Relay learning: a physically secure framework for clinical multi-site deep learning.

Big data serves as the cornerstone for constructing real-world deep learning systems across various domains. In medicine and healthcare, a single clinical site lacks sufficient data, thus necessitating the involvement of multiple sites. Unfortunately, concerns regarding data security and privacy hinder the sharing and reuse of data across sites. Existing approaches to multi-site clinical learning heavily depend on the security of the network firewall and system implementation. To address this issue, we propose Relay Learning, a secure deep-learning framework that physically isolates clinical data from external intruders while still leveraging the benefits of multi-site big data. We demonstrate the efficacy of Relay Learning in three medical tasks of different diseases and anatomical structures, including structure segmentation of retina fundus, mediastinum tumors diagnosis, and brain midline localization. We evaluate Relay Learning by comparing its performance to alternative solutions through multi-site validation and external validation. Incorporating a total of 41,038 medical images from 21 medical hosts, including 7 external hosts, with non-uniform distributions, we observe significant performance improvements with Relay Learning across all three tasks. Specifically, it achieves an average performance increase of 44.4%, 24.2%, and 36.7% for retinal fundus segmentation, mediastinum tumor diagnosis, and brain midline localization, respectively. Remarkably, Relay Learning even outperforms central learning on external test sets. In the meanwhile, Relay Learning keeps data sovereignty locally without cross-site network connections. We anticipate that Relay Learning will revolutionize clinical multi-site collaboration and reshape the landscape of healthcare in the future.

Miniaturized microfluidic-based nucleic acid analyzer to identify new biomarkers of biopsy lung cancer samples for subtyping.

Identifying new biomarkers is necessary and important to diagnose and treat malignant lung cancer. However, existing protein marker detection methods usually require complex operation steps, leading to a lag time for diagnosis. Herein, we developed a rapid, minimally invasive, and convenient nucleic acid biomarker recognition method, which enabled the combined specific detection of 11 lung cancer typing markers in a microliter reaction system after only one sampling. The primers for the combined specific detection of 11 lung cancer typing markers were designed and screened, and the microfluidic chip for parallel detection of the multiple markers was designed and developed. Furthermore, a miniaturized microfluidic-based analyzer was also constructed. By developing a microfluidic chip and a miniaturized nucleic acid analyzer, we enabled the detection of the mRNA expression levels of multiple biomarkers in rice-sized tissue samples. The miniaturized nucleic acid analyzer could detect ≥10 copies of nucleic acids. The cell volume of the typing reaction on the microfluidic chip was only 0.94 µL, less than 1/25 of that of the conventional 25-µL Eppendorf tube PCR method, which significantly reduced the testing cost and significantly simplified the analysis of multiple biomarkers in parallel. With a simple injection operation and reverse transcription loop-mediated isothermal amplification (RT-LAMP), real-time detection of 11 lung cancer nucleic acid biomarkers was performed within 45 min. Given these compelling features, 86 clinical samples were tested using the miniaturized nucleic acid analyzer and classified according to the cutoff values of the 11 biomarkers. Furthermore, multi-biomarker analysis was conducted by a machine learning model to classify different subtypes of lung cancer, with an average area under the curve (AUC) of 0.934. This method shows great potential for the identification of new nucleic acid biomarkers and the accurate diagnosis of lung cancer.

Toward human intervention-free clinical diagnosis of intracranial aneurysm via deep neural network.

Intracranial aneurysm (IA) is an enormous threat to human health, which often results in nontraumatic subarachnoid hemorrhage or dismal prognosis. Diagnosing IAs on commonly used computed tomographic angiography (CTA) examinations remains laborious and time consuming, leading to error-prone results in clinical practice, especially for small targets. In this study, we propose a fully automatic deep-learning model for IA segmentation that can be applied to CTA images. Our model, called Global Localization-based IA Network (GLIA-Net), can incorporate the global localization prior and generates the fine-grain three-dimensional segmentation. GLIA-Net is trained and evaluated on a big internal dataset (1,338 scans from six institutions) and two external datasets. Evaluations show that our model exhibits good tolerance to different settings and achieves superior performance to other models. A clinical experiment further demonstrates the clinical utility of our technique, which helps radiologists in the diagnosis of IAs.