Image Complexity Guided Network Compression for Biomedical Image Segmentation

Compression is a standard procedure for making convolutional neural networks (CNNs) adhere to some specific computing resource constraints. However, searching for a compressed architecture typically involves a series of time-consuming training/validation experiments to determine a good compromise between network size and performance accuracy. To address this, we propose an image complexity-guided network compression technique for biomedical image segmentation. Given any resource constraints, our framework utilizes data complexity and network architecture to quickly estimate a compressed model which does not require network training. Specifically, we map the dataset complexity to the target network accuracy degradation caused by compression. Such mapping enables us to predict the final accuracy for different network sizes, based on the computed dataset complexity. Thus, one may choose a solution that meets both the network size and segmentation accuracy requirements. Finally, the mapping is used to determine the convolutional layer-wise multiplicative factor for generating a compressed network. We conduct experiments using 5 datasets, employing 3 commonly-used CNN architectures for biomedical image segmentation as representative networks. Our proposed framework is shown to be effective for generating compressed segmentation networks, retaining up to ≈95% of the full-sized network segmentation accuracy, and at the same time, utilizing ≈32x fewer network trainable weights (average reduction) of the full-sized networks.

Administration of Antenatal Corticosteroids: Current State of Knowledge

The administration of a single course of corticosteroids before week 34 + 0 of gestation in cases with impending preterm birth is now standard procedure in obstetric care and firmly established in the guidelines of different countries. But despite the apparently convincing data, numerous aspects of this intervention have not yet been properly studied. It is still not clear which corticosteroid achieves the best results. There are very few studies on what constitutes an appropriate dose, circadian rhythms, the time frame in which corticosteroids are effective, and the balance between the risks and benefits of repeat administration. As the existing studies have rarely included patients before week 24 + 0 of gestation, we have very little information on the possible benefits of administering corticosteroids before this timepoint. If corticosteroids are administered antenatally after week 34 + 0 of gestation, the short-term benefit may be offset by the long-term adverse effect on psychomotor development. This present study summarizes the current state of knowledge regarding these issues.

Detection and Quantification of Stagonosporopsis cucurbitacearum in Seeds of Cucurbita maxima Using Droplet Digital Polymerase Chain Reaction

Stagonosporopsis cucurbitacearum is an important seedborne pathogen of squash (Cucurbita maxima). The aim of our work was to develop a rapid and sensitive diagnostic tool for detection and quantification of S. cucurbitacearum in squash seed samples, to be compared with blotter analysis, that is the current official seed test. In blotter analysis, 29 of 31 seed samples were identified as infected, with contamination from 1.5 to 65.4%. A new set of primers (DB1F/R) was validated in silico and in conventional, quantitative real-time PCR (qPCR) and droplet digital (dd) PCR. The limit of detection of S. cucurbitacearum DNA for conventional PCR was ∼1.82 × 10–2 ng, with 17 of 19 seed samples positive. The limit of detection for ddPCR was 3.6 × 10–3 ng, which corresponded to 0.2 copies/μl. Detection carried out with artificial samples revealed no interference in the absolute quantification when the seed samples were diluted to 20 ng. All seed samples that showed S. cucurbitacearum contamination in the blotter analysis were highly correlated with the absolute quantification of S. cucurbitacearum DNA (copies/μl) in ddPCR (R2 = 0.986; p ≤ 0.01). Our ddPCR protocol provided rapid detection and absolute quantification of S. cucurbitacearum, offering a useful support to the standard procedure.

Characterising the Area Under the Curve Loss Function Landscape

One of the most common metrics to evaluate neural network classifiers is the area under the receiver operating characteristic curve (AUC). However, optimisation of the AUC as the loss function during network training is not a standard procedure. Here we compare minimising the cross-entropy (CE) loss and optimising the AUC directly. In particular, we analyse the loss function landscape (LFL) of approximate AUC (appAUC) loss functions to discover the organisation of this solution space. We discuss various surrogates for AUC approximation and show their differences. We find that the characteristics of the appAUC landscape are significantly different from the CE landscape. The approximate AUC loss function improves testing AUC, and the appAUC landscape has substantially more minima, but these minima are less robust, with larger average Hessian eigenvalues. We provide a theoretical foundation to explain these results. To generalise our results, we lastly provide an overview of how the LFL can help to guide loss function analysis and selection.

Visco-Elastic Properties of Semi-Solid Alloys

Oscillation and creep experiments have been performed with Semi-Solid Material (SSM) AlSi7 with 35% solid fraction to investigate the early visco-elastic properties after shearing of the material in a Searle Rheometer. The preparation of the SSM has been done in situ using a standard procedure to guarantee for all experiments the same initial properties of the material. First, oscillation experiments at low amplitude allowed to study the evolution of material structure with time. Subsequently, creep experiments have been performed changing the resting period based on previous results. Creep experiments are characterized by exposing the material to a sudden increase of shear stress. The resolution in time has been 0.01 seconds, which allows observing the dynamics of the development of visco-elastic properties.The material exhibits viscoelastic properties that are becoming more pronounced with longer resting time. This is in accordance with previous experiments where the ratio between elastic and viscous properties increases with increasing resting time. The development of the elastic properties follows the increase of the yield stress due to the creation of an internal structure of the material, which starts immediately after stopping shearing. The investigation of the short-term response of SSM can be particularly relevant for industrial practice, where material deformation during die filling is very fast and the material flow does not take place in steady-state condition.

Lateral chest wall perforator flaps in partial breast reconstruction

Background Breast conserving surgery (BCS) has been a standard procedure for the treatment of breast cancer instead of mastectomy whenever possible. Lateral chest wall perforator flaps are one of the volume replacement techniques that participate in increasing the rate of BCS especially in small- to moderate-sized breasts with good cosmetic outcome. In this study, we tried to evaluate the outcome of those flaps as an oncoplastic procedure instead of the conventional flaps. Methods This study included 26 patients who underwent partial mastectomy with immediate reconstruction using lateral chest wall perforator flaps in the period from October 2019 to November 2020. The operative time, techniques, and complications were recorded. The cosmetic outcome was assessed 3 months post-radiation therapy through a questionnaire and photographic assessment. Results Lateral intercostal artery perforator (LICAP), lateral thoracic artery perforator (LTAP) and combined flaps were performed in 24, 1, and 1 patients, respectively. The mean operative time was 129.6 ± 13.2 min. The flap length ranged from 10 to 20 cm and its width from 5 to 9 cm. Overall patients’ satisfaction was observed to be 88.5% as either excellent or good and the photographic assessment was 96.2% as either excellent or good. Conclusions Lateral chest wall perforator flaps are reliable and safe option for partial breast reconstruction with an acceptable aesthetic outcome. In the era of oncoplastic breast surgery, they deserve to gain attention especially with the advantages of some modifications added to the classic technique.

Estimated Renal Metabolomics at Reperfusion Predicts One-Year Kidney Graft Function

Renal transplantation is the gold-standard procedure for end-stage renal disease patients, improving quality of life and life expectancy. Despite continuous advancement in the management of post-transplant complications, progress is still needed to increase the graft lifespan. Early identification of patients at risk of rapid graft failure is critical to optimize their management and slow the progression of the disease. In 42 kidney grafts undergoing protocol biopsies at reperfusion, we estimated the renal metabolome from RNAseq data. The estimated metabolites’ abundance was further used to predict the renal function within the first year of transplantation through a random forest machine learning algorithm. Using repeated K-fold cross-validation we first built and then tuned our model on a training dataset. The optimal model accurately predicted the one-year eGFR, with an out-of-bag root mean square root error (RMSE) that was 11.8 ± 7.2 mL/min/1.73 m2. The performance was similar in the test dataset, with a RMSE of 12.2 ± 3.2 mL/min/1.73 m2. This model outperformed classic statistical models. Reperfusion renal metabolome may be used to predict renal function one year after allograft kidney recipients.

The Modified Procedure of Totally Tubeless PNL

Mini percutaneous nephrolithotomy (mPNL) is a standard treatment for kidney stones larger than 1.5 cm, with the placement of a nephrostomy drainage at the end of it, which is considered the standard procedure, but tubeless/ totally tubeless mPNL techniques reduce postoperative discomfort in patients and shorten hospital stays. The aim of article was to compare the efficacy and safety of our proposed modified method of totally tubeless mPNL with control of the parenchymal canal, with existing methods of tubeless/totally tubeless mPNL. Novelty of the study presented by modified method of totally tubeless mPNL. During the period from 2018 to 2020 we performed 486 mPNL were performed in our clinic in total, among which 63 (12.9%) patients underwent tubeless PNL. Patients whose surgeries ended with using tubeless techniques were divided into three groups: Group I – 22 patients who had tubeless mPNL (with ureteral stent), Group II (20 patients) – totally tubeless mPNL with a safety thread (the proposed procedure), Group III (21 patients) – totally tubeless mPNL. In all three groups, the access point was most often made through the lower group of renal calyces: Group I – 12 (54.5%), Group II – 14 (70.0%), Group III – 13 (61.9%); then through the middle calyx: Group I – 8 (36.4%), Group II – 6 (30.0%), Group III – 7 (33.3%); and the upper calyx: Group І – 2 (9.1%), Group ІІ – 0%, Group ІІІ – 1 (4.8%), no differences in the distribution of access points between groups were found (p=0.67). There were no differences in the distribution of tract sizes between the groups (p=0.95) with tract dilatation to 16.5/17.5 Fr was performed most often: Group I – 12 (54.5%), in Group II – 11 (55.0%) and Group III – 11 (52.4%). The mean duration of surgery in Group I was 83.0±22.9 min, in Group II – 74.9±13.6 min, in Group III – 72.6±12.0 min (p=0.47). This study confirms the high effectiveness of totally tubeless mPNL. The proposed modification to perform totally tubeless mPNL allows you to have permanent postoperative control over the parenchymal channel and in case of postoperative bleeding it enables you to immediately insert nephrostomy drainage through the safety thread. Study contributes to practical methods as an intermediate step for surgeons who are considering transition to a totally tubeless PCNL technique.

Outcome Comparison of Endoscopic (Endonasal Trans-Sphenoidal) Repairs of CSF Leak vs. Transcranial Approach

OBJECTIVES: To compare the effectiveness of Endoscopic (endonasal transsphenoidal) repair of CSF leak with transcranial approach in terms of post-operative complications. METHODOLOGY: This study was conducted in the Department of Neurosurgery, Lady Reading Hospital, Peshawar. Total of 40 patients diagnosed according to inclusion criteria were enrolled and were divided into two groups. One group was treated with endonasal trans-sphenoidal repair, and another was treated with a trans-cranial approach. All patients were followed for 1 year. RESULTS: The mean age of enrolled patients was 35.4±11.6 years. There were 62.5% male and 37.5% female. In the endoscopic group the recurrence rate was observed in 3 (15%) of the patients while in the trans-cranial group the recurrence rate was observed in 2 (10%) of the patients. The overall recurrence and success rate was 8% and 92% respectively. About 4 patients developed an infection, which was treated successfully. CONCLUSION: It is concluded that the endoscopic approach is safe and effective. The endoscopic approach should be considered as standard procedure for treatment.

From Visual Grading and Dynamic Modulus of European Beech (Fagus sylvatica) Logs to Tensile Strength of Boards

The aim of the present paper is to assess the non-destructive indicating properties of Slovenian beech (Fagus sylvatica) logs and correlate them with the mechanical properties of the final product, which is boards. Beech logs were visually graded according to the standard procedure and vibrational frequencies were measured. Logs were further on sawn into boards which were also non-destructively tested in wet and dry conditions. Finally, the boards were experimentally tested in tension. Special focus was directed towards visual parameters of the beech logs and their influence on the overall quality of the output material. The longitudinal natural frequencies of the logs were studied as potential indicating properties. The results showed that a majority of the visual log grading parameters do not result in good quality timber in terms of strength and stiffness properties, and only few are decisive for the final classification. The coefficient of determination of the static MOE vs. dynamic MOE of logs was r2=0.13, whereas vs. the MOE of wet boards was r2=0.49. Using a few visual characteristics in combination with dynamic measurements of logs and of wet boards could help to increase the yield of high quality beech wood.