Automatic Acne Object Detection and Acne Severity Grading Using Smartphone Images and Artificial Intelligence.

Skin image analysis using artificial intelligence (AI) has recently attracted significant research interest, particularly for analyzing skin images captured by mobile devices. Acne is one of the most common skin conditions with profound effects in severe cases. In this study, we developed an AI system called AcneDet for automatic acne object detection and acne severity grading using facial images captured by smartphones. AcneDet includes two models for two tasks: (1) a Faster R-CNN-based deep learning model for the detection of acne lesion objects of four types, including blackheads/whiteheads, papules/pustules, nodules/cysts, and acne scars; and (2) a LightGBM machine learning model for grading acne severity using the Investigator's Global Assessment (IGA) scale. The output of the Faster R-CNN model, i.e., the counts of each acne type, were used as input for the LightGBM model for acne severity grading. A dataset consisting of 1572 labeled facial images captured by both iOS and Android smartphones was used for training. The results show that the Faster R-CNN model achieves a mAP of 0.54 for acne object detection. The mean accuracy of acne severity grading by the LightGBM model is 0.85. With this study, we hope to contribute to the development of artificial intelligent systems to help acne patients better understand their conditions and support doctors in acne diagnosis.

A Portable Droplet Magnetofluidic Device for Point-of-Care Detection of Multidrug-Resistant Candida auris.

Candida auris is an emerging multidrug-resistant fungal pathogen that can cause severe and deadly infections. To date, C. auris has spurred outbreaks in healthcare settings in thirty-three countries across five continents. To control and potentially prevent its spread, there is an urgent need for point-of-care (POC) diagnostics that can rapidly screen patients, close patient contacts, and surveil environmental sources. Droplet magnetofluidics (DM), which leverages nucleic acid-binding magnetic beads for realizing POC-amenable nucleic acid detection platforms, offers a promising solution. Herein, we report the first DM device-coined POC.auris-for POC detection of C. auris. As part of POC.auris, we have incorporated a handheld cell lysis module that lyses C. auris cells with 2 min hands-on time. Subsequently, within the palm-sized and automated DM device, C. auris and control DNA are magnetically extracted and purified by a motorized magnetic arm and finally amplified via a duplex real-time quantitative PCR assay by a miniaturized rapid PCR module and a miniaturized fluorescence detector-all in ≤30 min. For demonstration, we use POC.auris to detect C. auris isolates from 3 major clades, with no cross reactivity against other Candida species and a limit of detection of ∼300 colony forming units per mL. Taken together, POC.auris presents a potentially useful tool for combating C. auris.

Filtration-assisted magnetofluidic cartridge platform for HIV RNA detection from blood.

The ability to detect and quantify HIV RNA in blood is essential to sensitive detection of infections and monitoring viremia throughout treatment. Current options for point-of-care HIV diagnosis (i.e. lateral flow rapid tests) lack sensitivity for early detection and are unable to quantify viral load. HIV RNA diagnostics typically require extensive pre-processing of blood to isolate plasma and extract nucleic acids, in addition to expensive equipment for conducting nucleic acid amplification and fluorescence detection. Therefore, molecular HIV diagnostics is still mainly limited to clinical laboratories and there is an unmet need for high sensitivity point-of-care screening and at-home HIV viral load quantification. In this work, we outline a streamlined workflow for extraction of plasma from whole blood coupled with HIV RNA extraction and quantitative polymerase chain reaction (qPCR) in a portable magnetofluidic cartridge platform for use at the point-of-care. Viral particles were isolated from blood using manual filtration through a 3D-printed filter module in seconds followed by automated nucleic acid capture, purification, and transfer to qPCR using magnetic beads. Both nucleic acid extraction and qPCR were integrated within cartridges using compact instrumentation consisting of a motorized magnet arm, miniaturized thermocycler, and image-based fluorescence detection. We demonstrated detection down to 1000 copies of HIV viral particles from whole blood in <30 minutes.

Ensemble learning using traditional machine learning and deep neural network for diagnosis of Alzheimer's disease.

In recent years, Alzheimer's disease (AD) diagnosis using neuroimaging and deep learning has drawn great research attention. However, due to the scarcity of training neuroimaging data, many deep learning models have suffered from severe overfitting. In this study, we propose an ensemble learning framework that combines deep learning and machine learning. The deep learning model was based on a 3D-ResNet to exploit 3D structural features of neuroimaging data. Meanwhile, Extreme Gradient Boosting (XGBoost) machine learning was applied on a voxel-wise basis to draw the most significant voxel groups out of the image. The 3D-ResNet and XGBoost predictions were combined with patient demographics and cognitive test scores (Mini-Mental State Examination (MMSE) and Clinical Dementia Rating (CDR)) to give a final diagnosis prediction. Our proposed method was trained and validated on brain MRI brain images of the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. During the training phase, multiple data augmentation methods were employed to tackle overfitting. Our test set contained only baseline scans, i.e., the first visit scans since we aimed to investigate the ability of our approach in detecting AD during the first visit of AD patients. Our 5-fold cross-validation implementation achieved an average AUC of 100% during training and 96% during testing. Using the same computer, our method was much faster in scoring a prediction, approximately 10 min, than feature extraction-based machine learning methods, which often take many hours to score a prediction. To make the prediction explainable, we visualized the brain MRI image regions that primarily affected the 3D-ResNet model's prediction via heatmap. Lastly, we observed that proper generation of test sets was critical to avoiding the data leakage issue and ensuring the validity of results.

Label-free DNA biosensor based on SERS Molecular Sentinel on Nanowave chip.

Development of a rapid, cost-effective, label-free biosensor for DNA detection is important for many applications in clinical diagnosis, homeland defense, and environment monitoring. A unique label-free DNA biosensor based on Molecular Sentinel (MS) immobilized on a plasmonic 'Nanowave' chip, which is also referred to as a metal film over nanosphere (MFON), is presented. Its sensing mechanism is based upon the decrease of the surface-enhanced Raman scattering (SERS) intensity when Raman label tagged at one end of MS is physically separated from the MFON's surface upon DNA hybridization. This method is label-free as the target does not have to be labeled. The MFON fabrication is relatively simple and low-cost with high reproducibility based on depositing a thin shell of gold over close-packed arrays of nanospheres. The sensing process involves a single hybridization step between the DNA target sequences and the complementary MS probes on the Nanowave chip without requiring secondary hybridization or posthybridization washing, thus resulting in rapid assay time and low reagent usage. The usefulness and potential application of the biosensor for medical diagnostics is demonstrated by detecting the human radical S-adenosyl methionine domain containing 2 (RSAD2) gene, a common inflammation biomarker.