Pixel Intensity Research Articles

Continuous distribution of cancer cells in the cell cycle unveiled by AI-segmented imaging of 37,000 HeLa FUCCI cells

Classification of live or fixed cells based on their unlabeled microscopic images would be a powerful tool for cell biology and pathology. For such software, the first step is the generation of a ground truth database that can be used for training and testing AI classification algorithms. The Application of cells expressing fluorescent reporter proteins allows the building of ground truth datasets in a straightforward way. In this study, we present an automated imaging pipeline utilizing the Cellpose algorithm for the precise cell segmentation and measurement of fluorescent cellular intensities across multiple channels. We analyzed the cell cycle of HeLa–FUCCI cells expressing fluorescent red and green reporter proteins at various levels depending on the cell cycle state. To build the dataset, 37,000 fixed cells were automatically scanned using a standard motorized microscope, capturing phase contrast and fluorescent red/green images. The fluorescent pixel intensity of each cell was integrated to calculate the total fluorescence of cells based on cell segmentation in the phase contrast channel. It resulted in a precise intensity value for each cell in both channels. Furthermore, we conducted a comparative analysis of Cellpose 1.0 and Cellpose 2.0 in cell segmentation performance. Cellpose 2.0 demonstrated notable improvements, achieving a significantly reduced false positive rate of 2.7 % and 1.4 % false negative. The cellular fluorescence was visualized in a 2D plot (map) based on the red and green intensities of the FUCCI construct revealing the continuous distribution of cells in the cell cycle. This 2D map enables the selection and potential isolation of single cells in a specific phase. In the corresponding heatmap, two clusters appeared representing cells in the red and green states. Our pipeline allows the high-throughput and accurate measurement of cellular fluorescence providing extensive statistical information on thousands of cells with potential applications in developmental and cancer biology. Furthermore, our method can be used to build ground truth datasets automatically for training and testing AI cell classification. Our automated pipeline can be used to analyze thousands of cells within 2 h after putting the sample onto the microscope.

Convolutional neural network for improved event-based Shack-Hartmann wavefront reconstruction

Shack-Hartmann wavefront sensing is a technique for measuring wavefront aberrations, whose use in adaptive optics relies on fast position tracking of an array of spots. These sensors conventionally use frame-based cameras operating at a fixed sampling rate to report pixel intensities, even though only a fraction of the pixels have signal. Prior in-lab experiments have shown feasibility of event-based cameras for Shack-Hartmann wavefront sensing (SHWFS), asynchronously reporting the spot locations as log intensity changes at a microsecond time scale. In our work, we propose a convolutional neural network (CNN) called event-based wavefront network (EBWFNet) that achieves highly accurate estimation of the spot centroid position in real time. We developed a custom Shack-Hartmann wavefront sensing hardware with a common aperture for the synchronized frame- and event-based cameras so that spot centroid locations computed from the frame-based camera may be used to train/test the event-CNN-based centroid position estimation method in an unsupervised manner. Field testing with this hardware allows us to conclude that the proposed EBWFNet achieves sub-pixel accuracy in real-world scenarios with substantial improvement over the state-of-the-art event-based SHWFS. An ablation study reveals the impact of data processing, CNN components, and training cost function; and an unoptimized MATLAB implementation is shown to run faster than 800 Hz on a single GPU.

Deep learning-based optimization of field geometry for total marrow irradiation delivered with volumetric modulated arc therapy.

Total marrow (lymphoid) irradiation (TMI/TMLI) is a radiotherapy treatment used to selectively target the bone marrow and lymph nodes in conditioning regimens for allogeneic hematopoietic stem cell transplantation. A complex field geometry is needed to cover the large planning target volume (PTV) of TMI/TMLI with volumetric modulated arc therapy (VMAT). Five isocenters and ten overlapping fields are needed for the upper body, while, for patients with large anatomical conformation, two specific isocenters are placed on the arms. The creation of a field geometry is clinically challenging and is performed by a medical physicist (MP) specialized in TMI/TMLI. To develop convolutional neural networks (CNNs) for automatically generating the field geometry of TMI/TMLI. The dataset comprised 117 patients treated with TMI/TMLI between 2011 and 2023 at our Institute. The CNN input image consisted of three channels, obtained by projecting along the sagittal plane: (1) average CT pixel intensity within the PTV; (2) PTV mask; (3) brain, lungs, liver, bowel, and bladder masks. This "averaged" frontal view combined the information analyzed by the MP when setting the field geometry in the treatment planning system (TPS). Two CNNs were trained to predict the isocenters coordinates and jaws apertures for patients with (CNN-1) and without (CNN-2) isocenters on the arms. Local optimization methods were used to refine the models output based on the anatomy of the patient. Model evaluation was performed on a test set of 15 patients in two ways: (1) by computing the root mean squared error (RMSE) between the CNN output and ground truth; (2) with a qualitative assessment of manual and generated field geometries-scale: 1=not adequate, 4=adequate-carried out in blind mode by three MPs with different expertise in TMI/TMLI. The Wilcoxon signed-rank test was used to evaluate the independence of the given scores between manual and generated configurations (p<0.05 significant). The average and standard deviation values of RMSE for CNN-1 and CNN-2 before/after local optimization were 15±2/13±3mm and 16±2/18±4mm, respectively. The CNNs were integrated into a planning automation software for TMI/TMLI such that the MPs could analyze in detail the proposed field geometries directly in the TPS. The selection of the CNN model to create the field geometry was based on the PTV width to approximate the decision process of an experienced MP and provide a single option of field configuration. We found no significant differences between the manual and generated field geometries for any MP, with median values of 4versus 4 (p=0.92), 3versus 3 (p=0.78), 4versus 3 (p=0.48), respectively. Starting from October 2023, the generated field geometry has been introduced in our clinical practice for prospective patients. The generated field geometries were clinically acceptable and adequate, even for an MP with high level of expertise in TMI/TMLI. Incorporating the knowledge of the MPs into the development cycle was crucial for optimizing the models, especially in this scenario with limited data.

Gray level fuzzy deep neural networks for enhancing performance in lung disease detection: A comparative study with fuzzy logic methods

AbstractLung cancer is a deadly disease, and its early detection is crucial for effective treatment. In this context, accurate classification of lung cancers from computed tomography imaging is a vital research area. Irregularly detected gray matter in these images can affect classification outcomes, making accurate lung cancer detection difficult. To address this issue, researchers have developed a new approach that combines fuzzy logic and deep neural networks (DNNs) for extracting the hidden characteristic features of lung diseases from the images, and thereby, ensuring that only relevant features are used for classification. The proposed gray‐level fuzzy approach on DNNs (GL‐FDNN) is designed to accurately classify four distinguished lung cancer classes namely, Large cell carcinoma, Adenocarcinoma, Normal lung computed tomography, and Squamous cell carcinoma. After identifying the area of interest, its pixel intensity ratio is used to derive the fuzzy logic, which is then used to extract the most obscure gray matter features. Then, ResNet‐18 sorts the obscure features into categories and picks only relevant features to improve classification accuracy. The gray‐level fuzzy approach on DNNs technique was tested alongside some of the renowned, existing techniques for their relative performances. Experimental analysis was carried out on standard Kaggle datasets and the outcomes reveal that the proposed technique offers the highest level of lung disease classification accuracy (99.2%). It also improves the recall and precision factors. Thus, it can serve as a valuable diagnosis tool that can enhance the detection of lung cancer and the effectiveness of its treatment to save lives.

Comparison of Scheimpflug Corneal Tomography and Anterior Segment Optical Coherence Tomography Measurements in Corneal Cystinosis: A Case Series.

To report the clinical course and compare the utility of Scheimpflug tomography (ST) and anterior segment optical coherence tomography (AS-OCT) for central corneal thickness (CCT) and corneal densitometry (CD) assessment in patients with corneal crystals owing to nephropathic cystinosis. A retrospective chart analysis of three patients with nephropathic cystinosis and the presence of corneal cystine crystals in both eyes was performed. All patients underwent clinical examination and anterior segment photography, ST, and AS-OCT scans. Corneal densitometry was exported from built-in proprietary software for ST and from custom-made validated software for AS-OCT. Anterior segment optical coherence tomography images were rescaled to grayscale units from 0 (maximum transparency) to 100 (minimum transparency) to match built-in ST densitometry readings. Furthermore, the mean pixel intensity, representative of CD, was calculated from the pixels corresponding to the segmented cornea. All three patients had pathognomonic cystine crystals deposits in the cornea and were treated with cysteamine medications that resulted in clinical improvement. The CCT measured using ST exhibited a range from 560 to 958 μm. Conversely, when assessed with AS-OCT, the CCT varied within the range of 548 to 610 μm. Both examinations could be performed, but in the more severe cases, AS-OCT showed far greater utility to estimate CD. In four of six eyes examined, ST showed disproportionate CCT values, compared with the AS-OCT, whereas reliable CD measurements were only available in AS-OCT. The AS-OCT could be considered a baseline ocular measurement in cystinosis and in the evaluation of disease progression and treatment efficacy.

Visualized Experimental Study of Soil Temperature Distribution around Submarine Buried Offshore Pipeline Based on Transparent Soil

The temperature distribution around the offshore burial pipeline is an important factor affecting its safety design and economic operation. The traditional test method cannot obtain the continuous temperature distribution of soil owing to the constraints of placing measurement sensors in soil. The transparent soil model test is an alternative method to realize the visualization research of soil temperature. In this paper, a relationship between the temperature of transparent soil and pixel intensity was first established. Then, the transparent soil test and numerical simulation, considering the natural convection, were carried out to study the temperature distribution around the submarine pipeline during start-up and stable operation. The influence of buried depth and pipeline diameter was analyzed. The results suggest that the continuous temperature distribution can be obtained visually by using a transparent soil test, and the observed heating zone of influence extended to a radial distance of 2.6 pipe diameters. The numerical analysis results show that the influence zone of the temperature of pipeline is a distance of four pipeline diameters at a temperature difference of 45 °C. The buried depth and pipeline diameter have little influence on the influence zone. In addition, the contour curves of soil temperature around the pipeline with different diameter are similar in shape. With the decrease in the buried depth of pipeline, the temperature gradient of soil around the pipeline decreases, which is caused by the natural convection.

Nucleus segmentation of white blood cells in blood smear images by modeling the pixels' intensities as a set of three Gaussian distributions.

The precise segmentation of white blood cells (WBCs) within blood smear images is a significant challenge with implications for both medical research and image processing. Of particular importance is the often neglected task of accurately segmenting WBC nuclei, an aspect that currently lacks dedicated methodologies. This paper introduces a straightforward and efficient method designed to fill this critical gap, providing an effective solution for the efficient segmentation of WBC nuclei. In blood smear imagery, the distinctive coloration of WBCs contrasts with the hues of other blood components. The inherent obscurity of WBCs prompts their segmentation by isolating pixels with minimal intensities. To streamline this process, our proposed method employs the Laplacian pyramid technique to decorrelate pixels in blood smear images, thereby amplifying the contrast. Subsequently, the intensities of pixels constituting blood cells, encompassing WBCs and the background, are modeled using three Gaussian random variables. Capitalizing on this feature, we implement the Gaussian mixture model (GMM) clustering method to determine the optimal threshold value, facilitating a highly precise segmentation of WBC nuclei. The proposed method demonstrates the capability to process images containing a single WBC as well as effectively functioning with images containing multiple cells of this type. Evaluation of the method on the ALL-IDB, ALL-IDB2, CellaVision, and JTSC datasets yielded accuracy values of 0.9802, 0.9725, 0.9772, and 0.9730, respectively. Comparative analysis with state-of-the-art methods revealed a notably comparable performance, underscoring the effectiveness of the proposed approach. The method presented in this article is highly competitive for segmenting the nuclei of WBCs compared to state-of-the-art methods. The three main advantages of our method are its ability to process images containing one or more WBCs, the automatic calculation of threshold values for each processed image, eliminating the need for manual parameter adjustments. Lastly, the method is efficient, as its algorithmic complexity is approximately .

Infrared Thermographic Imaging as a Tool to Assess Inflammatory and Ischemic Response With Rubber Ring Tail Docking in Suffolk and Hampshire Lambs (Ovis Aries)

ABSTRACT Tail docking in lambs is associated with an acute inflammatory and ischemic response correlated to pain. The present study sought to quantify these responses by utilizing infrared thermographic imaging (IRT) after tail docking lambs. Images were taken for each lamb before tail docking (T0), within a minute after the band was placed (T1), at 5 minutes (T5), and 15 minutes (T15) respectively and analyzed for pixel intensity in areas on the lamb’s rump, tail base, and two locations below the docking band. The pixel intensity at 2 cm below the docking band did significantly decrease (p < 0.05) over the course of the study at 15 minutes, indicating the presence of ischemia, and a significant decrease in the pixel intensity of the rump and tail base was observed. The closest location to the rubber ring showed the presence of inflammation as an indirect indicator of acute pain . IRT is a valuable tool in the detection of inflammation and ischemic areas and has the potential to be used to quantify and measure the experience of inflammation and pain after tail docking.

Muscle Ultrasound Echo Intensity and Fiber Type Composition in Young Females.

Ultrasonography has been extensively used to evaluate skeletal muscle morphology. The echo intensity, i.e., the mean pixel intensity of a specific region of interest in an ultrasound image, may vary among muscles and individuals with several intramuscular parameters presumed to influence it. The purpose of this study was to investigate the correlation between muscle echo intensity and muscle fiber type composition in humans. Thirteen female physical education students (age: 22.3 ± 5.4 years, height: 1.63 ± 0.06 m, body mass: 59.9 ± 7.4 kg) with no history of systematic athletic training participated in the study. Body composition with dual X-ray absorptiometry, leg-press maximum strength (1-RM), echo intensity, and the cross-sectional area (CSA) of the vastus lateralis (VL) muscle according to ultrasonography were measured. Muscle biopsies were harvested from the VL site where the echo intensity was measured. VL echo intensity was not significantly correlated with the percentage of type I muscle fibers or with the percentage area of type I muscle fibers. However, when VL echo intensity was corrected for the subcutaneous fat thickness at the site of the measurement, it was significantly correlated with the percentage of type I muscle fibers (r = 0.801, p < 0.01) and the percentage area of type I muscle fibers (r = 0.852, p < 0.01). These results suggest that the echo intensity of the vastus lateralis muscle corrected for the subcutaneous fat thickness at the measurement site may provide an estimate of the muscle fiber type composition, at least in young moderately trained females.

Intelligent MRI diagnosis of neurological alterations in infants from 4 to 12 months

AbstractMagnetic resonance imaging is an essential tool for the identification of neurological problems since it provides relevant information on brain development. The aim of the present work was the detection of neurological alterations in newborns from 4 to 12 months of age by segmentation and analysis of lateral ventricles in magnetic resonance images. For this purpose, an automated deep approach based on U‐net is proposed to segment the cerebral ventricles of the newborn. Subsequently, for these regions, features were extracted based on the patient's clinical history and on the shape (area, roundness, normalized central moment, among others) and pixel intensity (mean gray value, contrast level, among others). Once the features were extracted, different types of intelligent models (Logistic Regression, k‐Nearest Neighbors (kNN), and a Convolutional Neural Network) were assessed to detect the presence of neurological alterations. The segmentation phase of the system was tested on 50 patients and the classification phase on 28 patients (11 healthy, 17 with neurological changes). The results show a DICE similarity coefficient of 0.89 and a volume ratio of 1.05 for the segmentation stage and an accuracy of 98%, precision of 100%, sensitivity of 92%, and specificity of 100% for the classification stage using kNN. The last one proved to be the most computationally feasible model, due to the time required for training and inference (0.36 s and 35.2e‐4 s, respectively), as well as the consumption of computational resources (0.1 GB RAM CPU). In conclusion, it is possible to detect neurological alterations in newborns aged 4 to 12 months by segmenting and classifying the lateral ventricles in magnetic resonance images, using image processing techniques, the U‐net, as well as the kNN algorithm. This proposed methodology could play an important role in the early diagnosis and treatment of neurological disorders.

Establishing a reference focal plane using convolutional neural networks and beads for brightfield imaging

Repeatability of measurements from image analytics is difficult, due to the heterogeneity and complexity of cell samples, exact microscope stage positioning, and slide thickness. We present a method to define and use a reference focal plane that provides repeatable measurements with very high accuracy, by relying on control beads as reference material and a convolutional neural network focused on the control bead images. Previously we defined a reference effective focal plane (REFP) based on the image gradient of bead edges and three specific bead image features. This paper both generalizes and improves on this previous work. First, we refine the definition of the REFP by fitting a cubic spline to describe the relationship between the distance from a bead’s center and pixel intensity and by sharing information across experiments, exposures, and fields of view. Second, we remove our reliance on image features that behave differently from one instrument to another. Instead, we apply a convolutional regression neural network (ResNet 18) trained on cropped bead images that is generalizable to multiple microscopes. Our ResNet 18 network predicts the location of the REFP with only a single inferenced image acquisition that can be taken across a wide range of focal planes and exposure times. We illustrate the different strategies and hyperparameter optimization of the ResNet 18 to achieve a high prediction accuracy with an uncertainty for every image tested coming within the microscope repeatability measure of 7.5 µm from the desired focal plane. We demonstrate the generalizability of this methodology by applying it to two different optical systems and show that this level of accuracy can be achieved using only 6 beads per image.

Infrared and Visual Image Fusion Based on a Local-Extrema-Driven Image Filter.

The objective of infrared and visual image fusion is to amalgamate the salient and complementary features of the infrared and visual images into a singular informative image. To accomplish this, we introduce a novel local-extrema-driven image filter designed to effectively smooth images by reconstructing pixel intensities based on their local extrema. This filter is iteratively applied to the input infrared and visual images, extracting multiple scales of bright and dark feature maps from the differences between continuously filtered images. Subsequently, the bright and dark feature maps of the infrared and visual images at each scale are fused using elementwise-maximum and elementwise-minimum strategies, respectively. The two base images, representing the final-scale smoothed images of the infrared and visual images, are fused using a novel structural similarity- and intensity-based strategy. Finally, our fusion image can be straightforwardly produced by combining the fused bright feature map, dark feature map, and base image together. Rigorous experimentation conducted on the widely used TNO dataset underscores the superiority of our method in fusing infrared and visual images. Our approach consistently performs on par or surpasses eleven state-of-the-art image-fusion methods, showcasing compelling results in both qualitative and quantitative assessments.

Reflectance characterization of Starlink-like satellite with HELIOS brightness testing facility

The HELIOS (High Efficiency Light Intensity Observation System) project addresses the issue of excessive brightness from LEO satellite constellations. The system aims to develop a reflectance model of a satellite prior to launch, enabling the validation of compliance with astronomer guidelines, which suggest a minimum magnitude of 7. This ensures that the constellation is not visible to the naked eye and does not excessively interfere with astronomical observations. Additionally, the reflectance model obtained can be used for attitude determination by comparing the light curves with those obtained from observations. This application extends to Space Traffic Management, debris reentry prediction, and active debris removal. The system utilizes a light source consisting of 14 high-power LEDs with different wavelengths that replicate the solar spectrum. The light is positioned on an arc and can slide along a rail, allowing sampling at various angles. The arc itself can rotate on its axis, enabling the placement of the light source at any point on the hemisphere. At the center of the hemisphere, a scaled 3D printed model of the satellite under study is placed, with surface finishes resembling the actual satellite. Creating models with varying levels of detail permits to assess their impact on the reflectance model. Finally, the robotic arm Arm4NDO (Arm For Near Distance Observations) permits the movement of a high-efficiency quantum camera mounted at its end across the entire hemisphere’s surface for data acquisition. The camera records a video, stopping at points of interest for capturing frames. To strike a balance between continuous data acquisition and reasonable testing times, both the light source and capture points are positioned every 15° in elevation and azimuth. The initial test campaign aimed to determine optimal camera settings (exposure, aperture, and ISO) to achieve good model visibility and a dark background. Preliminary tests focused on Starlink’s chassis and the most significant observation configurations. Frames of interest were extracted from the videos, and their luminous flux (sum of pixel intensity values) was derived. With the obtained data, curves normalized relative to the maximum were plotted, allowing for a comparison of flux variations among different configurations.

Data-driven diagnostics of boiling heat transfer on flat heaters from non-intrusive visualization

Rankine-based power generation cycles require the use of boiling heat exchangers for liquid–vapor phase change. However, operating boilers at too high heat fluxes can lead to heater meltdown due to onset of the boiling crisis. Despite recent advances in increasing this critical heat flux, the lack of fast and accurate diagnostics hinders boiling heat transfer safety at high heat fluxes. Here we show that neural networks fed on direct observation of bubble dynamics can be used to measure heat flux during nucleate boiling. The boiling heat transfer data was obtained with a horizontally oriented, circular flat heater with saturated water as the heat transfer fluid. A deep convolutional neural network (CNN) was trained using frames from videos acquired from 15 different heat dissipation values on the boiling curve to learn how these images correlate to the underlying heat flux. The CNN predicted heat fluxes with a mean square error of 14.6 W/cm2. Furthermore, the CNN presented high sensitivity to pixel intensity in image regions relevant to boiling heat transfer physics (e.g. bubble boundaries, contextual clues about bubble image distortion). Understanding this approach may allow future studies in unstable and critical boiling regimes that require fast transient heat flux diagnostics and cannot rely on slower thermohydraulic measurements.

Experimental characterisation of the spreading of polymeric powders in powder bed fusion additive manufacturing process at changing temperature conditions

This study investigates the spreadability behaviour of four different polymeric materials, namely Polyamide 6, Polyamide 6 Black, Polypropylene, and Thermoplastic Polyurethane, under different spreading speeds (30 and 3 mm/s) and powder bed temperatures (25, 80, 110 °C) using a purposely developed experimental apparatus. Macroscopic and microscopic images of the powder layer were taken after completing the powder spreading step. A thresholding-based image processing method was utilised to evaluate the fraction of the bed area not covered by particles (NCF), and the standard deviation of pixel intensities in grayscale images (SDG) was calculated to evaluate powder layer quality in macroscopic images. NCF and SDG can provide quantitative evaluation of the quality of the spread layer, NCF in a logarithmic scale ranking and SDG in a linear scale ranking. A wavelet analysis technique was developed on microscopic panorama images obtained with grazing light to characterise the surface roughness of the layer. Results indicate that the spreadability generally worsens much more significantly than powder flow properties at increasing temperatures and, remarkably, that flowability and spreadability are unrelated. As expected, the temperature effect on powder spreading changes for the different powders are mostly governed by the approach to the powder melting temperature. Minor effects on the final layer quality were also observed at changing spreading speed.

Classifying alzheimer's disease from SMRI data using a hybrid deep learning approaches

The chance of developing "Alzheimer's Disease (AD)" increases every 5 years after 65 years of age, making it a particularly common form of neurodegenerative disorder among the older population. The use of "Magnetic Resonance Imaging (MRI)" to diagnose AD has grown in popularity in recent years. A further benefit of MRI is that it provides excellent contrast and exquisite structural detail. As a result, some studies have used biological markers backed by "structural MRI (sMRI)" to separate AD populations, which indicate differences in brain tissue size and degradation of the nervous system. The lack of properly segmented regions and essential features by the existing models might affect classification accuracy for AD. The categorization of AD in this study is based on sMRI. In this research, the hybrid Deep-Learning Models "SegNet and ResNet (SegResNet)" have been proposed for segmentation, feature extraction, and to classify the AD. SegNet network is used to identify and segment specific brain areas. Edges and circles are the SegNet's first levels, whereas the deeper layers acquire more nuanced and useful features. SegNet's last deconvolution layer produces a wide range of segmented images linked to the 3 categorization labels "Cognitive Normal (CN)", "Mild Cognitive Impairment (MCI)", and "AD" which the machine has earlier found out. To increase classification performance, the attributes of each segmented sMRI image serve as strong features of the labels. To enhance the feature information used for classification, a feature vector is built by combining the values of the pixel intensity of the segmented sMRI images. ResNet-101 classifiers are then used for characterizing vectors to identify the presence or absence of AD or MCI in each sMRI image. In terms of detection and classification accuracy, the proposed SegResNet Model is superior to the existing KNN, EFKNN, AANFIS, and ACS approaches

Fully automated contrast selection of joint bright- and black-blood late gadolinium enhancement imaging for robust myocardial scar assessment

PurposeJoint bright- and black-blood MRI techniques provide improved scar localization and contrast. Black-blood contrast is obtained after the visual selection of an optimal inversion time (TI) which often results in uncertainties, inter- and intra-observer variability and increased workload. In this work, we propose an artificial intelligence-based algorithm to enable fully automated TI selection and simplify myocardial scar imaging. MethodsThe proposed algorithm first localizes the left ventricle using a U-Net architecture. The localized left cavity centroid is extracted and a squared region of interest (“focus box”) is created around the resulting pixel. The focus box is then propagated on each image and the sum of the pixel intensity inside is computed. The smallest sum corresponds to the image with the lowest intensity signal within the blood pool and healthy myocardium, which will provide an ideal scar-to-blood contrast. The image's corresponding TI is considered optimal. The U-Net was trained to segment the epicardium in 177 patients with binary cross-entropy loss. The algorithm was validated retrospectively in 152 patients, and the agreement between the algorithm and two magnetic resonance (MR) operators' prediction of TI values was calculated using the Fleiss' kappa coefficient. Thirty focus box sizes, ranging from 2.3mm2 to 20.3cm2, were tested. Processing times were measured. ResultsThe U-Net's Dice score was 93.0 ± 0.1%. The proposed algorithm extracted TI values in 2.7 ± 0.1 s per patient (vs. 16.0 ± 8.5 s for the operator). An agreement between the algorithm's prediction and the MR operators' prediction was found in 137/152 patients (κ= 0.89), for an optimal focus box of size 2.3cm2. ConclusionThe proposed fully-automated algorithm has potential of reducing uncertainties, variability, and workload inherent to manual approaches with promise for future clinical implementation for joint bright- and black-blood MRI.

A Novel Framework of Hybrid Optimization Techniques for Contrast-Enhancement in Cardiac MRI Medical Images

Medical image processing plays a crucial role in diagnosing and treating various diseases, with Magnetic Resonance Imaging (MRI) being a widely used modality. A previous study introduced an algorithm aimed at eliminating noise from MRI images through a filtering process, effectively suppressing mixed noise while preserving important structures like edges. However, limitations in optimizing parameters for image enhancement and occasional edge preservation failures were observed. Addressing the challenge of inadequate contrast enhancement in medical image processing, this study proposes a solution utilizing the hybrid cetacean optimization algorithm and Sand Cat Swarm Optimization (COA-SCSO) to enhance contrast in MRI images. The proposed approach focuses on improving the contrast nature in MRI images to mitigate image degradation during acquisition. Cardiac MRI images from SCMR consensus and AMRG Atlas databases are analyzed, emphasizing the consideration of spatial information for enhancement. The enhancement process involves integrating a cost function with a contrast measure based on multiple metrics to create a novel transformation function. Soft-computing approaches are employed to reduce time complexity, while expanding the search pattern through a transformation function in the spatial domain. This method enhances pixel intensity, contributing to increased image resolution. Performance evaluation of the proposed algorithm incorporates various parameters such as Peak Signal-to-Noise Ratio (PSNR), Mean Squared Error (MSE), Normalized Absolute Error (NAE), and Structural Similarity Index (SSIM). Results from analysis of ten cardiac MRI images indicate that the proposed algorithm achieves the highest PSNR (98) and SSIM (0.99) values, as well as the lowest NAE (-0.17) and MSE (0.16) values, compared to the particle swarm optimized texture-based histogram equalization (PSOTHE) method and modified sunflower optimization (MSFO) method. Overall, the COA-SCSO approach demonstrates promising results in enhancing contrast in MRI images, offering potential benefits for medical diagnosis and treatment planning. Further research could explore additional optimization techniques and validation on larger datasets to further validate its effectiveness in clinical practice.

Feasibility of proton dosimetry overriding planning CT with daily CBCT elaborated through generative artificial intelligence tools

Radiotherapy commonly utilizes cone beam computed tomography (CBCT) for patient positioning and treatment monitoring. CBCT is deemed to be secure for patients, making it suitable for the delivery of fractional doses. However, limitations such as a narrow field of view, beam hardening, scattered radiation artifacts, and variability in pixel intensity hinder the direct use of raw CBCT for dose recalculation during treatment. To address this issue, reliable correction techniques are necessary to remove artifacts and remap pixel intensity into Hounsfield Units (HU) values. This study proposes a deep-learning framework for calibrating CBCT images acquired with narrow field of view (FOV) systems and demonstrates its potential use in proton treatment planning updates. Cycle-consistent generative adversarial networks (cGAN) processes raw CBCT to reduce scatter and remap HU. Monte Carlo simulation is used to generate CBCT scans, enabling the possibility to focus solely on the algorithm’s ability to reduce artifacts and cupping effects without considering intra-patient longitudinal variability and producing a fair comparison between planning CT (pCT) and calibrated CBCT dosimetry. To showcase the viability of the approach using real-world data, experiments were also conducted using real CBCT. Tests were performed on a publicly available dataset of 40 patients who received ablative radiation therapy for pancreatic cancer. The simulated CBCT calibration led to a difference in proton dosimetry of less than 2%, compared to the planning CT. The potential toxicity effect on the organs at risk decreased from about 50% (uncalibrated) up the 2% (calibrated). The gamma pass rate at 3%/2 mm produced an improvement of about 37% in replicating the prescribed dose before and after calibration (53.78% vs 90.26%). Real data also confirmed this with slightly inferior performances for the same criteria (65.36% vs 87.20%). These results may confirm that generative artificial intelligence brings the use of narrow FOV CBCT scans incrementally closer to clinical translation in proton therapy planning updates.

METHODOLOGY FOR IMPROVING DEEP LEARNING-BASED CLASSIFICATION FOR CT SCAN COVID-19 IMAGES

Early identification of COVID-19 necessitates a precise interpretation of computed tomography (CT) chest images. Consistent and accurate photographs are critical for the correct diagnosis. In this paper, a novel pre-processing approach (mainly improves the contrast) based on gradient enhancement method (GCE) is proposed for prominent visualization of the diagnostic features from the COVID-19 CT images. This pre-processing stage helps in preserving the diagnostic information in the disease affected area. Edge information in the CT images helps the physicians and classification model for better classification. The edge features are preserved by improving the contrast by using multi-scale dependent dark pass filter. From the edge features and pixel intensities, cumulative distribution function (CDF) is computed. It is then mapped to a uniform distribution, resulting in contrast enhanced COVID-19 CT images. These pre-processed images are fed to the customized deep convolutional neural networks (CNNs) like AlexNet, VGG-19, ResNet-101, DenseNet-201, GoogleNet, MobileNet-v2, SqueezeNet, Inception-v3, Xception, and EfficientNet-b0 for classification. Introducing GCE as a pre-processing stage improves the COVID-19 classification accuracy by nearly 6%. Evaluation of the contrast enhancement by GCE technique is carried out on the CT images by contrast improvement index (CII), discrete entropy (DE), and Kullback–Leibler distance (KL-Distance) measures. The experimental findings reveal that the GCE method produces higher CII and DE values than the other enhancement methods available.