Virtual staining for pixel-wise and quantitative analysis of single cell images.

Immunocytochemical staining of microorganisms and cells has long been a popular method for examining their specific subcellular structures in greater detail. Recently, generative networks have emerged as an alternative to traditional immunostaining techniques. These networks infer fluorescence signatures from various imaging modalities and then virtually apply staining to the images in a digital environment. In numerous studies, virtual staining models have been trained on histopathology slides or intricate subcellular structures to enhance their accuracy and applicability. Despite the advancements in virtual staining technology, utilizing this method for quantitative analysis of microscopic images still poses a significant challenge. To address this issue, we propose a straightforward and automated approach for pixel-wise image-to-image translation. Our primary objective in this research is to leverage advanced virtual staining techniques to accurately measure the DNA fragmentation index in unstained sperm images. This not only offers a non-invasive approach to gauging sperm quality, but also paves the way for streamlined and efficient analyses without the constraints and potential biases introduced by traditional staining processes. This novel approach takes into account the limitations of conventional techniques and incorporates improvements to bolster the reliability of the virtual staining process. To further refine the results, we discuss various denoising techniques that can be employed to reduce the impact of background noise on the digital images. Additionally, we present a pixel-wise image matching algorithm designed to minimize the error caused by background noise and to prevent the introduction of bias into the analysis. By combining these approaches, we aim to develop a more effective and reliable method for quantitative analysis of virtually stained microscopic images, ultimately enhancing the study of microorganisms and cells at the subcellular level.

On the application of hydrodynamic cavitation on a chip in cellular injury and drug delivery.

Hydrodynamic cavitation (HC) is a phase change phenomenon, where energy release in a fluid occurs upon the collapse of bubbles, which form due to the low local pressures. During recent years, due to advances in lab-on-a-chip technologies, HC-on-a-chip (HCOC) and its potential applications have attracted considerable interest. Microfluidic devices enable the performance of controlled experiments by enabling spatial control over the cavitation process and by precisely monitoring its evolution. In this study, we propose the adjunctive use of HC to induce distinct zones of cellular injury and enhance the anticancer efficacy of Doxorubicin (DOX). HC caused different regions (lysis, necrosis, permeabilization, and unaffected regions) upon exposure of different cancer and normal cells to HC. Moreover, HC was also applied to the confluent cell monolayer following the DOX treatment. Here, it was shown that the combination of DOX and HC exhibited a more pronounced anticancer activity on cancer cells than DOX alone. The effect of HC on cell permeabilization was also proven by using carbon dots (CDs). Finally, the cell stiffness parameter, which was associated with cell proliferation, migration and metastasis, was investigated with the use of cancer cells and normal cells under HC exposure. The HCOC offers the advantage of creating well-defined zones of bio-responses upon HC exposure simultaneously within minutes, achieving cell lysis and molecular delivery through permeabilization by providing spatial control. In conclusion, micro scale hydrodynamic cavitation proposes a promising alternative to be used to increase the therapeutic efficacy of anticancer drugs.

Machine Learning Approach on High Risk Treadmill Exercise Test to Predict Obstructive Coronary Artery Disease by using P, QRS, and T waves' Features.

Treadmill Exercise Test (TET) results and patients' clinical symptoms influence cardiologists' decision to perform Coronary Angiography (CAG) which is an invasive procedure. Since TET has high false positive rates, it can cause an unnecessary invasive CAG. Our primary objective was to develop a machine learning model capable of optimizing TET performance based on electrocardiography (ECG) waves characteristics and signals. TET reports from 294 patients who underwent CAG following high risk TET were collected and categorized into those with critical CAD and others. The signal was converted to time series format. A dataset containing the P, QRS, and T wave times and amplitudes was created. Using this dataset, 5 machine learning algorithms were trained with 5-fold cross validation. All these models were then compared to the performance of cardiologists on V5 signal. The results from 5 machine learning models were clearly superior to the cardiologists' V5 signal performance (P < 0.0001). In addition, the XGBoost model, with an accuracy of 80.92±6.42% and an area under the curve (AUC) of 0.78±0.06, was the most successful model. Machine learning models can produce high-performance diagnoses using the V5 signal markers only as it does not require any clinical markers obtained from TET reports. This can lead to significant contributions to improving clinical prediction in non-invasive methods.

Acousto-holographic reconstruction of whole-cell stiffness maps.

Accurate assessment of cell stiffness distribution is essential due to the critical role of cell mechanobiology in regulation of vital cellular processes like proliferation, adhesion, migration, and motility. Stiffness provides critical information in understanding onset and progress of various diseases, including metastasis and differentiation of cancer. Atomic force microscopy and optical trapping set the gold standard in stiffness measurements. However, their widespread use has been hampered with long processing times, unreliable contact point determination, physical damage to cells, and unsuitability for multiple cell analysis. Here, we demonstrate a simple, fast, label-free, and high-resolution technique using acoustic stimulation and holographic imaging to reconstruct stiffness maps of single cells. We used this acousto-holographic method to determine stiffness maps of HCT116 and CTC-mimicking HCT116 cells and differentiate between them. Our system would enable widespread use of whole-cell stiffness measurements in clinical and research settings for cancer studies, disease modeling, drug testing, and diagnostics.

MobileSkin: Classification of Skin Lesion Images Acquired Using Mobile Phone-Attached Hand-Held Dermoscopes.

Dermoscopy is the visual examination of the skin under a polarized or non-polarized light source. By using dermoscopic equipment, many lesion patterns that are invisible under visible light can be clearly distinguished. Thus, more accurate decisions can be made regarding the treatment of skin lesions. The use of images collected from a dermoscope has both increased the performance of human examiners and allowed the development of deep learning models. The availability of large-scale dermoscopic datasets has allowed the development of deep learning models that can classify skin lesions with high accuracy. However, most dermoscopic datasets contain images that were collected from digital dermoscopic devices, as these devices are frequently used for clinical examination. However, dermatologists also often use non-digital hand-held (optomechanical) dermoscopes. This study presents a dataset consisting of dermoscopic images taken using a mobile phone-attached hand-held dermoscope. Four deep learning models based on the MobileNetV1, MobileNetV2, NASNetMobile, and Xception architectures have been developed to classify eight different lesion types using this dataset. The number of images in the dataset was increased with different data augmentation methods. The models were initialized with weights that were pre-trained on the ImageNet dataset, and then they were further fine-tuned using the presented dataset. The most successful models on the unseen test data, MobileNetV2 and Xception, had performances of 89.18% and 89.64%. The results were evaluated with the 5-fold cross-validation method and compared. Our method allows for automated examination of dermoscopic images taken with mobile phone-attached hand-held dermoscopes.

Deep convolutional neural networks for onychomycosis detection using microscopic images with KOH examination.

BACKGROUND: The diagnosis of superficial fungal infections is still mostly based on direct microscopic examination with potassium hydroxide solution. However, this method can be time consuming, and its diagnostic accuracy rates vary widely depending on the clinician's experience. OBJECTIVES: This study presents a deep neural network structure that enables the rapid solutions for these problems and can perform automatic fungi detection in grayscale images without dyes. METHODS: One hundred sixty microscopic full field photographs containing the fungal element, obtained from patients with onychomycosis, and 297 microscopic full field photographs containing dissolved keratin obtained from normal nails were collected. Smaller patches containing fungi (n = 1835) and keratin (n = 5238) were extracted from these full field images. In order to detect fungus and keratin, VGG16 and InceptionV3 models were developed by the use of these patches. The diagnostic performance of models was compared with 16 dermatologists by using 200 test patches. RESULTS: For the VGG16 model, the InceptionV3 model and 16 dermatologists, mean accuracy rates were 88.10 ± 0.8%, 88.78 ± 0.35% and 74.53 ± 8.57%, respectively; mean sensitivity rates were 75.04 ± 2.73%, 74.93 ± 4.52% and 74.81 ± 19.51%, respectively; and mean specificity rates were 92.67 ± 1.17%, 93.78 ± 1.74% and 74.25 ± 18.03%, respectively. The models were statistically superior to dermatologists according to rates of accuracy and specificity but not to sensitivity (p < .0001, p < .005 and p > .05, respectively). Area under curve values of the VGG16 and InceptionV3 models were 0.9339 and 0.9292, respectively. CONCLUSION: Our research demonstrates that it is possible to build an automated system capable of detecting fungi present in microscopic images employing the proposed deep learning models. It has great potential for fungal detection applications based on AI.

Micro-UFO (Untethered Floating Object): A Highly Accurate Microrobot Manipulation Technique.

A new microrobot manipulation technique with high precision (nano level) positional accuracy to move in a liquid environment with diamagnetic levitation is presented. Untethered manipulation of microrobots by means of externally applied magnetic forces has been emerging as a promising field of research, particularly due to its potential for medical and biological applications. The purpose of the presented method is to eliminate friction force between the surface of the substrate and microrobot. In an effort to achieve high accuracy motion, required magnetic force for the levitation of the microrobot was determined by finite element method (FEM) simulations in COMSOL (version 5.3, COMSOL Inc., Stockholm, Sweden) and verified by experimental results. According to position of the lifter magnet, the levitation height of the microrobot in the liquid was found analytically, and compared with the experimental results head-to-head. The stable working range of the microrobot is between 30 µm to 330 µm, and it was confirmed in both simulations and experimental results. It can follow the given trajectory with high accuracy (<1 µm error avg.) at varied speeds and levitation heights. Due to the nano-level positioning accuracy, desired locomotion can be achieved in pre-specified trajectories (sinusoidal or circular). During its locomotion, phase difference between lifter magnet and carrier magnet has been observed, and relation with drag force effect has been discussed. Without using strong electromagnets or bulky permanent magnets, our manipulation approach can move the microrobot in three dimensions in a liquid environment.