Anti-Aliasing Attention U-net Model for Skin Lesion Segmentation

The need for a lightweight and reliable segmentation algorithm is critical in various biomedical image-prediction applications. However, the limited quantity of data presents a significant challenge for image segmentation. Additionally, low image quality negatively impacts the efficiency of segmentation, and previous deep learning models for image segmentation require large parameters with hundreds of millions of computations, resulting in high costs and processing times. In this study, we introduce a new lightweight segmentation model, the mobile anti-aliasing attention u-net model (MAAU), which features both encoder and decoder paths. The encoder incorporates an anti-aliasing layer and convolutional blocks to reduce the spatial resolution of input images while avoiding shift equivariance. The decoder uses an attention block and decoder module to capture prominent features in each channel. To address data-related problems, we implemented data augmentation methods such as flip, rotation, shear, translate, and color distortions, which enhanced segmentation efficiency in the international Skin Image Collaboration (ISIC) 2018 and PH2 datasets. Our experimental results demonstrated that our approach had fewer parameters, only 4.2 million, while it outperformed various state-of-the-art segmentation methods.

Electrocardiogram Heartbeat Classification for Arrhythmias and Myocardial Infarction

An electrocardiogram (ECG) is a basic and quick test for evaluating cardiac disorders and is crucial for remote patient monitoring equipment. An accurate ECG signal classification is critical for real-time measurement, analysis, archiving, and transmission of clinical data. Numerous studies have focused on accurate heartbeat classification, and deep neural networks have been suggested for better accuracy and simplicity. We investigated a new model for ECG heartbeat classification and found that it surpasses state-of-the-art models, achieving remarkable accuracy scores of 98.5% on the Physionet MIT-BIH dataset and 98.28% on the PTB database. Furthermore, our model achieves an impressive F1-score of approximately 86.71%, outperforming other models, such as MINA, CRNN, and EXpertRF on the PhysioNet Challenge 2017 dataset.

The COVIDTW study: Clinical predictors of COVID-19 mortality and a novel AI prognostic model using chest X-ray

BACKGROUND

There is a lack of published research on the impact of the first wave of the COVID-19 pandemic in Taiwan. We investigated the mortality risk factors among critically ill patients with COVID-19 in Taiwan during the initial wave. Furthermore, we aim to develop a novel AI mortality prediction model using chest X-ray (CXR) alone.

METHOD

We retrospectively reviewed the medical records of patients with COVID-19 at Taipei Tzu Chi Hospital from May 15 to July 15 2021. We enrolled adult patients who received invasive mechanical ventilation. The CXR images of each enrolled patient were divided into 4 categories (1st, pre-ETT, ETT, and WORST). To establish a prediction model, we used the MobilenetV3-Small model with "Imagenet" pretrained weights, followed by high Dropout regularization layers. We trained the model with these data with Five-Fold Cross-Validation to evaluate model performance.

RESULT

A total of 64 patients were enrolled. The overall mortality rate was 45%. The median time from symptom onset to intubation was 8 days. Vasopressor use and a higher BRIXIA score on the WORST CXR were associated with an increased risk of mortality. The areas under the curve of the 1st, pre-ETT, ETT, and WORST CXRs by the AI model were 0.87, 0.92, 0.96, and 0.93 respectively.

CONCLUSION

The mortality rate of COVID-19 patients who receive invasive mechanical ventilation was high. Septic shock and high BRIXIA score were clinical predictors of mortality. The novel AI mortality prediction model using CXR alone exhibited a high performance.

Organ- and species-specific biological activity of rosmarinic acid

Rosmarinic acid (RA), a compound found in several plant species, has beneficial properties, including anti-inflammatory and antibacterial effects. We investigated the toxicity, anti-inflammatory, and antifibrotic effects of RA using precision-cut liver slices (PCLS) and precision-cut intestinal slices (PCIS) prepared from human, mouse, and rat tissue. PCLS and PCIS were cultured up to 48 h in the absence or presence of RA. Gene expression of the inflammatory markers: IL-6, IL-8/CXCL1/KC, and IL-1β, as well as the fibrosis markers: pro-collagen 1a1, heat shock protein 47, α-smooth muscle actin, fibronectin (Fn2) and plasminogen activator inhibitor-1 (PAI-1) were evaluated by qPCR. RA was only toxic in murine PCIS. RA failed to mitigate the inflammatory response in most models, while it clearly reduced IL-6 and CXCL1/KC gene expression in murine PCIS at non-toxic concentrations. With regard to fibrosis, RA decreased the gene levels of Fn2 and PAI-1 in murine PCLS, and Fn2 in murine PCIS. Yet, no effect was observed on the gene expression of fibrosis markers in human and rat PCIS. In conclusion, we observed clear organ- and species-specific effects of RA. RA had little influence on inflammation. However, our study further establishes RA as a potential candidate for the treatment of liver fibrosis.

A BAC library and paired-PCR approach to mapping and completing the genome sequence of Sulfolobus solfataricus P2

The original strategy used in the Sulfolobus solfataricus genome project was to sequence non overlapping, or minimally overlapping, cosmid or lambda inserts without constructing a physical map. However, after only about two thirds of the genome sequence was completed, this approach became counter-productive because there was a high sequence bias in the cosmid and lambda libraries. Therefore, a new approach was devised for linking the sequenced regions which may be generally applicable. BAC libraries were constructed and terminal sequences of the clones were determined and used for both end mapping and PCR screening. The PCR approaches included a novel chromosome walking method termed "paired-PCR". 21 gaps were filled by BAC end sequence analyses and 6 gaps were filled by PCR including three large ones by paired-PCR. The complete map revealed that 0.9 Mb remained to be sequenced and 34 BAC clones were selected for walking over small gaps and preparing template libraries for larger ones. It is concluded that an optimal strategy for sequencing microorganism genomes involves construction of a high-resolution physical map by BAC end analyses, PCR screening and paired-PCR chromosome walking after about half the genome sequence has been accumulated.

Cytochrome P450 and cyclooxygenase metabolites contribute to the endothelin-1 afferent arteriolar vasoconstrictor and calcium responses

Arachidonic acid metabolites contribute to the endothelin-1 (ET-1)-induced decrease in renal blood flow, but the vascular sites of action are unknown. Experiments performed in vitro used the rat juxtamedullary nephron preparation combined with videomicroscopy. The response of afferent arterioles to ET-1 was determined before and after cytochrome P450 (CYP450) or cyclooxygenase (COX) inhibition. Afferent arteriolar diameter averaged 20+/-1 microm (n=17) at a renal perfusion pressure of 100 mm Hg. Superfusion with 0.001 to 10 nmol/L ET-1 caused a graded decrease in diameter of the afferent arteriole. Vessel diameter decreased by 30+/-2% and 41+/-2% in response to 1 and 10 nmol/L ET-1, respectively. The afferent arteriolar response to ET-1 was significantly attenuated during administration of the CYP450 hydroxylase inhibitor N-methylsulfonyl-12,12-dibromododec-11-enamide (DDMS), such that afferent arteriolar diameter decreased by 19+/-3% and 22+/-3% in response to 1 and 10 nmol/L ET-1, respectively. COX inhibition also greatly attenuated the vasoconstriction elicited by ET-1, whereas the CYP450 epoxygenase inhibitor N-methylsulfonyl-6-(2-proparglyoxyphenyl) hexanamide enhanced the ET-1-mediated vascular response. Additional studies were performed using freshly isolated smooth muscle cells prepared from preglomerular microvessels. Renal microvascular smooth muscle cells were loaded with the calcium-sensitive dye fura 2 and studied by use of single-cell fluorescence microscopy. Basal renal microvascular smooth muscle cell [Ca(2+)](i) averaged 95+/-3 nmol/L (n=42). ET-1 (10 nmol/L) increased microvascular smooth muscle cell [Ca(2+)](i) to a peak value of 731+/-75 nmol/L before stabilizing at 136+/-8 nmol/L. Administration of DDMS or the COX inhibitor indomethacin significantly attenuated the renal microvascular smooth muscle cell calcium response to ET-1. These data demonstrate that CYP450 hydroxylase and COX arachidonic acid metabolites contribute importantly to the afferent arteriolar diameter and renal microvascular smooth muscle cell calcium responses elicited by ET-1.

Endothelin-mediated calcium signaling in preglomerular smooth muscle cells

This study was performed to test the hypothesis that endothelin peptides differentially influence intracellular calcium concentration ([Ca(2+)](i)) in preglomerular microvascular smooth muscle cells (MVSMC), in part through activation of endothelin (ET)(A) receptors. Experiments were performed in vitro with the use of single MVSMC freshly isolated from rat preglomerular microvessels. The effect of ET-1, ET-2, and ET-3 on [Ca(2+)](i) was measured with the use of the calcium-sensitive dye, fura 2, and standard fluorescence microscopy techniques. Baseline [Ca(2+)](i) averaged 84+/-3 nmol/L (n=141 cells from 23 dispersions). ET-1 concentrations of 1, 10, and 100 nmol/L evoked peak increases in [Ca(2+)](i) of 48+/-16, 930+/-125, and 810+/-130 nmol/L, respectively. The time course of the [Ca(2+)](i) response was biphasic, beginning with a rapid initial increase followed by a sustained plateau phase or a period during which [Ca(2+)](i) oscillated sharply. Similar responses were observed after ET-2 administration. In contrast, ET-3 stimulated monophasic increases in [Ca(2+)](i) of only 14+/-5, 33+/-16, and 44+/-19 nmol/L at peptide concentrations of 1, 10, and 100 nmol/L, respectively. These responses are significantly smaller than responses to ET-1 or ET-2, respectively. The relative contributions of calcium mobilization and calcium influx in the response to ET-1 were also evaluated. Removal of calcium from the bathing medium did not significantly alter the peak response to 10 nmol/L ET-1 but abolished the late phase elevation of [Ca(2+)](i). These data demonstrate that endothelin peptides increase [Ca(2+)](i) in preglomerular MVSMC. The concentration-response profiles are consistent with the response involving activation of ET(A) receptors. Furthermore, these results suggest that ET-1 increases [Ca(2+)](i) by stimulating both the release of intracellular calcium and the influx of calcium from the extracellular medium.

Purinoceptor-mediated calcium signaling in preglomerular smooth muscle cells

-The current studies were performed to determine the contribution of calcium mobilization and voltage-dependent calcium influx to the increase in [Ca2+]i elicited by ATP and UTP. Suspensions of freshly isolated smooth muscle cells were prepared from preglomerular microvessels by enzymatic digestion and loaded with the Ca2+-sensitive dye fura 2. The effect of ATP and UTP on [Ca2+]i was studied on single cells with standard microscope-based fluorescence photometry techniques. Resting [Ca2+]i averaged 80+/-3 nmol/L (n=219 single cells from 58 dispersions). ATP (100 micromol/L) increased [Ca2+]i to a peak value of 845+/-55 nmol/L (n=70 single cells from 38 dispersions) before stabilizing at 124+/-81 nmol/L. Similarly, 100 micromol/L UTP (n=39 single cells from 26 dispersions) stimulated a peak increase in [Ca2+]i of 1426+/-584 nmol/L before reaching a stable plateau of 123+/-10 nmol/L. The [Ca2+]i response to ATP and UTP was also assessed in the absence of extracellular calcium. In these studies, exposure to 100 micromol/L ATP induced a transient peak increase in [Ca2+]i, with the plateau phase being totally abolished. In contrast, exposure to 100 micromol/L UTP under calcium-free conditions resulted in no detectable change in the UTP-mediated increase in [Ca2+]i. The role of L-type calcium channels in the response was assessed with the calcium channel antagonist diltiazem. Incubation with diltiazem (10 micromol/L) markedly reduced the response to ATP, whereas the response to UTP was only slightly reduced. These data demonstrate that both ATP and UTP directly stimulate a biphasic increase in [Ca2+]i in renal microvascular smooth muscle cells. Furthermore, the data suggest that the elevation of [Ca2+]i elicited by ATP is largely dependent on calcium influx through L-type calcium channels, whereas the response to UTP appears to derive primarily from mobilization of calcium from intracellular stores.

National rehabilitation hospital limb classification for exercise, research, and clinical trials in post-polio patients

A need exists for an objective classification of polio patients for clinical and research purposes that takes into account the focal, asymmetric, and frequent subclinical nature of polio lesions. In order to prescribe a safe, effective exercise program, we developed a five-level (Classes I-V) limb-specific classification system based on remote and recent history, physical examination, and a four-extremity electrodiagnostic study (EMG/NCS). Class I limbs have no history of remote or recent weakness, normal strength, and a normal EMG. Class II limbs have no history of remote or recent weakness (or if remote history of weakness, full recovery occurred), normal strength and EMG evidence of prior anterior horn cell disease (AHCD). Class III limbs have a history of remote weakness with variable recovery, no new weakness, decreased strength, and EMG evidence of prior AHCD. Class IV limbs have a history of remote weakness with variable recovery, new clinical weakness, decreased strength, and EMG evidence of AHCD. Class V limbs have a history of severe weakness with little-to-no recovery, severely decreased strength and atrophy, and few-to-no motor units on EMG. In a prospective study of 400 limbs in 100 consecutive post-polio patients attending our clinic, 94 (23%) limbs were Class I, 88 (22%) were Class II, 95 (24%) were Class III, 75 (19%) were Class IV, and 48 (12%) were Class V. Guidelines for the use of this classification in a clinical/research setting are presented along with sample case histories and class-specific exercise recommendations.