Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

Surgery Surgery

Using a New Deep Learning Method for 3D Cephalometry in Patients With Cleft Lip and Palate.

In The Journal of craniofacial surgery

Deep learning algorithms based on automatic 3-dimensional (D) cephalometric marking points about people without craniomaxillofacial deformities has achieved good results. However, there has been no previous report about cleft lip and palate. The purpose of this study is to apply a new deep learning method based on a 3D point cloud graph convolutional neural network to predict and locate landmarks in patients with cleft lip and palate based on the relationships between points. The authors used the PointNet++ model to investigate the automatic 3D cephalometric marking points. And the mean distance error of the center coordinate position and the success detection rate (SDR) were used to evaluate the accuracy of systematic labeling. A total of 150 patients were enrolled. The mean distance error for all 27 landmarks was 1.33 mm, and 9 landmarks (30%) showed SDRs at 2 mm over 90%, and 3 landmarks (35%) showed SDRs at 2 mm under 70%. The automatic 3D cephalometric marking points take 16 seconds per dataset. In summary, our training sets were derived from the cleft lip with/without palate computed tomography to achieve accurate results. The 3D cephalometry system based on the graph convolutional neural network algorithm may be suitable for 3D cephalometry system in cleft lip and palate cases. More accurate results may be obtained if the cleft lip and palate training set is expanded in the future.

Xu Meng, Liu Bingyang, Luo Zhaoyang, Ma Hengyuan, Sun Min, Wang Yongqian, Yin Ningbei, Tang Xiaojun, Song Tao

2023-Mar-22

Radiology Radiology

Impact of inactivated COVID-19 vaccines on lung injury in B.1.617.2 (Delta) variant-infected patients.

In Annals of clinical microbiology and antimicrobials

BACKGROUND : Chest computerized tomography (CT) scan is an important strategy that quantifies the severity of COVID-19 pneumonia. To what extent inactivated COVID-19 vaccines could impact the COVID-19 pneumonia on chest CT is not clear.

METHODS : This study recruited 357 SARS-COV-2 B.1.617.2 (Delta) variant-infected patients admitted to the Second Hospital of Nanjing from July to August 2021. An artificial intelligence-assisted CT imaging system was used to quantify the severity of COVID-19 pneumonia. We compared the volume of infection (VOI), percentage of infection (POI) and chest CT scores among patients with different vaccination statuses.

RESULTS : Of the 357 Delta variant-infected patients included for analysis, 105 were unvaccinated, 72 were partially vaccinated and 180 were fully vaccinated. Fully vaccination had the least lung injuries when quantified by VOI (median VOI of 222.4 cm3, 126.6 cm3 and 39.9 cm3 in unvaccinated, partially vaccinated and fully vaccinated, respectively; p < 0.001), POI (median POI of 7.60%, 3.55% and 1.20% in unvaccinated, partially vaccinated and fully vaccinated, respectively; p < 0.001) and chest CT scores (median CT score of 8.00, 6.00 and 4.00 in unvaccinated, partially vaccinated and fully vaccinated, respectively; p < 0.001). After adjustment for age, sex, comorbidity, time from illness onset to hospitalization and viral load, fully vaccination but not partial vaccination was significantly associated with less lung injuries quantified by VOI {adjust coefficient[95%CI] for "full vaccination": - 106.10(- 167.30,44.89); p < 0.001}, POI {adjust coefficient[95%CI] for "full vaccination": - 3.88(- 5.96, - 1.79); p = 0.001} and chest CT scores {adjust coefficient[95%CI] for "full vaccination": - 1.81(- 2.72, - 0.91); p < 0.001}. The extent of reduction of pulmonary injuries was more profound in fully vaccinated patients with older age, having underlying diseases, and being female sex, as demonstrated by relatively larger absolute values of adjusted coefficients. Finally, even within the non-severe COVID-19 population, fully vaccinated patients were found to have less lung injuries.

CONCLUSION : Fully vaccination but not partially vaccination could significantly protect lung injury manifested on chest CT. Our study provides additional evidence to encourage a full course of vaccination.

Lai Miao, Wang Kai, Ding Chengyuan, Yin Yi, Lin Xiaoling, Xu Chuanjun, Hu Zhiliang, Peng Zhihang

2023-Mar-21

Artificial intelligence (AI), COVID-19, COVID-19 vaccines, Chest CT, Lung injury

Radiology Radiology

Compositional Zero-Shot Domain Transfer with Text-to-Text Models

ArXiv Preprint

Label scarcity is a bottleneck for improving task performance in specialised domains. We propose a novel compositional transfer learning framework (DoT5 - domain compositional zero-shot T5) for zero-shot domain transfer. Without access to in-domain labels, DoT5 jointly learns domain knowledge (from MLM of unlabelled in-domain free text) and task knowledge (from task training on more readily available general-domain data) in a multi-task manner. To improve the transferability of task training, we design a strategy named NLGU: we simultaneously train NLG for in-domain label-to-data generation which enables data augmentation for self-finetuning and NLU for label prediction. We evaluate DoT5 on the biomedical domain and the resource-lean subdomain of radiology, focusing on NLI, text summarisation and embedding learning. DoT5 demonstrates the effectiveness of compositional transfer learning through multi-task learning. In particular, DoT5 outperforms the current SOTA in zero-shot transfer by over 7 absolute points in accuracy on RadNLI. We validate DoT5 with ablations and a case study demonstrating its ability to solve challenging NLI examples requiring in-domain expertise.

Fangyu Liu, Qianchu Liu, Shruthi Bannur, Fernando Pérez-García, Naoto Usuyama, Sheng Zhang, Tristan Naumann, Aditya Nori, Hoifung Poon, Javier Alvarez-Valle, Ozan Oktay, Stephanie L. Hyland

2023-03-23

General General

Fault Prognosis of Turbofan Engines: Eventual Failure Prediction and Remaining Useful Life Estimation

ArXiv Preprint

In the era of industrial big data, prognostics and health management is essential to improve the prediction of future failures to minimize inventory, maintenance, and human costs. Used for the 2021 PHM Data Challenge, the new Commercial Modular Aero-Propulsion System Simulation dataset from NASA is an open-source benchmark containing simulated turbofan engine units flown under realistic flight conditions. Deep learning approaches implemented previously for this application attempt to predict the remaining useful life of the engine units, but have not utilized labeled failure mode information, impeding practical usage and explainability. To address these limitations, a new prognostics approach is formulated with a customized loss function to simultaneously predict the current health state, the eventual failing component(s), and the remaining useful life. The proposed method incorporates principal component analysis to orthogonalize statistical time-domain features, which are inputs into supervised regressors such as random forests, extreme random forests, XGBoost, and artificial neural networks. The highest performing algorithm, ANN-Flux, achieves AUROC and AUPR scores exceeding 0.95 for each classification. In addition, ANN-Flux reduces the remaining useful life RMSE by 38% for the same test split of the dataset compared to past work, with significantly less computational cost.

Joseph Cohen, Xun Huan, Jun Ni

2023-03-23

Radiology Radiology

Mesoporous nanodrug delivery system: a powerful tool for a new paradigm of remodeling of the tumor microenvironment.

In Journal of nanobiotechnology

Tumor microenvironment (TME) plays an important role in tumor progression, metastasis and therapy resistance. Remodeling the TME has recently been deemed an attractive tumor therapeutic strategy. Due to its complexity and heterogeneity, remodeling the TME still faces great challenges. With the great advantage of drug loading ability, tumor accumulation, multifactor controllability, and persistent guest molecule release ability, mesoporous nanodrug delivery systems (MNDDSs) have been widely used as effective antitumor drug delivery tools as well as remolding TME. This review summarizes the components and characteristics of the TME, as well as the crosstalk between the TME and cancer cells and focuses on the important role of drug delivery strategies based on MNDDSs in targeted remodeling TME metabolic and synergistic anticancer therapy.

Hang Yinhui, Liu Yanfang, Teng Zhaogang, Cao Xiongfeng, Zhu Haitao

2023-Mar-21

Mesoporous nanodrug delivery systems, Metabolism, Targeted remodeling, Tumor microenvironment, Tumor therapy

Surgery Surgery

Radiomics nomogram for preoperative differentiation of pulmonary mucinous adenocarcinoma from tuberculoma in solitary pulmonary solid nodules.

In BMC cancer

OBJECTIVE : To develop and validate predictive models using clinical parameters, radiomic features and a combination of both for preoperative differentiation of pulmonary nodular mucinous adenocarcinoma (PNMA) from pulmonary tuberculoma (PTB).

METHOD : A total of 124 and 53 patients with PNMA and PTB, respectively, were retrospectively analyzed from January 2017 to November 2022 in The Fourth Affiliated Hospital of Hebei Medical University (Ligang et al., A machine learning model based on CT and clinical features to distinguish pulmonary nodular mucinous adenocarcinoma from tuberculoma, 2023). A total of 1037 radiomic features were extracted from contrast-enhanced computed tomography (CT). The patients were randomly divided into a training group and a test group at a ratio of 7:3. The least absolute shrinkage and selection operator (LASSO) algorithm was used for radiomic feature selection. Three radiomics prediction models were applied: logistic regression (LR), support vector machine (SVM) and random forest (RF). The best performing model was adopted, and the radiomics score (Radscore) was then computed. The clinical model was developed using logistic regression. Finally, a combined model was established based on clinical factors and radiomics features. We externally validated the three models in a group of 68 patients (46 and 22 patients with PNMA and PTB, respectively) from Xing Tai People's Hospital (30 and 14 patients with PNMA and PTB, respectively) and The First Hospital of Xing Tai (16 and 8 patients with PNMA and PTB, respectively). The area under the receiver operating characteristic (ROC) curve (AUC) value and decision curve analysis were used to evaluate the predictive value of the developed models.

RESULTS : The combined model established by the logistic regression method had the best performance. The ROC-AUC (also a decision curve analysis) of the combined model was 0.940, 0.990 and 0.960 in the training group, test group and external validation group, respectively, and the combined model showed good predictive performance for the differentiation of PNMA from PTB. The Brier scores of the combined model were 0.132 and 0.068 in the training group and test group, respectively.

CONCLUSION : The combined model incorporating radiomics features and clinical parameters may have potential value for the preoperative differentiation of PNMA from PTB.

Zhang Junjie, Hao Ligang, Qi MingWei, Xu Qian, Zhang Ning, Feng Hui, Shi Gaofeng

2023-Mar-21

Contrast-enhanced CT, Mucinous adenocarcinoma, Nomogram, Radiomics, Tuberculoma