most for females and males greater than 50 years of age compared with those 50 years or younger. Mortality rates remained relatively constant. These findings suggest that incidence rates of thyroid cancer are increasing in Sao Paulo, following a similar trend as seen in many developed countries. Further research looking at tumor stage and size could help to understand the potential causes of this increase. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr LB-390.

Understanding how news media frame political issues is important due to its impact on public attitudes, yet hard to automate. Computational approaches have largely focused on classifying the frame of a full news article while framing signals are often subtle and local. Furthermore, automatic news analysis is a sensitive domain, and existing classifiers lack transparency in their predictions. This paper addresses both issues with a novel semi-supervised model, which jointly learns to embed local information about the events and related actors in a news article through an auto-encoding framework, and to leverage this signal for document-level frame classification. Our experiments show that: our model outperforms previous models of frame prediction; we can further improve performance with unlabeled training data leveraging the semi-supervised nature of our model; and the learnt event and actor embeddings intuitively corroborate the document-level predictions, providing a nuanced and interpretable article frame representation.

Label consistency-based deep semisupervised NMF for tumor recognition

Regarding the healthcare delivered to patients with euthyroid goiter in general practice, and on the basis of our own research experience in developing a guideline on thyroid disorders in general practice for the German College for General Practitioners and Family Physicians (Deutsche Gesellschaft fur Allgemeinmedizin, DEGAM), we wish to draw attention to the following. We have noticed with some surprise that, for some years now, authors of German scientific articles have been citing the Papillon study when reporting the prevalence rates of goiter and thyroid nodules for the German population. What remains completely unconsidered is the question of the relevance of these prevalence values for the German population. When looking at the selection of the study participants, for example (voluntary participation of “people in work,” who themselves (!) reported no prior thyroid-related findings), one may assume a high selection bias. These prevalence rates are not representative for the German population. Most cases of goiter are asymptomatic. Do all persons with asymptomatic changes of the thyroid (mostly incidental findings) have to be referred for further diagnostic/therapeutic measures? Or is watchful waiting sufficient, provided the patients have received comprehensive information? It is worth mentioning that thyroid cancer is rare. Estimated, age-standardized incidence rates of thyroid cancer according to the Robert Koch-Institute are 3.1/100 000 in men and 6.3/100 000 in women (1). Most thyroid changes are of benign origin (23). Against the background of very low incidence rates, the question arises of how effective further diagnostic measures are, especially in cases of asymptomatic euthyroid goiter. Their predictive value is too low due to the very low prevalence rates, especially in general practice. On the basis of the current state of knowledge, the DEGAM guideline for thyroid disorders will appeal to general practitioners to adopt an approach of watchful waiting in patients with asymptomatic thyroid changes (always depending on the clinical symptoms, in accordance with patients’ own wishes, and provided comprehensive information has been given).

Thyroid cancer has a high prevalence all over the world. Accurate thyroid nodule diagnosis can lead to effective treatment and decrease the mortality rate. Ultrasound imaging is a safe, portable, and inexpensive tool for thyroid nodule monitoring. However, the widespread use of ultrasound has also resulted in over-diagnosis and over-treatment of nodules. There is also large variability in the assessment and characterization of nodules. Thyroid nodule classification requires precise delineation of the nodule boundary which is tedious and time- consuming. Automatic segmentation of nodule boundaries is highly desirable, however, it is challenging due to the wide range of nodule appearances, shapes, and sizes. In this study, we propose an end-to-end pipeline for nodule segmentation and classification. A residual dilated UNet (resDUnet) model is proposed for nodule segmentation. The output of resDUnet is fed to two rule-based classifiers to categorize the composition and echogenicity of the segmented nodule. We evaluate our segmentation method on a large dataset of 352 ultrasound images reviewed by a certified radiologist. When compared with ground-truth, resDUnet gives a higher Dice score than the standard UNet (82% vs. 81%). Our method requires minimal user interaction and it is robust to reasonable variations in the user-specified region-of-interest. We expect the proposed method to reduce variability in thyroid nodule assessment which results in more efficient and cost-effective monitoring of thyroid cancer.

Background Incidence rate of thyroid cancer is steadily increasing due to overdiagnosis and overtreatment. Thyroid ultrasound is commonly used to diagnose thyroid cancer. The aim of this study is to examine the accuracy of using deep convolutional neural network (DCNN) models to improve diagnosis of thyroid cancer by analyzing sonographic imaging data from clinical thyroid ultrasound. Methods A total of 131,731 sonographic images from 17,627 thyroid cancer patients and 180,668 sonographic images from 25,325 controls used as training set were obtained from Tianjin Cancer Hospital. Images from anatomical sites that did not have cancer according to location sign on the image were not included. All thyroid cancer patients and 13·2% of controls (51,255 images) were confirmed by pathological reports. DCNN is a specific type of neural network optimized for image recognition. We trained two DCNN models on the training set and subsequently evaluated the performance on one independent internal (Tianjin, 1,118 individuals) and two external (Jilin,154 individuals; Weihai, 1,420 individuals) validation sets. Individuals in the validation sets all have pathological examinations. We compared the specificity/sensitivity of DCNN models with the performance of six thyroid ultrasound radiologists on these three validation sets. Findings DCNN model achieved high performance in identifying thyroid cancer patients versus six experience radiologists: for Tianjin validation set, sensitivity was 92·2% versus 96·9% (95% CI 89·7% - 94·3% vs. 93·9% - 98·6%; p = 0·003), and specificity was 85·6% versus 59·4% (95% CI 82·4% - 88·4% vs. 53% - 65·6%; p < 0·0001); for Jilin validation set, sensitivity was 84·3% versus 92·9% (95% CI 73·6% - 91·9% vs. 84·1% - 97·6%; p = 0·05), and specificity was 86·9% versus 57·1% (95% CI 77·8% - 93·3% vs. 45·9% - 67·9%; p < 0·0001); for Weihai validation set, sensitivity was 84·5% versus 89% (95% CI 81·2% - 87·4% vs. 81·9% - 94%; p = 0·2), and specificity was 87·5% versus 68·6% (95% CI 85·1% - 89·6% vs. 60·7% - 75·8%; p < 0·0001). Interpretation DCNN models exhibited high accuracy, sensitivity, and specificity in identifying thyroid cancer patients at levels comparable to or higher than six experienced radiologists. Conferred by the high specificity of DCNN models, the rate of overdiagnosis and overtreatment of patients with thyroid cancer is expected to decrease. This supports future application of the deep learning models to clinical practice for thyroid cancer diagnosis. However, further validation of these DCNN models in prospective clinical trials is warranted. Funding The Program for Changjiang Scholars and Innovative Research Team in University in China (IRT_14R40), National Natural Science Foundation of China (31801117). Citation Format: Xiangchun Li, Sheng Zhang, Qiang Zhang, Xi Wei, Yi Pan, Jing Zhao, Xiaojie Xin, Xiaoqing Wang, Fan Yang, Jianxin Li, Meng Yang, Qinghua Wang, Xiangqian Zheng, Yanhui Zhao, Lun Zhang, Xudong Wang, Zhimin Zheng, Christopher T. Whitlow, Metin N. Gurcan, Boris Pasche, Ming Gao, Wei Zhang, Kexin Chen. Diagnosis of thyroid cancer using deep convolutional neural network models applied to sonographic images from clinical ultrasound exams [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 1394.

This research proposes an image segmentation model based on an improved U-Net network (rcU-Net) in response to the phenomena of misinterpretation and missed diagnosis in the process of artificial diagnosis and screening due to the variability of thyroid nodule size and unclear edges in ultrasound images. This paper uses the TN3K dataset as the experimental dataset. The superiority of the proposed model is validated through comparative experiments and ablation experiments. Experimental results show that the proposed model achieves an accuracy of 95.61%, an AUC of 90.67%, a specificity of 98.12%, and a Dice coefficient of 75.69% on the thyroid ultrasound image dataset. The deep learning model proposed in this paper performs well in the segmentation task of thyroid nodule ultrasound images, providing new solutions for the identification of small nodules, nodule edge segmentation, and noise interference in thyroid nodule segmentation tasks.

The ultrasound image is a commonly used imaging modality to treat thyroid nodules due to its rapid imaging speed and the ability for multiple anatomical and soft-tissue visualization. However, considering the variability of the position, shape, size and intensity of nodules, the accurate segmentation of thyroid nodules is challenging. In this paper, we proposed an automatic segmentation network of thyroid nodules by combining fast-RCNN and spatial-channel attentive U-Net. In our model, fast-RCNN is firstly utilized to identify the rough position of nodules. The predicted boxes are then utilized to crop the image patches containing only nodules and then build a new database. Next, a spatial-channel attentive U-Net is designed and trained to realize the nodule segmentation. By comparing the proposed network with other state-of-the-art models, we can gain superior results in discarding similar non-nodule structures and preserving the boundary information of nodules.