SCAC: A Semi-Supervised Learning Approach for Cervical Abnormal Cell Detection.

Cervical abnormal cell detection plays a crucial role in the early screening of cervical cancer. In recent years, some deep learning-based methods have been proposed. However, these methods rely heavily on large amounts of annotated images, which are time-consuming and labor-intensive to acquire, thus limiting the detection performance. In this paper, we present a novel Semi-supervised Cervical Abnormal Cell detector (SCAC), which effectively utilizes the abundant unlabeled data. We utilize Transformer as the backbone of SCAC to capture long-range dependencies to mimic the diagnostic process of pathologists. In addition, in SCAC, we design a Unified Strong and Weak Augment strategy (USWA) that unifies two data augmentation pipelines, implementing consistent regularization in semi-supervised learning and enhancing the diversity of the training data. We also develop a Global Attention Feature Pyramid Network (GAFPN), which utilizes the attention mechanism to better extract multi-scale features from cervical cytology images. Notably, we have created an unlabeled cervical cytology image dataset, which can be leveraged by semi-supervised learning to enhance detection accuracy. To the best of our knowledge, this is the first publicly available large unlabeled cervical cytology image dataset. By combining this dataset with two publicly available annotated datasets, we demonstrate that SCAC outperforms other existing methods, achieving state-of-the-art performance. Additionally, comprehensive ablation studies are conducted to validate the effectiveness of USWA and GAFPN. These promising results highlight the capability of SCAC to achieve high diagnostic accuracy and extensive clinical applications.

Programmable Photoswitchable Microcapsules Enable Precise and Tailored Drug Delivery from Microfluidics.

The development of precision personalized medicine poses a significant need for the next generation of advanced diagnostic and therapeutic technologies, and one of the key challenges is the development of highly time-, space-, and dose-controllable drug delivery systems that respond to the complex physiopathology of patient populations. In response to this challenge, an increasing number of stimuli-responsive smart materials are integrated into biomaterial systems for precise targeted drug delivery. Among them, responsive microcapsules prepared by droplet microfluidics have received much attention. In this study, we present a UV-visible light cycling mediated photoswitchable microcapsule (PMC) with dynamic permeability-switching capability for precise and tailored drug release. The PMCs were fabricated using a programmable pulsed aerodynamic printing (PPAP) technique, encapsulating an aqueous core containing magnetic nanoparticles and the drug doxorubicin (DOX) within a poly(lactic-co-glycolic acid) (PLGA) composite shell modified by PEG-b-PSPA. Selective irradiation of PMCs with ultraviolet (UV) or visible light (Vis) allows for high-precision time-, space-, and dose-controlled release of the therapeutic agent. An experimentally validated theoretical model was developed to describe the drug release pattern, holding promise for future customized programmable drug release applications. The therapeutic efficacy and value of patternable cancer cell treatment activated by UV radiation is demonstrated by our experimental results. After in vitro transcatheter arterial chemoembolization (TACE), PMCs can be removed by external magnetic fields to mitigate potential side effects. Our findings demonstrate that PMCs have the potential to integrate embolization, on-demand drug delivery, magnetic actuation, and imaging properties, highlighting their immense potential for tailored drug delivery and embolic therapy.

Investigation of lethal thresholds of nanosecond pulsed electric field in rabbit VX2 hepatic tumors through finite element analysis and verification with a single-needle bipolar electrode: A prospective strategy employing three-dimensional comparisons.

Pulsed electric field has emerged as a promising modality for the solid tumor ablation with the advantage in treatment planning, however, the accurate prediction of the lesion margin requires the determination of the lethal electric field (E) thresholds. Herein we employ the highly repetitive nanosecond pulsed electric field (RnsPEF) to ablate the normal and VX2 tumor-bearing livers of rabbits. The ultrasound-guided surgery is operated using the conventional double- and newly devised single-needle bipolar electrodes. Finite element analysis is also introduced to simulate the E distribution in the practical treatments. Two- and three-dimensional investigations are performed on the image measurements and reconstructed calcification models on micro-CT, respectively. Specially, an algorithm considering the model surface, volume and shape is employed to compare the similarities between the simulative and experimental models. Blood vessel injury, temperature and synergistic efficacy with doxorubicin (DOX) are also investigated. According to the three-dimensional calculation, the overall E threshold is 4536.4 ± 618.2 V/cm and the single-needle bipolar electrode is verified to be effective in tissue ablation. Vessels are well preserved and the increment of temperature is limited. Synergy of RnsPEF and DOX shows increased apoptosis and improved long-term tumor survival. Our study presents a prospective strategy for the evaluation of the lethal E threshold, which can be considered to guide the future clinical treatment planning for RnsPEF.

Erratum to "A Low-Cost High-Performance Data Augmentation for Deep Learning-Based Skin Lesion Classification".

[This corrects the article DOI: 10.34133/2022/9765307.].

Evaluation of exogenous therapeutic protein activity under confinement and crowding effects.

Dysfunction of intracellular proteins is frequently associated with various diseases, such as cancer. The exogenous proteins in cells are usually assembled with specific configurations due to physiological confinement/crowding to exhibit novel features in the protein structure, folding or conformational stability, distinguished with their behaviors in buffer solutions. Here, we synthesized exogenous proteins under confined/crowded conditions, to explore protein activity within cells. The findings suggested that the confinement and crowding effects on protein activity are heterogeneous; they showed an inhibitory effect on HRP by decreasing Km from ∼9.5- and ∼21.7-fold and Vmax from ∼6.8- and ∼20.2-fold lower than that of dilute solutions. Interestingly, the effects on Cyt C seem to be more complicated, and crowding exerts a positive effect by increasing Km ∼ 3.6-fold and Vmax ∼ 1.5-fold higher than that of dilute solutions; however, confinement exhibits a negative effect by decreasing Km ∼2.0 and Vmax ∼8.3 times. Additionally, in contrast to traditional nanoparticle-based confinement models, we synthesized a biodegradable nanoparticle to mimic the confined space, and the biggest advantage of this novel model is that the particles can be degraded and thus it can provide more intuitive observations of the properties of the target proteins under confinement and after release. Furthermore, we also evaluated protein activity in different cellular environments, indicating that the exogenous protein activity was closely related to the crowdedness of cellular environments, and the inhibition of protein activity in MDA-MB-231 cancer cells was more obvious than in HEK293 normal cells. Finally, SAXS analysis revealed the correlation between the protein conformation and the different environments. Our work will provide a unique method for precisely assessing whether the target cellular environments are native matrix in which specific exogenous protein drugs are delivered to function or whether they display a therapeutic role, which is of great significance for screening and development of new drugs.

Modality preserving U-Net for segmentation of multimodal medical images.

Background: Recent advances in artificial intelligence and digital image processing have inspired the use of deep neural networks for segmentation tasks in multimodal medical imaging. Unlike natural images, multimodal medical images contain much richer information regarding different modal properties and therefore present more challenges for semantic segmentation. However, there is no report on systematic research that integrates multi-scaled and structured analysis of single-modal and multimodal medical images. Methods: We propose a deep neural network, named as Modality Preserving U-Net (MPU-Net), for modality-preserving analysis and segmentation of medical targets from multimodal medical images. The proposed MPU-Net consists of a modality preservation encoder (MPE) module that preserves the feature independency among the modalities and a modality fusion decoder (MFD) module that performs a multiscale feature fusion analysis for each modality in order to provide a rich feature representation for the final task. The effectiveness of such a single-modal preservation and multimodal fusion feature extraction approach is verified by multimodal segmentation experiments and an ablation study using brain tumor and prostate datasets from Medical Segmentation Decathlon (MSD). Results: The segmentation experiments demonstrated the superiority of MPU-Net over other methods in the segmentation tasks for multimodal medical images. In the brain tumor segmentation tasks, the Dice scores (DSCs) for the whole tumor (WT), the tumor core (TC) and the enhancing tumor (ET) regions were 89.42%, 86.92%, and 84.59%, respectively. In the meanwhile, the 95% Hausdorff distance (HD95) results were 3.530, 4.899 and 2.555, respectively. In the prostate segmentation tasks, the DSCs for the peripheral zone (PZ) and the transitional zone (TZ) of the prostate were 71.20% and 90.38%, respectively. In the meanwhile, the 95% HD95 results were 6.367 and 4.766, respectively. The ablation study showed that the combination of single-modal preservation and multimodal fusion methods improved the performance of multimodal medical image feature analysis. Conclusions: In the segmentation tasks using brain tumor and prostate datasets, the MPU-Net method has achieved the improved performance in comparison with the conventional methods, indicating its potential application for other segmentation tasks in multimodal medical images.

Stereotactic co-axial projection imaging for augmented reality neuronavigation: a proof-of-concept study.

Background: Lack of intuitiveness and poor hand-eye coordination present a major technical challenge in neurosurgical navigation. Methods: We developed an integrated dexterous stereotactic co-axial projection imaging (sCPI) system featuring orthotopic image projection for augmented reality (AR) neurosurgical navigation. The performance characteristics of the sCPI system, including projection resolution and navigation accuracy, were quantitatively verified. The resolution of the sCPI was tested with a USAF1951 resolution test chart. The stereotactic navigation accuracy of the sCPI was measured using a calibration panel with a 7×7 circle array pattern. In benchtop validation, the navigation accuracy of the sCPI and the BrainLab Kick Navigation Station was compared using a skull phantom with 8 intracranial targets. Finally, we demonstrated the potential clinical application of sCPI through a clinical trial. Results: The resolution test showed that the resolution of the sCPI was 1.3 mm. In a stereotactic navigation accuracy test, the maximum and minimum error of the sCPI was 2.9 and 0.3 mm, and the mean error was 1.5 mm. The stereotactic navigation accuracy test also showed that the navigation error of the sCPI would increase with the pitch and yaw angle, but there was no obvious difference in navigation errors caused by different yaw directions, which meant that the navigation error is unbiased across all directions. The benchtop validation showed that the average navigation errors for the sCPI system and the Kick Navigation Station were 1.4±0.8 and 1.8±0.7 mm, the medians were 1.3 and 1.9 mm, and the average preparation times were 3 min 24 sec and 6 min 8 sec, respectively. The clinical feasibility of sCPI-assisted neurosurgical navigation was demonstrated in a clinical study. In comparison with the BrainLab device, the sCPI system required less time for preoperative preparation and enhanced the clinician experience in intraoperative visualization and navigation. Conclusions: The sCPI technique can be potentially used in many surgical applications for intuitive visualization of medical information and intraoperative guidance of surgical trajectories.

A Redox-Activatable and Targeted Photosensitizing Agent to Deliver Doxorubicin for Combining Chemotherapy and Photodynamic Therapy.

Currently, tumors have become a serious disease threatening human health and life in modern society. Photo-chemo combination therapy is considered to be an important method to improving the efficiency of tumor treatment, especially in the treatment of multi-drug-resistant tumors. However, the application of photo-chemo combination therapy has been limited by the poor water solubility of photosensitizers, low tumor targeting, and high side effects of chemotherapy drugs. In order to solve these problems, a smart nano drug delivery platform FA-PEG-ss-PLL(-g-Ce6) designed and synthesized by us. The smart nano drug carrier uses folic acid (FA) as the targeting group, polyethylene glycol (PEG) as the hydrophilic end, Ce6-grafted polylysine (PLL(-g-Ce6)) as the hydrophobic end, and Chlorin e6 (Ce6) as the photosensitizer of photodynamic therapy, and it connects PEG to PLL by a redox-responsive cleavable disulfide linker (-ss-). Finally, the combination of tumor chemotherapy and photodynamic therapy (PDT) is realized by loading with anticancer drug doxorubicin (DOX) to the intelligent carrier. In vitro experiments showed that the drug loading content (DLC%) of DOX@FA-PEG-ss-PLL(-g-Ce6) nanoparticles (DOX@FPLC NPs) was as high as 14.83%, and the nanoparticles had good serum stability, reduction sensitivity and hemocompatibility. From the cytotoxicity assays in vitro, we found that under 664 nm laser irradiation DOX@FPLC NPs showed stronger toxicity to MCF-7 cells than did DOX, Ce6 + laser, and DOX + Ce6 + laser. Moreover, the antitumor efficiency in vivo and histopathological analysis showed that DOX@FPLC NPs under 664 nm laser irradiation exhibited higher antitumor activity and lower systemic toxicity than single chemotherapy. These results suggested that the FA-PEG-ss-PLL(-g-Ce6) nano drug delivery platform has considerable potential for the combination of chemotherapy and PDT.

A Single Model Deep Learning Approach for Alzheimer's Disease Diagnosis.

Early and accurate diagnosis of Alzheimer's disease (AD) and its prodromal period mild cognitive impairment (MCI) is essential for the delayed disease progression and the improved quality of patients' life. The emerging computer-aided diagnostic methods that combine deep learning with structural magnetic resonance imaging (sMRI) have achieved encouraging results, but some of them are limit of issues such as data leakage, overfitting, and unexplainable diagnosis. In this research, we propose a novel end-to-end deep learning approach for automated diagnosis of AD. This approach has the following differences from the current approaches: (1) Convolutional Neural Network (CNN) models of different structures and capacities are evaluated systemically and the most suitable model is adopted for AD diagnosis; (2) A data augmentation strategy named Two-stage Random RandAugment (TRRA) is proposed to alleviate the overfitting issue caused by limited training data and to improve the classification performance in AD diagnosis; (3) An explainable method of Grad-CAM++ is introduced to generate the visually explainable heatmaps to make our model more transparent. Our approach has been evaluated on two publicly accessible datasets for two classification tasks of AD vs. cognitively normal (CN) and progressive MCI (pMCI) vs. stable MCI (sMCI). The experimental results indicate that our approach outperforms the state-of-the-art approaches, including those using multi-model and three-dimensional (3D) CNN methods. The resultant heatmaps from our approach also highlight the lateral ventricle and some regions of cortex, which have been proved to be affected by AD.

A Low-Cost High-Performance Data Augmentation for Deep Learning-Based Skin Lesion Classification.

Objective and Impact Statement. There is a need to develop high-performance and low-cost data augmentation strategies for intelligent skin cancer screening devices that can be deployed in rural or underdeveloped communities. The proposed strategy can not only improve the classification performance of skin lesions but also highlight the potential regions of interest for clinicians' attention. This strategy can also be implemented in a broad range of clinical disciplines for early screening and automatic diagnosis of many other diseases in low resource settings. Methods. We propose a high-performance data augmentation strategy of search space 101, which can be combined with any model through a plug-and-play mode and search for the best argumentation method for a medical database with low resource cost. Results. With EfficientNets as a baseline, the best BACC of HAM10000 is 0.853, outperforming the other published models of "single-model and no-external-database" for ISIC 2018 Lesion Diagnosis Challenge (Task 3). The best average AUC performance on ISIC 2017 achieves 0.909 (±0.015), exceeding most of the ensembling models and those using external datasets. Performance on Derm7pt archives the best BACC of 0.735 (±0.018) ahead of all other related studies. Moreover, the model-based heatmaps generated by Grad-CAM++ verify the accurate selection of lesion features in model judgment, further proving the scientific rationality of model-based diagnosis. Conclusion. The proposed data augmentation strategy greatly reduces the computational cost for clinically intelligent diagnosis of skin lesions. It may also facilitate further research in low-cost, portable, and AI-based mobile devices for skin cancer screening and therapeutic guidance.

Single Model Deep Learning on Imbalanced Small Datasets for Skin Lesion Classification.

Deep convolutional neural network (DCNN) models have been widely explored for skin disease diagnosis and some of them have achieved the diagnostic outcomes comparable or even superior to those of dermatologists. However, broad implementation of DCNN in skin disease detection is hindered by small size and data imbalance of the publically accessible skin lesion datasets. This paper proposes a novel single-model based strategy for classification of skin lesions on small and imbalanced datasets. First, various DCNNs are trained on different small and imbalanced datasets to verify that the models with moderate complexity outperform the larger models. Second, regularization DropOut and DropBlock are added to reduce overfitting and a Modified RandAugment augmentation strategy is proposed to deal with the defects of sample underrepresentation in the small dataset. Finally, a novel Multi-Weighted New Loss (MWNL) function and an end-to-end cumulative learning strategy (CLS) are introduced to overcome the challenge of uneven sample size and classification difficulty and to reduce the impact of abnormal samples on training. By combining Modified RandAugment, MWNL and CLS, our single DCNN model method achieved the classification accuracy comparable or superior to those of multiple ensembling models on different dermoscopic image datasets. Our study shows that this method is able to achieve a high classification performance at a low cost of computational resources and inference time, potentially suitable to implement in mobile devices for automated screening of skin lesions and many other malignancies in low resource settings.

A photo-responsive membrane for tailored drug delivery with spatially and temporally controlled release.

Accurate delivery of therapeutics to tumor regions and effective sparing of normal tissue structures are important principles for the treatment of widespread metastases or malignant lesions in close proximity to vital organs. However, the currently available drug delivery techniques do not support precise drug release within the identified disease margins. We propose a tailored drug delivery strategy that utilizes a photo-responsive material in combination with tumor margin imaging for automated and tailored release of therapeutics. As a proof of concept, a poly(ethylene oxide)-b-PSPA (PEO-b-PSPA) diblock copolymer is synthesized by spiropyran (SP) polymerization. A photo-responsive membrane (PRM) is formed and irradiated with light sources of different wavelengths. Switching irradiation between ultraviolet light (UV) and green light (Vis) controls the permeability of the PRM in coincidence with the programmed irradiation patterns. The dynamic process of photo-switchable drug permeation through the PRM is modeled and compared with the experimental results. The strategy of tailored drug release is verified using both regular geometric shapes and metastatic cancer images. The therapeutic effect of this tailored drug release strategy is demonstrated in vitro in human breast cancer cells. Our pilot study implies the technical potential of using photo-responsive carriers for image-guided chemotherapy with precisely controlled drug release patterns.

Applying limiting entropy to quantify the alignment of collagen fibers by polarized light imaging.

Collagen alignment has shown clinical significance in a variety of diseases. For instance, vulvar lichen sclerosus (VLS) is characterized by homogenization of collagen fibers with increasing risk of malignant transformation. To date, a variety of imaging techniques have been developed to visualize collagen fibers. However, few works focused on quantifying the alignment quality of collagen fiber. To assess the level of disorder of local fiber orientation, the homogeneity index (HI) based on limiting entropy is proposed as an indicator of disorder. Our proposed methods are validated by verification experiments on Poly Lactic Acid (PLA) filament phantoms with controlled alignment quality of fibers. A case study on 20 VLS tissue biopsies and 14 normal tissue biopsies shows that HI can effectively characterize VLS tissue from normal tissue (P < 0.01). The classification results are very promising with a sensitivity of 93% and a specificity of 95%, which indicated that our method can provide quantitative assessment for the alignment quality of collagen fibers in VLS tissue and aid in improving histopathological examination of VLS.

Treatment and relapse in breast cancer show significant correlations to noninvasive testing using urinary and plasma DNA.

Aim: Circulating tumor DNA is promising for routine monitoring of breast cancer. Noninvasive testing allows regular probing using plasma and urine samples. Methods: Peripheral blood and simultaneous urine collection from patients were quantified. Concordance between methods were made. Serial time-point measurements were correlated to disease outcome. Results: Index measurements demonstrate over 90% concordance with biopsy. Receiver operating characteristics curves showed over 0.95 for both plasma and urine results comparing with controls. Patients with lower risk of relapse experienced greater declines in detected DNA levels. Maximal declines were registered at 4.0- and 6.8-fold for plasma and urine results, respectively. Conclusion: Measuring and monitoring DNA levels complement existing testing regimes and provides better risk profiling of patients for possible relapse.

Multimodal 3D Printing of Phantoms to Simulate Biological Tissue.

Biomedical optical imaging is playing an important role in diagnosis and treatment of various diseases. However, the accuracy and the reproducibility of an optical imaging device are greatly affected by the performance characteristics of its components, the test environment, and the operations. Therefore, it is necessary to calibrate these devices by traceable phantom standards. However, most of the currently available phantoms are homogeneous phantoms that cannot simulate multimodal and dynamic characteristics of biological tissue. Here, we show the fabrication of heterogeneous tissue-simulating phantoms using a production line integrating a spin coating module, a polyjet module, a fused deposition modeling (FDM) module, and an automatic control framework. The structural information and the optical parameters of a "digital optical phantom" are defined in a prototype file, imported to the production line, and fabricated layer-by-layer with sequential switch between different printing modalities. Technical capability of such a production line is exemplified by the automatic printing of skin-simulating phantoms that comprise the epidermis, dermis, subcutaneous tissue, and an embedded tumor.

Fundus-simulating phantom for calibration of retinal vessel oximetry devices.

Retinal vessel oxygen supply is important for retinal tissue metabolism. Commonly used retinal vessel oximetry devices are based on dual-wavelength spectral measurement of oxyhemoglobin and deoxyhemoglobin. However, there is no traceable standard for reliable calibration of these devices. In this study, we developed a fundus-simulating phantom that closely mimicked the optical properties of human fundus tissues. Microchannels of precisely controlled topological structures were produced by soft lithography to simulate the retinal vasculature. Optical properties of the phantom were adjusted by adding scattering and absorption agents to simulate different concentrations of fundus pigments. The developed phantom was used to calibrate the linear correlation between oxygen saturation (SO2) level and optical density ratio in a dual-wavelength oximetry device. The obtained calibration factors were used to calculate the retinal vessel SO2 in both eyes of five volunteers aged between 24 and 27 years old. The test results showed that the mean arterial and venous SO2 levels after phantom calibration were coincident with those after empirical value calibration, indicating the potential clinical utility of the produced phantom as a calibration standard.

Simulating orientation and polarization characteristics of dense fibrous tissue by electrostatic spinning of polymeric fibers.

Phantoms simulating polarization characteristics of soft tissue play an important role in the development, calibration, and validation of diagnostic polarized imaging devices and of therapeutic strategy, in both laboratory and clinical settings. We propose to fabricate optical phantoms that simulate polarization characteristics of dense fibrous tissues by bonding electrospun polylactic acid (PLA) fibers between polydimethylsiloxane (PDMS) substrate with a groove. Increasing the rotational speed of an electrospinning collector helps improve the orientation of the electrospun fibers. The phantoms simulate the polarization characteristics of dense fibrous tissue of collagenous fibroma and healthy skin with high fidelity. Our experiments demonstrate the technical potential of using such phantoms for validation and calibration of polarimetric medical devices.

Fabrication of a multilayer tissue-mimicking phantom with tunable optical properties to simulate vascular oxygenation and perfusion for optical imaging technology.

Vast research has been carried out to fabricate tissue-mimicking phantoms, due to their convenient use and ease of storage, to assess and validate the performance of optical imaging devices. However, to the best of our knowledge, there has been little research on the use of multilayer tissue phantoms for optical imaging technology, although their structure is closer to that of real skin tissue. In this work, we design, fabricate, and characterize multilayer tissue-mimicking phantoms, with a morphological mouse ear blood vessel, that contain an epidermis, a dermis, and a hypodermis. Each tissue-mimicking phantom layer is characterized individually to match specific skin tissue layer characteristics. The thickness, optical properties (absorption coefficient and reduced scattering coefficient), oxygenation, and perfusion of skin are the most critical parameters for disease diagnosis and for some medical equipment. These phantoms can be used as calibration artifacts and help to evaluate optical imaging technologies.

Design of a portable phantom device to simulate tissue oxygenation and blood perfusion.

We propose a portable phantom system for calibration and validation of medical optical devices in a clinical setting. The phantom system comprises a perfusion module and an exchangeable tissue-simulating phantom that simulates tissue oxygenation and blood perfusion. The perfusion module consists of a peristaltic pump, two liquid storage units, and two pressure suppressors. The tissue-simulating phantom is fabricated by a three-dimensional (3D) printing process with microchannels embedded to simulate blood vessels. Optical scattering and absorption properties of biologic tissue are simulated by mixing graphite powder and titanium dioxide powder with clear photoreactive resin at specific ratios. Tissue oxygen saturation (StO2) and blood perfusion are simulated by circulating the mixture of blood and intralipid at different oxygenation levels and flow rates. A house-made multimodal imaging system that combines multispectral imaging and laser speckle imaging are used for non-invasive detection of phantom oxygenation and perfusion, and the measurements are compared with those of a commercial Moor device as well as numerical simulation. By acquiring multimodal imaging data from one phantom and applying the calibration factors in different settings, we demonstrate the technical feasibility to calibrate optical devices for consistent measurements. By simulating retina tissue vasculature and acquiring functional images at different tissue oxygenation and blood perfusion levels, we demonstrate the clinical potential to simulate tissue anomalies. Our experiments imply the clinical potential of a portable, low-cost, and traceable phantom standard to calibrate and validate medical optical devices for improved performance.