Inherent Variation Research Articles

Additive Manufacturing Technologies in Dentistry (A Review)

Background: Significant advancements have been observed in the additive manufacturing (AM) technology industry in recent decades. Due to the inherent variations among each AM manufacturing technique, new areas of investigation continually arise and require consideration. Additionally, the novel applications of additive manufacturing present new difficulties and possibilities for targeted focus. The aim of this manuscript is to conduct a comprehensive literature review that describes the various processing methods, precision levels, types of materials utilized, and potential applications of 3D printing technology in the field of dentistry. Data: An online search was conducted on databases including Research Gate, Google Scholar, and PubMed to identify potential applications of AM technologies in the dental industry. The most relevant studies on the subject were selected, including English-language articles published between 2006 and 2022. Conclusion: It is feasible to incorporate a variety of AM techniques in dentistry, which has led to improved workflow and acceptable clinical results. Moreover, the different technologies of 3D printing have a broad array of potential applications, enabling the development of novel and optimized techniques to produce dental products.

Physical unclonable function using photonic spin Hall effect

This study presents a novel method leveraging surface wave-assisted photonic spin Hall effect (PSHE) to construct physical unclonable functions (PUFs). PUFs exploit inherent physical variations to generate unique Challenge–Response pairs, which are critical for hardware security and arise from manufacturing discrepancies, device characteristics, or timing deviations. We explore PSHE generation-based PUF design, expanding existing design possibilities. With recent applications in precise sensing and computing, PSHE offers promising performance metrics for our proposed PUFs, including an inter-Hamming distance of 47.50% , an average proportion of unique responses of 62.5% , and a Pearson correlation coefficient of − 0.198. The PUF token demonstrates robustness to simulated noise. Additionally, we evaluate security using a machine learning-based attack model, employing a multi-layer perceptron (MLP) regression model with a randomized search method. The average accuracy of successful attack prediction is 9.70% for the selected dataset. Our novel PUF token exhibits high non-linearity due to the PSHE effect, resilience to MLP-based attacks, and sensitivity to process variation.

Association between specific human leukocyte antigen alleles and development of thyroid immune-related adverse event.

Aim: Inherent variations in human leukocyte antigen (HLA) alleles have been revealed epidemiologically to influence the development of autoimmune diseases. HLA alleles may thus also be associated with the development of immune-related adverse events (irAEs), such as thyroid irAE.Materials & methods: In this case-control study, 71 cancer patients who received immune checkpoint inhibitors were enrolled and HLA-genotyped and the frequency of HLA alleles was compared.Results: A*26:01, DPA1*01:03 and DPB1*02:01 were significantly more frequent in patients with thyroid irAE than in patients without any irAEs (35.0 vs3.2% [p=0.004], 80.0 vs45.2% [p=0.020] and 55.0 vs25.8% [p=0.044], respectively).Conclusion: A*26:01, DPA1*01:03 and DPB1*02:01 appear to be associated with thyroid irAE.

CarbonScaler: Leveraging Cloud Workload Elasticity for Optimizing Carbon-Efficiency

Due to inherent variations in energy's carbon intensity, temporal shifting has become a key method in reducing the carbon footprint of batch workloads. However, temporally shifting workloads involves searching for periods with lower carbon intensity, which increases the workload's completion time. In this paper, we present CarbonScaler, a new approach that reduces carbon emissions of batch workloads without extending their completion time. Our approach relies on applications' ability to change their compute demand by scaling the workload based on fluctuations in energy's carbon intensity. We present a carbon-aware scheduling algorithm, a Kubernetes-based prototype, and an analytic tool to guide the carbon-efficient deployment of batch applications in the cloud. We evaluate CarbonScaler using real-world applications and show that it can yield 33% carbon savings without extending the completion time and up to 32% extra carbon savings over state-of-the-art suspend-resume policies when completion time is flexible.

Visualizing Anthocyanins: Colorimertic Analysis of Blue Maize

Anthocyanin, vibrant pigments found in a wide range of plants, including maize, contribute to the red, blue, and purple hues observed in fruits, vegetables, and grains. The inherent color variations in maize, including natural shades of purple, red, blue, and even rainbow colors, pose a significant challenge in accurately assessing maize maturity. This study recognizes the importance of visualizing the distinct blue purplish anthocyanin coloration to determine the optimal harvest time for blue maize, particularly among small-scale producers. To address this crucial need, this research project presents the development of the MaizeMeter, an advanced colorimeter specifically designed to analyze maize color based on anthocyanin pigmentation. Leveraging the power of Internet of Things (IoT) implementation, the MaizeMeter provides real-time monitoring and interpretation of anthocyanin color values. The proposed methodology encompasses the calibration of the color sensor and the prototyping of the MaizeMeter, culminating in the establishment of a comprehensive database of anthocyanin color profiles in blue maize. The generated anthocyanin color database by the MaizeMeter will serve as a vital tool for small-scale farmers and researchers, enabling more efficient and accurate assessment of maize maturity in the future.

Inherent versus random variation in fitness of elephant seals: offspring quality and quantity

Variation in reproductive success is the basis of evolution and allows species to respond to the environment, but only when it is based on fixed individual variation that is heritable. Several recent studies suggest that observed variation in reproduction is due to chance, not inherent individual differences. Our aim was to quantify inherent versus neutral variation in fitness of northern elephant seal ( Mirounga angustirostris (Gill, 1866)) females, including both quality and quantity of their offspring. Using 44 years of observations at Año Nuevo in California, we assembled lifetime pup production of 1065 individual females and mass at weaning for 2120 of their pups. Females varied significantly in mean lifetime mass of their pups, with 28% of the variance due to fixed individual differences among mothers. Variation was repeatable over 6 years of a mother’s lifetime and heritable ( h = 0.48). Moreover, pup mass at weaning was associated with future lifetime fitness, since larger pups had a higher chance of surviving to breed. Larger pups, however, did not produce more offspring once breeding, and lifetime pup production was not heritable. Traits related to offspring quality in elephant seals were inherently different among females, but variation in pup production was neutral.

A Pretty Kettle of Fish: A Review on the Current Challenges in Mediterranean Teleost Reproduction.

The Mediterranean region is facing several environmental changes and pollution issues. Teleosts are particularly sensitive to these challenges due to their intricate reproductive biology and reliance on specific environmental cues for successful reproduction. Wild populations struggle with the triad of climate change, environmental contamination, and overfishing, which can deeply affect reproductive success and population dynamics. In farmed species, abiotic factors affecting reproduction are easier to control, whereas finding alternatives to conventional diets for farmed teleosts is crucial for enhancing broodstock health, reproductive success, and the sustainability of the aquaculture sector. Addressing these challenges involves ongoing research into formulating specialized diets, optimizing feeding strategies, and developing alternative and sustainable feed ingredients. To achieve a deeper comprehension of these challenges, studies employing model species have emerged as pivotal tools. These models offer advantages in understanding reproductive mechanisms due to their well-defined physiology, genetic tractability, and ease of manipulation. Yet, while providing invaluable insights, their applicability to diverse species remains constrained by inherent variations across taxa and oversimplification of complex environmental interactions, thus limiting the extrapolation of the scientific findings. Bridging these gaps necessitates multidisciplinary approaches, emphasizing conservation efforts for wild species and tailored nutritional strategies for aquaculture, thereby fostering sustainable teleost reproduction in the Mediterranean.

Experimental investigation on the mechanical properties of cement mortar containing recycled tire crumbs reinforced with polypropylene, oil palm and banana fibers

The automotive industry discards a large number of end-of-life tires annually, creating a growing global environmental hazard. The incorporation of crumb rubber derived from waste tires into construction materials is considered one of several alternatives for sustainably reusing these waste materials. The addition of fibers as reinforcements would address the typical degradation caused by the crumb rubber in cement composites. Fibers sourced from plants and recycled synthetics offer a practical and eco-friendly reinforcement material that is both cost-effective and easy to procure. However, two challenges emerge from these choices. Firstly, natural fibers exhibit substantial regional variations, resulting in differences in their properties and inconsistent performance. Secondly, recycled materials also demonstrate variation due to disparities in production and manufacturing processes. As a result, incorporating these fibers into cement-based materials has produced conflicting outcomes due to these inherent variations, rendering practical application unfavorable. Therefore, this study attempts to promote the adoption of sustainable practices by addressing the lack of available performance-based data for the utilization of these fiber variants in the design of cement mixtures. The data would allow engineers to integrate these mixtures into commercial designs for practical in-situ construction. A total of 18 novel cement-based mixtures containing different types of fibers, chemical admixture, and discarded crumb waste were formulated and evaluated in fresh and hardened properties. A multifaceted approach was also undertaken to further hybridize the fibers and quantify the synergistic effects within the investigated cement matrice. The results indicate that the incorporation of polypropylene and oil palm fibers enhances compressive and tensile strength, although there is a general weakening observed in flexural strength. In the case of hybrid fibers, a successful combination of polypropylene and banana fibers is evident, showing a positive hybrid effect coefficient compared to a negative coefficient for polypropylene-oil palm fiber combinations.

Carbon Binder Domain Inhomogeneity in Silicon-Monoxide/Graphite Composite Anode by 2D Multiphysics Modeling.

The Carbon-binder domain (CBD) plays a pivotal role in the performance of lithium-ion battery electrodes. The heterogeneous distribution of CBD across the electrode has garnered significant attention. However, a thorough understanding of how this CBD inhomogeneity affects anode performance remains a crucial pursuit, especially when considering the inherent material variations present in the SiO/Graphite (SiO/Gr) composite anode. In this study, an electro-chemo-mechanical model is established that provides a detailed geometric description of the particles. This model allows to quantitatively uncover the effects of CBD inhomogeneity on the fundamental behaviors of the SiO/Gr composite anode. The findings indicate that reducing the proportion of CBD in the upper domain (near the anode surface) compared to the lower domain (near the current collector) positively influences electrochemical performance, particularly in terms of capacity and Li plating. However, such an arrangement introduces potential risks of mechanical failures, and it is recommended to incorporate a higher proportion of CBD alongside the SiO particles. Finally, an anode design with a lower CBD proportion in the upper domain exhibits superior rate performance. This study represents a pioneering modeling exploration of CBD inhomogeneity, offering a promising multiphysics model with significant potential for informing advanced battery design considerations.

Addressing class-imbalanced learning in real-time aero-engine gas-path fault diagnosis via feature filtering and mapping

Condition-based maintenance of aero-engines requires real-time gas-path fault diagnosis, which is crucial for reducing costs and enhancing aircraft attendance. It is imperative to ensure a low false alarm rate in fault diagnosis algorithms. Inherent variations between engines and age-related performance deterioration can lead to an increase in false alarms, necessitating online algorithm tuning. However, insufficient fault data in online tuning may lead to class imbalance and algorithmic forgetfulness of fault features, compromising effectiveness. To tackle this issue, this paper introduces a novel hybrid diagnosis algorithm that utilizes feature filtering and mapping. This method entails an engine model, a feature extractor, a feature filter, a fault classifier, and several feature mappers. The feature mappers are designed to generate fault features based on the incoming fault free data for rebalancing the tuning dataset. And the feature filter is introduced to prevent weight collapse during the online tuning. Validation through a 400-hour life cycle simulation test and a real data test indicate that the method outperforms the benchmark methods. It overcomes the class-imbalanced learning problem, showcasing its potential for aero-engine gas-path fault diagnosis.

Fertility Assessment of Sexed Semen with Synchronized and Spontaneous Estrus in Crossbred Cows

Background: With the commercial introduction of sexed semen, its use has been recommended for insemination at observed estrus, avoiding the use of a fixed-time insemination due to inherent variation in success of timed programmes. The use of the Ovsynch protocol with sexed semen at defined conditions such as normal clinical and reproductive status of lactating cows might result in a pregnancy rate comparable to that achievable at insemination upon detection of spontaneous estrus with sexed semen. Considering this rationale, present research was implemented with the effect of fertility on synchronized and spontaneous estrus with sexed semen. Methods: Post-partum multiparous crossbred cows (n=28) were categorized as a group I (Spontaneous; n=14) and group II (Ovsynch; n=14). The fourteen cows from group I were inseminated with sexed semen at observed estrus after initial treatment with vitamin AD3E and H, urimin and supplemented with chelated mixture and the fourteen cows from the group II were synchronized with Ovsynch synchronization protocol (Inj. Buserline acetate 10 µg I/M) on day 0, (Inj. Cloprostenol sodium 500 µg) on day 7 and (Inj. Buserline acetate 10 µg) I/M on day 9. Thereafter, timed artificial insemination (TAI) was performed 16-20 hours after 2nd dose of GnRH with sexed semen. Result: The percent intense type of intensity recorded numerically higher in induced estrus. The first service conception rate was higher in spontaneous estrus with the overall conception rate 39.28%. Cows with 60-100 days in milk was showed high conception rate with sexed semen, as the parity increases the conception rate decreases with sexed semen The overall proportion of female calves born was 90.90%.

Multi-timescale neuromodulation strategy for closed-loop deep brain stimulation in Parkinson’s disease

Objective. Beta triggered closed–loop deep brain stimulation (DBS) shows great potential for improving the efficacy while reducing side effect for Parkinson’s disease. However, there remain great challenges due to the dynamics and stochasticity of neural activities. In this study, we aimed to tune the amplitude of beta oscillations with different time scales taking into account influence of inherent variations in the basal ganglia–thalamus–cortical circuit. Approach. A dynamic basal ganglia–thalamus–cortical mean–field model was established to emulate the medication rhythm. Then, a dynamic target model was designed to embody the multi-timescale dynamic of beta power with milliseconds, seconds and minutes. Moreover, we proposed a closed-loop DBS strategy based on a proportional–integral–differential (PID) controller with the dynamic control target. In addition, the bounds of stimulation amplitude increments and different parameters of the dynamic target were considered to meet the clinical constraints. The performance of the proposed closed-loop strategy, including beta power modulation accuracy, mean stimulation amplitude, and stimulation variation were calculated to determine the PID parameters and evaluate neuromodulation performance in the computational dynamic mean–field model. Main results. The Results show that the dynamic basal ganglia–thalamus–cortical mean–field model simulated the medication rhythm with the fasted and the slowest rate. The dynamic control target reflected the temporal variation in beta power from milliseconds to minutes. With the proposed closed-loop strategy, the beta power tracked the dynamic target with a smoother stimulation sequence compared with closed-loop DBS with the constant target. Furthermore, the beta power could be modulated to track the control target under different long-term targets, modulation strengths, and bounds of the stimulation increment. Significance. This work provides a new method of closed–loop DBS for multi-timescale beta power modulation with clinical constraints.

Group sparsity residual constraint model with weighted log-sum penalty for image restoration

In the field of image restoration, non-convex penalties and non-local self-similarity (NSS) prior have been found to enhance image edges and textures. However, NSS-based methods often overlook the inherent variations within patch groups by treating them as a single entity for processing. Despite the existence of numerous convex or non-convex regularizers, few of them have been able to adaptively adjust penalty strength based on NSS information in order to preserve fine details in an image. To address this issue, we propose a novel group sparsity residual constraint model with weighted log-sum penalty (GSRC-Log) for image restoration in this paper. The proposed regularization term is parameterized by an adaptive scalar parameter that relates to NSS, allowing the model to prioritize high NSS group sparse coefficients while penalizing low NSS coefficients. For image denoising, we provide detailed analysis on the optimal condition of our proposed model and prove that it has a simple closed-form global solution. For image restoration tasks, we utilize the ADMM algorithm to solve the GSRC-Log problem and empirically verify its convergence. Extensive experimental results on both image denoising and compressive sensing (CS) recovery demonstrate that our proposed GSRC-Log outperforms many popular or state-of-the-art methods.

Matheuristics for mixed-model assembly line balancing problem with fuzzy stochastic processing time

Our work aims to investigate methods for solving the mixed-model assembly line balancing problem (MALBP) under uncertainty with the objective of minimizing the number of workstations. Specifically, we model task processing time as fuzzy stochastic variables (FRVs) due to the inherent uncertainties and variations in the manufacturing environment. Additionally, we introduce a ranking method for FRVs and propose a mathematical model to address MALBP. The recently developed Red Fox Optimization (RFO) algorithm is also discretized for the first time to support solving this problem. Finally, matheuristic algorithms combine a metaheuristic such as the popular Genetic Algorithm (GA), Particle Swarm Optimization (PSO), or the Discretized Red Fox Optimization (DRFO) algorithm with the Mixed-Integer Programming (MIP) model to generate the best solution in a reasonable time. Our comparative results demonstrate that the GA-MIP combination outperforms the others in both objective value and computational time.

Establishment of a novel protocol for assessing the severity of subarachnoid hemorrhage in circle Willis perforation mouse model

The Circle of Willis perforation (cWp) mouse model is a key tool in subarachnoid hemorrhage (SAH) research; however, inconsistent bleeding volumes can challenge experimental reliability. To address this issue, we introduced the ROB Scoring System, a novel protocol integrating Rotarod Tests (RT), Open-field Tests (OT) video analysis, and daily Body Weight Loss (BWL) monitoring to precisely categorize SAH severity. Forty C57BL/6 mice underwent cWp SAH induction, categorized by ROB into severity subgroups (severe, moderate, mild). Validation compared ROB trends in subgroups, and ROB outcomes with autopsy results on postoperative days three and seven for acute and sub-acute evaluations. Mortality rates were analyzed via the survival log-rank test, revealing a significant difference among SAH subgroups (P < 0.05). Strong correlations between ROB grades and autopsy findings underscored its precision. Notably, the severe group exhibited 100% mortality within 4 days post SAH onset. Single parameters (RT, OT, BWL) were insufficient for distinguishing SAH severity levels. The ROB score represents a significant advancement, offering an objective method for precise categorization and addressing inherent bleeding variations in the cWp SAH model. This standardized protocol enhances the reliability and effectiveness of the SAH translational research, providing a valuable tool for future investigations into this critical area.

Abstract PO5-26-09: Digital image analysis and a novel set of cell line samples as aids in the development of a quantitative external quality assessment programme for Ki-67

Abstract Background and Aims Ki-67 is a well-established biomarker of proliferation in breast cancer (BC). However, its value in treatment decision making is hampered by a lack of analytical reproducibility. Regular participation in external quality assessment (EQA) substantially improves inter-laboratory concordance. Pre-requisites in establishing a fit-for-purpose EQA for Ki-67 are a well-validated testing substrate and a reproducible method of assessing Ki-67 scores in participants’ submitted materials. We report here on the results of work using cell line controls analysed by digital image analysis (DIA) as a first step towards providing such an EQA. Materials and Methods A formalin-fixed paraffin embedded (FFPE) cell line microarray (CLMA) was designed and produced in conjunction with Array Sciences LLC (Sausalito, USA). It was comprised of cores taken from a pure population of Sf9 caterpillar cells, which have been shown to be completely unreactive with most commercial antibodies to human Ki-67, together with cores of Sf9 cells mixed with four different human BC cell lines. These were BT-20 (85% BC cells), ZR-75-1 (75%), BT-474 (65%) and BT-483 (55%). Sections from the CLMA were mounted onto glass microscope slides together with sections from a FFPE tonsil sample and two BC samples; one BC showed high (~30%, BC-high) and the second, low proliferation (~5%, BC-low). In both BC samples proliferating cancer cells were distributed homogeneously throughout the the block. Unstained sections were distributed to laboratories participating in the Scheme’s Ki-67 BC programme; after routine IHC-staining for Ki-67 returned slides were centrally visually assessed for stain quality and subjected to DIA using an in-house application developed on Visiopharm software (Visiopharm A/S, Hoersholm, Denmark). Results Slides were returned by 37 laboratories. Analysis of Ki-67 scores obtained on the Sf9 core identified two distinct groups. The first (n = 27) were negative or showed low Ki-67 scores (mean = 1.1%, 95% CIs: 0.2-1.9%), the second (n = 10) displayed a step-change in scores (mean = 49.9%, 95% CIs: 33.2-66.7%); the means of the two groups were significantly different (P&amp;lt; 0.0001). When Ki-67 scores for each of the tissue samples were dichotomized into similar groups, the means of those groups differed significantly for the two BC samples (P&amp;lt; 0.001), but not for tonsil. Quality scores generated by visual assessment did not differ significantly between the two groups. However, when slides bearing Sf9 cores demonstrating aberrant Ki-67 scores were visually examined nuclear staining was clearly visible, and non-specific nuclear staining could also be identified in the matched BC tissue samples, but not in the tonsil sections. Correlation of Ki-67 scores between the four BC cell line cores and each of the BC tissue samples was examined using Pearson’s correlation statistics. The r statistic range was 0.57-0.71 in comparisons between BC cell line cores and the BC-high sample, and 0.53-0.70 for those with BC-low; in each case BT-483 showed the highest correlation score and BT-20 the lowest. A similar analysis was undertaken between the tonsil and the two BC tissues. The r statistic for correlation between tonsil and BC-high scores was 0.72; it was 0.64 for those between tonsil and BC-low. Conclusions By using a pure population of Sf9 cells we have developed a sensitive indicator of non-specific nuclear staining in IHC preparations stained for Ki-67 which identifies the presence of the artefact quantifiably. Cores made from Sf9/BC cell line mixtures (especially BT-483) produce Ki-67 scores which correlate with those obtained in breast cancer samples at a similar level to those achieved between tonsil and BC samples; this is true for both high and the low proliferation ranges. Cell line mixtures can be adjusted to show Ki-67 scores in the clinically relevant ranges, and they do not show the inherent biological variations seen in tissue controls such as tonsil. Citation Format: Andrew Dodson, Fitim Berisha, Dawn Wilkingson, Lila Zabaglo, Suzanne Parry. Digital image analysis and a novel set of cell line samples as aids in the development of a quantitative external quality assessment programme for Ki-67 [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-26-09.

A deep learning based multi-model approach for predicting drug-like chemical compound’s toxicity

Ensuring the safety and efficacy of chemical compounds is crucial in small-molecule drug development. In the later stages of drug development, toxic compounds pose a significant challenge, losing valuable resources and time. Early and accurate prediction of compound toxicity using deep learning models offers a promising solution to mitigate these risks during drug discovery. In this study, we present the development of several deep-learning models aimed at evaluating different types of compound toxicity, including acute toxicity, carcinogenicity, hERG_cardiotoxicity (the human ether-a-go-go related gene caused cardiotoxicity), hepatotoxicity, and mutagenicity. To address the inherent variations in data size, label type, and distribution across different types of toxicity, we employed diverse training strategies. Our first approach involved utilizing a graph convolutional network (GCN) regression model to predict acute toxicity, which achieved notable performance with Pearson R 0.76, 0.74, and 0.65 for intraperitoneal, intravenous, and oral administration routes, respectively. Furthermore, we trained multiple GCN binary classification models, each tailored to a specific type of toxicity. These models exhibited high area under the curve (AUC) scores, with an impressive AUC of 0.69, 0.77, 0.88, and 0.79 for predicting carcinogenicity, hERG_cardiotoxicity, mutagenicity, and hepatotoxicity, respectively. Additionally, we have used the approved drug dataset to determine the appropriate threshold value for the prediction score in model usage. We integrated these models into a virtual screening pipeline to assess their effectiveness in identifying potential low-toxicity drug candidates. Our findings indicate that this deep learning approach has the potential to significantly reduce the cost and risk associated with drug development by expediting the selection of compounds with low toxicity profiles. Therefore, the models developed in this study hold promise as critical tools for early drug candidate screening and selection.

Intraductal magnetic resonance imaging of cholangiocarcinoma -a practical possibility.

Intraductal T2 mapping based on a catheter receiver is proposed as a method of visualizing the extent of intraductal and periductal cholangiocarcinoma (CCA). Compared to external receivers, internal receivers provide locally enhanced signal-to-noise ratios by virtue of their lower field-of-view for body noise, allowing smaller voxels and higher resolution. However, inherent radial sensitivity variation and segmentation for patient safety both distort image brightness. We discuss simulated T2 weighted images and T2 maps, and in vitro images obtained using a thin film catheter receiver of a freshly resected liver specimen containing a polypoid intraductal tumor from a patient with CCA. T2 mapping provides a simple method of compensating non-uniform signal reception patterns of catheter receivers, allowing the visualization of tumor extent without contrast enhancement and potentially quantitative tissue characterization. Potential advantages and disadvantages of in vivo intraductal imaging are considered.

Domain generalization for retinal vessel segmentation via Hessian-based vector field

Blessed by vast amounts of data, learning-based methods have achieved remarkable performance in countless tasks in computer vision and medical image analysis. Although these deep models can simulate highly nonlinear mapping functions, they are not robust with regard to the domain shift of input data. This is a significant concern that impedes the large-scale deployment of deep models in medical images since they have inherent variation in data distribution due to the lack of imaging standardization. Therefore, researchers have explored many domain generalization (DG) methods to alleviate this problem. In this work, we introduce a Hessian-based vector field that can effectively model the tubular shape of vessels, which is an invariant feature for data across various distributions. The vector field serves as a good embedding feature to take advantage of the self-attention mechanism in a vision transformer. We design paralleled transformer blocks that stress the local features with different scales. Furthermore, we present a novel data augmentation method that introduces perturbations in image style while the vessel structure remains unchanged. In experiments conducted on public datasets of different modalities, we show that our model achieves superior generalizability compared with the existing algorithms. Our code and trained model are publicly available at https://github.com/MedICL-VU/Vector-Field-Transformer.

Oriented feature pyramid network for small and dense wheat heads detection and counting

Wheat head detection and counting using deep learning techniques has gained considerable attention in precision agriculture applications such as wheat growth monitoring, yield estimation, and resource allocation. However, the accurate detection of small and dense wheat heads remains challenging due to the inherent variations in their size, orientation, appearance, aspect ratios, density, and the complexity of imaging conditions. To address these challenges, we propose a novel approach called the Oriented Feature Pyramid Network (OFPN) that focuses on detecting rotated wheat heads by utilizing oriented bounding boxes. In order to facilitate the development and evaluation of our proposed method, we introduce a novel dataset named the Rotated Global Wheat Head Dataset (RGWHD). This dataset is constructed by manually annotating images from the Global Wheat Head Detection (GWHD) dataset with oriented bounding boxes. Furthermore, we incorporate a Path-aggregation and Balanced Feature Pyramid Network into our architecture to effectively extract both semantic and positional information from the input images. This is achieved by leveraging feature fusion techniques at multiple scales, enhancing the detection capabilities for small wheat heads. To improve the localization and detection accuracy of dense and overlapping wheat heads, we employ the Soft-NMS algorithm to filter the proposed bounding boxes. Experimental results indicate the superior performance of the OFPN model, achieving a remarkable mean average precision of 85.77% in oriented wheat head detection, surpassing six other state-of-the-art models. Moreover, we observe a substantial improvement in the accuracy of wheat head counting, with an accuracy of 93.97%. This represents an increase of 3.12% compared to the Faster R-CNN method. Both qualitative and quantitative results demonstrate the effectiveness of the proposed OFPN model in accurately localizing and counting wheat heads within various challenging scenarios.