Computer Vision Algorithms Research Articles

Development of an Algorithm for Semantic Segmentation of Earth Remote Sensing Data to Determine Phytoplankton Populations

Introduction. Computer vision is widely used for semantic segmentation of Earth remote sensing (ERS) data. The method allows monitoring ecosystems, including aquatic ones. Algorithms that maintain the quality of semantic segmentation of ERS images are in demand, specifically, to identify areas with phytoplankton, where water blooms— the cause of suffocation — are possible. The objective of the study is to create an algorithm that processes satellite data as input information for the formation and checking of mathematical models of hydrodynamics, which are used to monitor the state of water bodies. Various algorithms for semantic segmentation are described in the literature. New research focuses on enhancing the reliability of recognition — often using neural networks. This approach is modified in the presented work. To develop the direction, a new set of information from open sources and synthetic data are proposed. They are aimed at improving the generalization ability of the model. For the first time, the contour area of the phytoplankton population is compared to the database — and thus the boundary conditions are formed for the implementation of mathematical models and the construction of boundary-adaptive grids.Materials and Methods. The set of remote sensing images was supplemented with the author's augmentation algorithm in Python. Computer vision segmented areas of phytoplankton populations in the images. The U-Net convolutional neural network (CNN) was trained on the basis of NVIDIA Tesla T4 computing accelerators.Results. To automate the detection of phytoplankton distribution areas, a computer vision algorithm based on the U-Net CNN was developed. The model was evaluated by the calculated values of the main quality metrics related to segmentation tasks. The following metric values were obtained: Precision = 0.89, Recall = 0.88, F1 = 0.87, Dice = 0.87, and IoU = 0.79. Graphical visualization of the results of CNN learning on the training and validation sets showed good quality of model learning. This is evidenced by small changes in the loss function at the end of training. The segmentation performed by the model turned out to be close to manual marking, which indicated the high quality of the proposed solution. The area of the segmented region of the phytoplankton population was calculated by the area of one pixel. The result obtained for the original image was 51202.5 (based on information about the number of pixels related to the bloom of blue-green algae). The corresponding result of the modeling was 51312.Discussion and Conclusion. The study expands theoretical and practical knowledge on the use of convolutional neural networks for semantic segmentation of space imagery data. Given the results of the work, it is possible to assess the potential for automating the process of semantic segmentation of remote sensing data to determine the boundaries of phytoplankton populations using artificial intelligence. The use of the proposed computer vision model to obtain contours of water bloom due to phytoplankton will provide for the creation of databases — the basis for environmental monitoring of water resources and predictive modeling of hydrobiological processes.

EarMonitor: Non-clinical Assessment of Ear Health Conditions Using a Low-cost Endoscope Camera on Smartphones

Hearing loss affects 20% of the global population, a rate that is increasing dramatically as the world's population ages. Early prevention and identification of ear diseases can significantly reduce the risk of becoming disabled with hearing impairment. We propose EarMonitor, an interactive, vision-based ear health monitoring system that enables users to examine their ear conditions with a low-cost hand-held endoscope. EarMonitor can detect six ear health conditions suitable for self-assessment. It can particularly recognize complications from ear diseases, helping users better understand the results. In the wild, our computer vision algorithm achieves a detection sensitivity of 0.949 for earwax buildup and blockage in 100 external auditory canal photos; our deep learning model achieves an average detection sensitivity of 0.861 for the other five conditions considering complications in 350 tympanic membrane photos. We validated EarMonitor 's effectiveness through a user study involving 17 participants and two experts, leading to valuable insights regarding the design and interpretation of non-clinical assessment devices.

Multimodal Large Language Models as Built Environment Auditing Tools

This research showcases the transformative potential of large language models (LLMs) for built environment auditing from street-view images. By empirically testing the performances of two multimodal LLMs, ChatGPT and Gemini, we confirmed that LLM-based audits strongly agree with virtual audits processed by a conventional deep learning-based method (DeepLabv3+), which has been widely adopted by existing studies on urban visual analytics. Unlike conventional field or virtual audits that require labor-intensive manual inspection or technical expertise to run computer vision algorithms, our results show that LLMs can offer an intuitive tool despite the user’s level of technical proficiency. This would allow a broader range of policy and planning stakeholders to employ LLM-based built environment auditing instruments for smart urban infrastructure management.

Optimization of building thermal environment and VR industrial heritage landscape design enhanced by computer vision algorithms

With the acceleration of urbanization, the optimization of thermal environment of buildings has become an important topic to improve energy efficiency and environmental sustainability. This paper aims to enhance the optimization of building thermal environment through computer vision algorithm, and explore its application in VR industrial heritage landscape design, so as to achieve the dual goals of building energy conservation and cultural inheritance. In this study, advanced computer vision technology is used to monitor and analyze the thermal performance of buildings in real time. Through the construction of thermal environment model, combined with deep learning algorithm to extract building thermal energy data, and use VR technology for visual display, so as to provide scientific basis for design decisions. The research objects include several representative industrial heritage buildings, and analyze their thermal energy consumption characteristics and optimization potential. The results show that the application of the computer vision algorithm reduces the heat loss of the building and significantly improves the energy efficiency. At the same time, the VR visual model intuitively shows the design effect after thermal optimization, enhancing the communication and understanding between designers and stakeholders. Therefore, the combination of computer vision and VR technology can effectively improve the optimization ability of building thermal environment and promote the sustainable design and operation of traditional industrial heritage.

DFUCare: deep learning platform for diabetic foot ulcer detection, analysis, and monitoring.

Diabetic foot ulcers (DFUs) are a severe complication among diabetic patients, often leading to amputation or even death. Early detection of infection and ischemia is essential for improving healing outcomes, but current diagnostic methods are invasive, time-consuming, and costly. There is a need for non-invasive, efficient, and affordable solutions in diabetic foot care. We developed DFUCare, a platform that leverages computer vision and deep learning (DL) algorithms to localize, classify, and analyze DFUs non-invasively. The platform combines CIELAB and YCbCr color space segmentation with a pre-trained YOLOv5s algorithm for wound localization. Additionally, deep-learning models were implemented to classify infection and ischemia in DFUs. The preliminary performance of the platform was tested on wound images acquired using a cell phone. DFUCare achieved an F1-score of 0.80 and a mean Average Precision (mAP) of 0.861 for wound localization. For infection classification, we obtained a binary accuracy of 79.76%, while ischemic classification reached 94.81% on the validation set. The system successfully measured wound size and performed tissue color and textural analysis for a comparative assessment of macroscopic wound features. In clinical testing, DFUCare localized wounds and predicted infected and ischemic with an error rate of less than 10%, underscoring the strong performance of the platform. DFUCare presents an innovative approach to wound care, offering a cost-effective, remote, and convenient healthcare solution. By enabling non-invasive and accurate analysis of wounds using mobile devices, this platform has the potential to revolutionize diabetic foot care and improve clinical outcomes through early detection of infection and ischemia.

Rockfall mapping and susceptibility evaluation based on UAV high-resolution imagery and support vector machine method

Abstract This study aims to develop a comprehensive method for automated rockfall mapping and susceptibility assessment using unmanned aerial vehicle (UAV) tilt photography and the support vector machine (SVM) algorithm. By employing Jinzhong Town in the mountainous forest landscapes of Southwest China as a case study, we leverage photogrammetry principles and computer vision algorithms to generate high-precision, high-resolution digital surface models (DSMs), and digital orthophoto maps through a UAV remote sensing system. The rockfall inventory is accurately and automatically mapped using the object-based classification method and SVM algorithm. The automated rockfall identification method achieves a 93% accuracy with a Kappa coefficient of 0.7967. Statistical analyses of spatial distribution characteristics reveal a significant correlation between rockfall volume and area with a correlation coefficient (R²) of 0.92316 under logarithmic coordinates. In addition, a power function relationship describes the link between rockfall volume and slope height (R² = 0.87), while the relationship with sliding distance is characterized by a weaker linear correlation (R² = 0.65). Rockfall runout distance also shows a significant linear correlation with slope height (R² = 0.79) but exhibits a less-pronounced association with rockfall volume (R² = 0.58). The SVM model employed to assess rockfall susceptibility indicates high accuracy (area under the curve = 0.896), affirming its efficacy in rockfall susceptibility assessment. Our findings underscore the utility of UAV remote sensing for rockfall information extraction and susceptibility evaluation, particularly in challenging mountainous forest environments characterized by intricate topography and geological complexities.

A novel solar radiation forecasting model based on time series imaging and bidirectional long short‐term memory network

AbstractThe instability of solar energy is the biggest challenge to its successful integration with modern power grids, and accurate prediction of long‐term solar radiation can effectively solve this problem. In this study, we proposed a novel long‐term solar radiation prediction model based on time series imaging and bidirectional long short‐term memory network. First, inspired by the computer vision algorithm, the recursive graph algorithm is used to transform the one‐dimensional time series into two‐dimensional images, and then convolutional neural network is used to extract the features from the images, thus, the deeper features in the original solar radiation data can be mined. Second, to solve the problem of low accuracy of long‐term solar radiation prediction, a hybrid model BiLSTM‐Transformer is used to predict long‐term solar radiation. The hybrid prediction model can capture the long‐term dependencies, thereby further improving the accuracy of the prediction model. The experimental results show that the hybrid model proposed in this study is superior to other single models and hybrid models in long‐term solar radiation prediction accuracy. The accuracy and stability of the hybrid model are verified by many tests.

Estimating Power, Performance, and Area for On-Sensor Deployment of AR/VR Workloads Using an Analytical Framework

Augmented Reality and Virtual Reality have emerged as the next frontier of intelligent image sensors and computer systems. In these systems, 3D die stacking stands out as a compelling solution, enabling in situ processing capability of the sensory data for tasks such as image classification and object detection at low power, low latency, and a small form factor. These intelligent 3D CMOS Image Sensor (CIS) systems present a wide design space, encompassing multiple domains (e.g., computer vision algorithms, circuit design, system architecture, and semiconductor technology, including 3D stacking) that have not been explored in-depth so far. This article aims to fill this gap. We first present an analytical evaluation framework, STAR-3DSim, dedicated to rapid pre-RTL evaluation of 3D-CIS systems capturing the entire stack from the pixel layer to the on-sensor processor layer. With STAR-3DSim, we then propose several knobs for PPA (power, performance, area) improvement of the Deep Neural Network (DNN) accelerator that can provide up to 53%, 41%, and 63% reduction in energy, latency, and area, respectively, across a broad set of relevant AR/VR workloads. Last, we present full-system evaluation results by taking image sensing, cross-tier data transfer, and off-sensor communication into consideration.

PSII-4 How to define good mothering ability using computer vision of free farrowing sows?

Abstract In modern swine production, most of the sows are housed in a cage system in the farrowing unit. Although this system is intended to reduce preweaning mortality, the welfare of the sow might be compromised as it greatly restricts their natural behavior. Free farrowing systems allow sows to move around in the farrowing pen, which might stimulate their nesting behavior, shorten the farrowing duration and improve locomotion score, therefore, optimizing the longevity of the sows. However, the risk of increased preweaning mortality in free farrowing systems is greater compared with cage systems. Previous studies suggested that sow behavior, also known as mothering ability, can influence preweaning mortality as an increased level of posture changes could lead to increased mortality. With grunts, a sow could initiate the suckling of the piglets but also terminate it by hiding her udder from the piglets. However, there is conflicting evidence on what characteristics of maternal ability are correlated with piglet mortality, and most studies relied on limited manual observations, which might be biased. Using computer vision algorithms, the behavior and the movement pattern of a large number of sows could be monitored at a high density and on a continuous basis. Hence, this study intended to investigate the mothering ability of sows in free farrowing systems based on large quantities and unbiased observations using computer vision. Forty cameras were installed above free farrowing pens to observe the behavior of the sow. The dataset is continuously growing and will add data of 8 sows every week, totaling 416 sows on a yearly basis. The performance of the sows and their piglets are recorded, including body weight (BW), backfat, total number born, number weaned, birth weight, and weaning weight. Computer vision algorithms (YOLO) will be trained to estimate the contour and posture of the sow and the position of the piglets. Preliminary results based on a small subset of animals showed that the algorithm is able to detect the sows and their postures accurately, and it is now being extended to detect piglet locations. This will allow us to find beneficial behavior of sow and piglets, estimate genetic parameters for potential implementation in breeding programs, and eventually improve sow performance in a welfare-friendly free farrowing system.

A computational framework for aesthetic preferences in architecture using computer vision and artificial neural networks

This paper introduces a computational aesthetics framework utilizing computer vision (CV) and artificial neural networks (ANN) to predict the aesthetic preferences of groups of people for architecture. It relies on part-to-whole theories from aesthetics and cognitive psychology. A survey of a group of people on preferences of images is held to record an average hedonic response (AHR). CV algorithms MSER and SAM recognize parts in images. Birkhoff’s aesthetic measure formula is adapted by employing the number of parts and their connections. These quantities are used as input layers of an ANN, and the AHR is the target output. The ANN evaluates images to output a predicted hedonic response (PHR), which is tested as a criterion in parametric design space navigation and in mapping the latent space of GANs. We conclude that such a framework is a heuristic method for better understanding the design and latent spaces and exploring designs.

AnnotateXR: An Extended Reality Workflow for Automating Data Annotation to Support Computer Vision Applications

Abstract Computer vision (CV) algorithms require large annotated datasets that are often labor-intensive and expensive to create. We propose AnnotateXR, an extended reality (XR) workflow to collect various high-fidelity data and auto-annotate it in a single demonstration. AnnotateXR allows users to align virtual models over physical objects, tracked with six degrees-of-freedom (6DOF) sensors. AnnotateXR utilizes a hand tracking capable XR head-mounted display coupled with 6DOF information and collision detection to enable algorithmic segmentation of different actions in videos through its digital twin. The virtual–physical mapping provides a tight bounding volume to generate semantic segmentation masks for the captured image data. Alongside supporting object and action segmentation, we also support other dimensions of annotation required by modern CV, such as human–object, object–object, and rich 3D recordings, all with a single demonstration. Our user study shows AnnotateXR produced over 112,000 annotated data points in 67 min.

Automated fish counting system based on instance segmentation in aquaculture

Non-invasive methods of fish counting are crucial for the effective management of aquaculture and aquatic resources. However, accurate and reliable estimation of fish populations in real-time is a significant challenge due to severe occlusion, body deformation, and rapid movement of fish. Recent advancements in computer vision and deep learning algorithms offer substantial potential to revolutionize fish counting methodologies, thereby enhancing conservation efforts and promoting the sustainable management of aquatic resources. This paper presents a comprehensive dataset of 256,580 fish instances annotated with mask information at the pixel level. To address the challenges associated with geometric variations in fish size, pose, perspective, and body shape, we propose a lightweight instance segmentation model based on YOLOv8, which integrates the CSP bottleneck with dual convolutions and the deformable convolutional networks (DConv). In addition, an efficient feature fusion network incorporating multi-scale information was introduced to improve segmentation accuracy by enhancing the representation of feature information, and a decoupled head with attention mechanisms was designed to detect and segment fish bodies in real time. Our proposed approach achieved an outstanding mAP50 of 98.1 % at 60.6 FPS, with model parameters and floating point operations per second (FLOPS) amounting to 25.4 M and 96.9 G, respectively. The instance segmentation-based counting system exhibited impressive performance metrics, with Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and R-Square (R2) values of 1.19, 1.65, and 0.992, respectively. The proposed method demonstrates strong potential for practical application as an intelligent solution for biomass estimation in recirculating aquaculture systems.

Enhancing concrete structure maintenance through automated crack detection: A computer vision approach

This paper presents the development of an Artificial Intelligence (AI) and Machine Learning (ML) model designed to detect cracks on concrete surfaces. The objective is to enhance the automation, precision, and performance of crack detection using the computer vision algorithm. Employing a ML approach and the YOLOv9 algorithm, this study developed a system to accurately identify concrete cracks from a diverse dataset. A total of 16,301 images of concrete surfaces, balanced between those with and without cracks, were utilized. The dataset was split into various sets with different ratios to ensure comprehensive model training. A transfer-learning methodology was employed to optimize the model's performance. The accuracy of the model was measured in each experiment to determine the optimal result. The most successful experiment resulted in a model with a mean Average Precision (mAP) of 94.6%, a Precision of 94.1%, and a Recall of 88.4%. These results demonstrate the effectiveness of AI and ML in concrete crack detection.

Review of Modern Forest Fire Detection Techniques: Innovations in Image Processing and Deep Learning

Fire detection and extinguishing systems are critical for safeguarding lives and minimizing property damage. These systems are especially vital in combating forest fires. In recent years, several forest fires have set records for their size, duration, and level of destruction. Traditional fire detection methods, such as smoke and heat sensors, have limitations, prompting the development of innovative approaches using advanced technologies. Utilizing image processing, computer vision, and deep learning algorithms, we can now detect fires with exceptional accuracy and respond promptly to mitigate their impact. In this article, we conduct a comprehensive review of articles from 2013 to 2023, exploring how these technologies are applied in fire detection and extinguishing. We delve into modern techniques enabling real-time analysis of the visual data captured by cameras or satellites, facilitating the detection of smoke, flames, and other fire-related cues. Furthermore, we explore the utilization of deep learning and machine learning in training intelligent algorithms to recognize fire patterns and features. Through a comprehensive examination of current research and development, this review aims to provide insights into the potential and future directions of fire detection and extinguishing using image processing, computer vision, and deep learning.

Health Warnings on Instagram Advertisements for Synthetic Nicotine E-Cigarettes and Engagement

Synthetic nicotine is increasingly used in e-cigarette liquids along with flavors to appeal to youths. Regulatory loopholes have allowed tobacco manufacturers to use social media to target youths. To analyze the extent to which synthetic nicotine e-cigarette brands have implemented US Food and Drug Administration (FDA) health warning requirements and to evaluate the association between health warnings and user engagement on Instagram. In this cross-sectional study, posts from 25 brands were analyzed across a 14-month period (August 2021 to October 2022). A content analysis was paired with Warning Label Multi-Layer Image Identification, a computer vision algorithm designed to detect the presence of health warnings and whether the detected health warning complied with FDA guidelines by (1) appearing on the upper portion of the advertisement and (2) occupying at least 20% of the advertisement's area. Data analysis was performed from March to June 2024. Synthetic nicotine e-cigarette advertisement on Instagram. The outcome variables were user engagement (number of likes and comments). Negative binomial regression analyses were used to evaluate the association between the presence and characteristics of health warnings and user engagement. Of a total of 2071 posts, only 263 (13%) complied with both FDA health warning requirements. Among 924 posts with health warnings, 732 (79%) displayed warnings in the upper image portion, and 270 (29%) had a warning covering at least 20% of the pixel area. Posts with warnings received fewer comments than posts without warnings (mean [SD], 1.8 [2.5] vs 5.4 [11.7] comments; adjusted incident rate ratio [aIRR], 0.70; 95% CI, 0.57-0.86; P < .001). For posts containing warnings, a larger percentage of the warning label's pixel area was associated with fewer comments (aIRR, 0.96; 95% CI, 0.93-0.99; P = .003). Flavored posts with health warnings placed in the upper image portion received more likes than posts with warnings in the lower portion (mean [SD], 34.6 [35.2] vs 19.9 [19.2] likes; aIRR, 1.48; 95% CI, 1.07-2.06; P = .02). In this cross-sectional study of synthetic nicotine brand Instagram accounts, 87% of sampled posts did not adhere to FDA health warning requirements in tobacco promotions. Enforcement of FDA compliant health warnings on social media may reduce youth engagement with tobacco marketing.

Nitroglycerin challenge identifies microcirculatory target for improved resuscitation in patients with circulatory shock

BackgroundCirculatory shock and multi-organ failure remain major contributors to morbidity and mortality in critically ill patients and are associated with insufficient oxygen availability in the tissue. Intrinsic mechanisms to improve tissue perfusion, such as up-regulation of functional capillary density (FCD) and red blood cell velocity (RBCv), have been identified as maneuvers to improve oxygen extraction by the tissues; however, their role in circulatory shock and potential use as resuscitation targets remains unknown. To fill this gap, we examined the baseline and maximum recruitable FCD and RBCv in response to a topical nitroglycerin stimulus (FCDNG, RBCvNG) in patients with and without circulatory shock to test whether this may be a method to identify the presence and magnitude of a microcirculatory reserve capacity important for identifying a resuscitation target.MethodsSublingual handheld vital microscopy was performed after initial resuscitation in mechanically ventilated patients consecutively admitted to a tertiary medical ICU. FCD and RBCv were quantified using an automated computer vision algorithm (MicroTools). Patients with circulatory shock were retrospectively identified via standardized hemodynamic and clinical criteria and compared to patients without circulatory shock.Results54 patients (57 ± 14y, BMI 26.3 ± 4.9 kg/m2, SAPS 56 ± 19, 65% male) were included, 13 of whom presented with circulatory shock. Both groups had similar cardiac index, mean arterial pressure, RBCv, and RBCvNG. Heart rate (p < 0.001), central venous pressure (p = 0.02), lactate (p < 0.001), capillary refill time (p < 0.01), and Mottling score (p < 0.001) were higher in circulatory shock after initial resuscitation, while FCD and FCDNG were 10% lower (16.9 ± 4.2 and 18.9 ± 3.2, p < 0.01; 19.3 ± 3.1 and 21.3 ± 2.9, p = 0.03). Nitroglycerin response was similar in both groups, and circulatory shock patients reached FCDNG similar to baseline FCD found in patients without shock.ConclusionCritically ill patients suffering from circulatory shock were found to present with a lower sublingual FCD. The preserved nitroglycerin response suggests a dysfunction of intrinsic regulation mechanisms to increase the microcirculatory oxygen extraction capacity associated with circulatory shock and identifies a potential resuscitation target. These differences in microcirculatory hemodynamic function between patients with and without circulatory shock were not reflected in blood pressure or cardiac index.

A Review of Vision-Based Pothole Detection Methods Using Computer Vision and Machine Learning.

Potholes and other road surface damages pose significant risks to vehicles and traffic safety. The current methods of in situ visual inspection for potholes or cracks are inefficient, costly, and hazardous. Therefore, there is a pressing need to develop automated systems for assessing road surface conditions, aiming to efficiently and accurately reconstruct, recognize, and locate potholes. In recent years, various methods utilizing (a) computer vision, (b) three-dimensional (3D) point clouds, or (c) smartphone data have been employed to map road surface quality conditions. Machine learning and deep learning techniques have increasingly enhanced the performance of these methods. This review aims to provide a comprehensive overview of cutting-edge computer vision and machine learning algorithms for pothole detection. It covers topics such as sensing systems for acquiring two-dimensional (2D) and 3D road data, classical algorithms based on 2D image processing, segmentation-based algorithms using 3D point cloud modeling, machine learning, deep learning algorithms, and hybrid approaches. The review highlights that hybrid methods combining traditional image processing and advanced machine learning techniques offer the highest accuracy in pothole detection. Machine learning approaches, particularly deep learning, demonstrate superior adaptability and detection rates, while traditional 2D and 3D methods provide valuable baseline techniques. By reviewing and evaluating existing vision-based methods, this paper clarifies the current landscape of pothole detection technologies and identifies opportunities for future research and development. Additionally, insights provided by this review can inform the design and implementation of more robust and effective systems for automated road surface condition assessment, thereby contributing to enhanced roadway safety and infrastructure management.

Artificial Intelligence Technology on Layered Water Injection in Oilfield Development Process

Since the 1960s, researchers have worked to advance water flooding technology to address challenges like high viscosity, low fluidity, and depleting reservoirs, aiming to prevent oil fields from becoming unproductive. The integration of artificial intelligence (AI), computer vision, and advanced algorithms like BP neural networks has recently revolutionized this field. These technological advancements have upgraded water injection methodologies, overcoming past limitations and enabling real-time monitoring and dynamic control of water injection into different oil layers. This 'intelligent layering' ensures optimized water management, enhancing overall recovery rates. This overview highlights the progression of water injection techniques, critiques traditional methods' shortcomings, and delves into the cutting-edge applications of AI-driven intelligent layering systems. It serves as a valuable guide for oil industry stakeholders, equipment manufacturers, and research institutions seeking to refine water injection practices and boost hydrocarbon extraction efficiency.

A Computer Vision Algorithm to Predict Superior Mesenteric Artery Margin Status for Patients with Pancreatic Ductal Adenocarcinoma.

To evaluate the feasibility of developing a computer vision algorithm that uses preoperative computed tomography (CT) scans to predict superior mesenteric artery (SMA) margin status in patients undergoing Whipple for pancreatic ductal adenocarcinoma (PDAC), and to compare algorithm performance to that of expert abdominal radiologists and surgical oncologists. Complete surgical resection is the only chance to achieve a cure for PDAC; however, current modalities to predict vascular invasion have limited accuracy. Adult patients with PDAC who underwent Whipple and had preoperative contrast-enhanced CT scans were included (2010-2022). The SMA was manually annotated on the CT scans, and we trained a U-Net algorithm for SMA segmentation and a ResNet50 algorithm for predicting SMA margin status. Radiologists and surgeons reviewed the scans in a blinded fashion. SMA margin status per pathology reports was the reference. Two hundred patients were included. Forty patients (20%) had a positive SMA margin. For the segmentation task, the U-Net model achieved a Dice Similarity Coefficient of 0.90. For the classification task, all readers demonstrated limited sensitivity, although the algorithm had the highest sensitivity at 0.43 (versus 0.23 and 0.36 for the radiologists and surgeons, respectively). Specificity was universally excellent, with the radiologist and algorithm demonstrating the highest specificity at 0.94. Finally, the accuracy of the algorithm was 0.85 versus 0.80 and 0.76 for the radiologists and surgeons, respectively. We demonstrated the feasibility of developing a computer vision algorithm to predict SMA margin status using preoperative CT scans, highlighting its potential to augment the prediction of vascular involvement.

AI-based detection and identification of low-level nuclear waste: a comparative analysis

Ensuring environmental safety and regulatory compliance at Department of Energy (DOE) sites demands an efficient and reliable detection system for low-level nuclear waste (LLW). Unlike existing methods that rely on human effort, this paper explores the integration of computer vision algorithms to automate the identification of such waste across DOE facilities. We evaluate the effectiveness of multiple algorithms in classifying nuclear waste materials and their adaptability to newly emerging LLW. Our research introduces and implements five state-of-the-art computer vision models, each representing a different approach to the problem. Through rigorous experimentation and validation, we evaluate these algorithms based on performance, speed, and adaptability. The results reveal a noteworthy trade-off between detection performance and adaptability. YOLOv7 shows the best performance and requires the highest effort to detect new LLW. Conversely, OWL-ViT has lower performance than YOLOv7 and requires minimal effort to detect new LLW. The inference speed does not strongly correlate with performance or adaptability. These findings offer valuable insights into the strengths and limitations of current computer vision algorithms for LLW detection. Each developed model provides a specialized solution with distinct advantages and disadvantages, empowering DOE stakeholders to select the algorithm that aligns best with their specific needs.