Diagnosis of Salivary Gland Tumors Using Transfer Learning with Fine-Tuning and Gradual Unfreezing.

Ultrasound is the primary tool for evaluating salivary gland tumors (SGTs); however, tumor diagnosis currently relies on subjective features. This study aimed to establish an objective ultrasound diagnostic method using deep learning. We collected 446 benign and 223 malignant SGT ultrasound images in the training/validation set and 119 benign and 44 malignant SGT ultrasound images in the testing set. We trained convolutional neural network (CNN) models from scratch and employed transfer learning (TL) with fine-tuning and gradual unfreezing to classify malignant and benign SGTs. The diagnostic performances of these models were compared. By utilizing the pretrained ResNet50V2 with fine-tuning and gradual unfreezing, we achieved a 5-fold average validation accuracy of 0.920. The diagnostic performance on the testing set demonstrated an accuracy of 89.0%, a sensitivity of 81.8%, a specificity of 91.6%, a positive predictive value of 78.3%, and a negative predictive value of 93.2%. This performance surpasses that of other models in our study. The corresponding Grad-CAM visualizations were also presented to provide explanations for the diagnosis. This study presents an effective and objective ultrasound method for distinguishing between malignant and benign SGTs, which could assist in preoperative evaluation.

Macrophage-Hitchhiked Orally Administered ß-Glucans-Functionalized Nanoparticles as "Precision-Guided Stealth Missiles" for Targeted Pancreatic Cancer Therapy.

The prognosis in cases of pancreatic ductal adenocarcinoma (PDAC) with current treatment modalities is poor owing to the highly desmoplastic tumor microenvironment (TME). Herein, a ß-glucans-functionalized zinc-doxorubicin nanoparticle system (ßGlus-ZnD NPs) that can be orally administered, is developed for targeted PDAC therapy. Following oral administration in PDAC-bearing mice, ßGlus-ZnD NPs actively target/transpass microfold cells, overcome the intestinal epithelial barrier, and then undergo subsequent phagocytosis by endogenous macrophages (ßGlus-ZnD@Mϕ). As hitchhiking cellular vehicles, ßGlus-ZnD@Mϕ transits through the intestinal lymphatic system and enters systemic circulation, ultimately accumulating in the tumor tissue as a result of the tumor-homing and "stealth" properties that are conferred by endogenous Mϕ. Meanwhile, the Mϕ that hitchhikes ßGlus-ZnD NPs is activated to produce matrix metalloproteinases, destroying the desmoplastic stromal barrier, and differentiates toward the M1 -like phenotype, modulating the TME and recruiting effector T cells, ultimately inducing apoptosis of the tumor cells. The combination of ßGlus-ZnD@Mϕ and immune checkpoint blockade effectively inhibits the growth of the primary tumor and suppresses the development of metastasis. It thus represents an appealing approach to targeted PDAC therapy.

Fire safety study on high-flow nasal oxygen in shared-airway surgeries with diathermy and laser: simulation based on a physical model.

High-flow nasal oxygen (HFNO) has been used in "tubeless" shared-airway surgeries but whether HFNO increased the fire hazard is yet to be examined. We used a physical model for simulation to explore fire safety through a series of ignition trials. An HFNO device was attached to a 3D-printed nose with nostrils connected to a degutted raw chicken. The HFNO device was set at twenty combinations of different oxygen concentration and gas flow rate. An electrocautery and diode laser were applied separately to a fat cube in the cavity of the chicken. Ten 30 s trials of continuous energy source application were conducted. An additional trial of continuous energy application was conducted if no ignition was observed for all the ten trials. A total of eight short flashes were observed in one hundred electrocautery tests; however, no continuous fire was observed among them. There were thirty-six events of ignition in one hundred trials with laser, twelve of which turned into violent self-sustained fires. The factors found to be related to a significantly increased chance of ignition included laser application, lower gas flow, and higher FiO2. The native tissue and smoke can ignite and turn into violent self-sustained fires under HFNO and continuous laser strikes, even in the absence of combustible materials. The results suggest that airway surgeries must be performed safely with HFNO if only a short intermittent laser is used in low FiO2.

Risk stratification of heart failure from one drop of blood using hand-held biosensor for BNP detection.

Continued risk assessment by evaluating cardiac biomarkers in healthy and unhealthy individuals can lower the mortality rate of cardiovascular diseases (CVDs). In this research, we have developed a hand-held biosensor system to rapidly screen for brain natriuretic peptide (BNP) from a single drop of whole blood. The sensor methodology is based on extended gate design of electrical double layer (EDL) field effect transistor (FET), that can directly detect BNP in whole blood, without extensive sample pre-treatments, thereby eliminating the limitations of charge screening in high ionic strength solutions. A simple sensor array chip is fabricated to integrate with the MOSFET sensor system. Sensing characteristics are elucidated using purified BNP samples in 1 × PBS (with 4% BSA), spiked BNP samples in whole blood and clinical whole blood samples. The blood cells can be gravitationally separated without the use of any external actuation. The sensor exhibits very high sensitivity over wide dynamic range of detection. The sensing characteristics are not adversely affected by the presence of background proteins or blood cells, even without gravitational blood cell separation. Thus, the biosensor system can allow users to perform rapid whole blood diagnostics with minimal user protocols, in 5 min. The features of high sensitivity, cost-effectiveness and convenience of usage empower this technology to revolutionize the mobile diagnostics and healthcare industry.

Single Drop Whole Blood Diagnostics: Portable Biomedical Sensor for Cardiac Troponin I Detection.

Detection of disease biomarkers from whole blood is very important in disease prevention and management. However, new generation assays like point-of-care or mobile diagnostics face a myriad of challenges in detecting proteins from whole blood. In this research, we have designed, fabricated, and characterized a portable biomedical sensor for the detection of cardiac troponin I (cTnI) directly from whole blood, without sample pretreatments. The sensing methodology is based on an extended gate electrical double layer (EDL) gated field effect transistor (FET) biosensor that can offer very high sensitivity, a wide dynamic range, and high selectivity to target analyte. The sensing methodology is not impeded by electrostatic screening and can be applied to all types of FET sensors. A portable biomedical system is designed to carry out the diagnostic assay in a very simple and rapid manner, that allows the user to screen for target protein from a single drop of blood, in 5 min. This biomedical sensor can be used in hospitals and homes alike, for early detection of cTnI which is a clinical marker for acute myocardial infarction. This sensing methodology could potentially revolutionize the modern health care industry.

Detection of osteogenic differentiation by differential mineralized matrix production in mesenchymal stromal cells by Raman spectroscopy.

Mesenchymal stromal cells (MSCs) hold great potential in skeletal tissue engineering and regenerative medicine. However, conventional methods that are used in molecular biology to evaluate osteogenic differentiation of MSCs require a relatively large amount of cells. Cell lysis and cell fixation are also required and all these steps are time-consuming. Therefore, it is imperative to develop a facile technique which can provide real-time information with high sensitivity and selectivity to detect the osteogenic maturation of MSCs. In this study, we use Raman spectroscopy as a biosensor to monitor the production of mineralized matrices during osteogenic induction of MSCs. In summary, Raman spectroscopy is an excellent biosensor to detect the extent of maturation level during MSCs-osteoblast differentiation with a non-disruptive, real-time and label free manner. We expect that this study will promote further investigation of stem cell research and clinical applications.