U-Net Convolutional Neural Network Research Articles

Development of UroSAM: a machine learning model to automatically identify kidney stone composition from endoscopic video.

Introduction Chemical composition analysis is important in prevention counseling for kidney stone disease. Advances in laser technology have made dusting techniques more prevalent, but this offers no consistent way to collect enough material to send for chemical analysis, leading many to forgo this test. We developed a novel machine learning (ML) model to effectively assess stone composition based on intraoperative endoscopic video data. Methods Two endourologists performed ureteroscopy for kidney stones ≥ 10mm. Representative videos were recorded intraoperatively. Individual frames were extracted from the videos and the stone was outlined by human tracing. An ML model, UroSAM, was built and trained to automatically identify kidney stones in the images and predict the majority stone composition: calcium oxalate monohydrate (COM), dihydrate (COD), calcium phosphate (CAP), or uric acid (UA). UroSAM was built on top of the publicly available Segment Anything Model (SAM) and incorporated a U-Net convolutional neural network (CNN). Discussion A total of 78 ureteroscopy videos were collected; 50 were used for the model after exclusions (32 COM, 8 COD, 8 CAP, 2 UA). The ML model segmented the images with 94.77% precision. Dice coefficient (0.9135) and Intersection over Union (0.8496) confirmed good segmentation performance of the ML model. A video-wise evaluation demonstrated 60% correct classification of stone composition. Subgroup analysis showed correct classification in 84.4% of COM videos. A post-hoc adaptive threshold technique was used to mitigate biasing of the model towards COM due to data imbalance - this improved the overall correct classification to 62% while improving the classification of COD, CAP, and UA videos. Conclusions This study demonstrates the successful development of UroSAM, an ML model that precisely identifies kidney stones from natural endoscopic video data. More high quality video data will improve the performance of the model in classifying the majority stone composition.

Clinical application of high-resolution spiral CT scanning in the diagnosis of auriculotemporal and ossicle

Precision and intelligence in evaluating the complexities of middle ear structures are required to diagnose auriculotemporal and ossicle-related diseases within otolaryngology. Due to the complexity of the anatomical details and the varied etiologies of illnesses such as trauma, chronic otitis media, and congenital anomalies, traditional diagnostic procedures may not yield accurate diagnoses. This research intends to enhance the diagnosis of diseases of the auriculotemporal region and ossicles by combining High-Resolution Spiral Computed Tomography (HRSCT) scanning with Deep Learning Techniques (DLT). This study employs a deep learning method, Convolutional Neural Network-UNet (CNN-UNet), to extract sub-pixel information from medical photos. This method equips doctors and researchers with cutting-edge resources, leading to groundbreaking discoveries and better patient healthcare. The research effort is the interaction between the CNN-UNet model and high-resolution Computed Tomography (CT) scans, automating activities including ossicle segmentation, fracture detection, and disruption cause classification, accelerating the diagnostic process and increasing clinical decision-making. The suggested HRSCT-DLT model represents the integration of high-resolution spiral CT scans with the CNN-UNet model, which has been fine-tuned to address the nuances of auriculotemporal and ossicular diseases. This novel combination improves diagnostic efficiency and our overall understanding of these intricate diseases. The results of this study highlight the promise of combining high-resolution CT scanning with the CNN-UNet model in otolaryngology, paving the way for more accurate diagnosis and more individualized treatment plans for patients experiencing auriculotemporal and ossicle-related disruptions.

An efficient finite element mesh generation methodology based on μCT images of multi-layer woven composites

High-fidelity models are essential for accurate finite element (FE) simulations of composite material behavior. This paper proposes an efficient meshing methodology based on micro-Computed Tomography (μCT) images. U-Net convolutional neural network was used for image segmentation. Connected yarns were then separated using an improved procedure based on watershed algorithm and geometric transformations. The proposed Constrained Delaunay-Advancing Front Technique (CD-AFT) surface reconstruction algorithm extracts point cloud of yarns from segmented images and outputs high-quality and smooth orientable manifold watertight triangulated surface. Intersecting meshes of yarns are separated through node position detection and Laplacian moving. Experimental results show that proposed methodology is capable of accomplishing mesh generation for different mesh sizes. Compared with commercial software, it has obvious advantages in mesh quality and size control. Since the proposed method operates independently of commercial software and manual operation, it facilitates the automated generation of numerous high-fidelity models from μCT images for FE simulations.

Design of sports achievement prediction system based on U-net convolutional neural network in the context of machine learning

Sports plays a pivotal role in national development. To accurately predict college students' sports performance and motivate them to improve their physical fitness, this study constructs a sports achievement prediction system by using a U-Net Convolutional Neural Network (CNN) in machine learning. Firstly, the current state of physical education teachers’ instructional proficiency is investigated and analyzed to identify existing problems. Secondly, an improved U-Net-based sports achievement prediction system is proposed. This method enhances the utilization and propagation of network features by incorporating dense connections, thus addressing gradient disappearance issues. Simultaneously, an improved mixed loss function is introduced to alleviate class imbalance. Finally, the effectiveness of the proposed system is validated through testing, demonstrating that the improved U-Net CNN algorithm yields superior results. Specifically, the prediction accuracy of the improved network for sports performance surpasses that of the original U-Net by 4.22 % and exceeds that of DUNet by 5.22 %. Compared with other existing prediction networks, the improved U-Net CNN model exhibits a superior achievement prediction ability. Consequently, the proposed system enhances teaching and learning efficiency and offers insights into applying artificial intelligence technology to smart classroom development.

Near-real-time multiparametric seismic and visual monitoring of explosive activity at Sabancaya volcano, Peru

This study presents the development of a multiparametric system that utilizes artificial intelligence techniques to identify and analyze volcanic explosions in near real-time. The study analyzed 1343 explosions recorded between 2019 and 2021, along with seismic, meteorological, and visible image data from the Sabancaya volcano. Deep learning algorithms like the U-Net convolutional neural network were used to segment and measure volcanic plumes in images, while boosting-based machine learning ensembles were used to classify seismic events related to ash plumes. The findings demonstrate that these approaches effectively handle large amounts of data generated during seismic and eruptive crises. The U-Net network achieved precise segmentation of volcanic plumes with over 98% accuracy and the ability to generalize to new data. The CatBoost classifier achieved an average accuracy of 94.5% in classifying seismic events. These approaches enable the real-time estimation of eruptive parameters without human intervention, contributing to the development of early warning systems for volcanic hazards. In conclusion, this study highlights the feasibility of using seismic signals and images to detect and characterize volcanic explosions in near real-time, making a significant contribution to the field of volcanic monitoring.

Application of a U-Net Neural Network to the Puccinia sorghi-Maize Pathosystem.

Computer vision approaches to analyze plant disease data can be both faster and more reliable than traditional, manual methods. However, the requirement of manually annotating training data for the majority of machine learning applications can present a challenge for pipeline development. Here, we describe a machine learning approach to quantify Puccinia sorghi incidence on maize leaves utilizing U-Net convolutional neural network models. We analyzed several U-Net models with increasing amounts of training image data, either randomly chosen from a large data pool or randomly chosen from a subset of disease time course data. As the training dataset size increases, the models perform better, but the rate of performance decreases. Additionally, the use of a diverse training dataset can improve model performance and reduce the amount of annotated training data required for satisfactory performance. Models with as few as 48 whole-leaf training images are able to replicate the ground truth results within our testing dataset. The final model utilizing our entire training dataset performs similarly to our ground truth data, with an intersection over union value of 0.5002 and an F1 score of 0.6669. This work illustrates the capacity of U-Nets to accurately answer real-world plant pathology questions related to quantification and estimation of plant disease symptoms. [Formula: see text] Copyright © 2024 The Author(s). This is an open access article distributed under the CC BY-NC-ND 4.0 International license.

UroAngel: a single-kidney function prediction system based on computed tomography urography using deep learning.

Accurate estimation of the glomerular filtration rate (GFR) is clinically crucial for determining the status of obstruction, developing treatment strategies, and predicting prognosis in obstructive nephropathy (ON). We aimed to develop a deep learning-based system, named UroAngel, for non-invasive and convenient prediction of single-kidney function level. We retrospectively collected computed tomography urography (CTU) images and emission computed tomography diagnostic reports of 520 ON patients. A 3D U-Net model was used to segment the renal parenchyma, and a logistic regression multi-classification model was used to predict renal function level. We compared the predictive performance of UroAngel with the Modification of Diet in Renal Disease (MDRD), Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equations, and two expert radiologists in an additional 40 ON patients to validate clinical effectiveness. UroAngel based on 3D U-Net convolutional neural network could segment the renal cortex accurately, with a Dice similarity coefficient of 0.861. Using the segmented renal cortex to predict renal function stage had high performance with an accuracy of 0.918, outperforming MDRD and CKD-EPI and two radiologists. We proposed an automated 3D U-Net-based analysis system for direct prediction of single-kidney function stage from CTU images. UroAngel could accurately predict single-kidney function in ON patients, providing a novel, reliable, convenient, and non-invasive method.

Raman difference spectroscopy and U-Net convolutional neural network for molecular analysis of cutaneous neurofibroma.

In Neurofibromatosis type 1 (NF1), peripheral nerve sheaths tumors are common, with cutaneous neurofibromas resulting in significant aesthetic, painful and functional problems requiring surgical removal. To date, determination of adequate surgical resection margins-complete tumor removal while attempting to preserve viable tissue-remains largely subjective. Thus, residual tumor extension beyond surgical margins or recurrence of the disease may frequently be observed. Here, we introduce Shifted-Excitation Raman Spectroscopy in combination with deep neural networks for the future perspective of objective, real-time diagnosis, and guided surgical ablation. The obtained results are validated through established histological methods. In this study, we evaluated the discrimination between cutaneous neurofibroma (n = 9) and adjacent physiological tissues (n = 25) in 34 surgical pathological specimens ex vivo at a total of 82 distinct measurement loci. Based on a convolutional neural network (U-Net), the mean raw Raman spectra (n = 8,200) were processed and refined, and afterwards the spectral peaks were assigned to their respective molecular origin. Principal component and linear discriminant analysis was used to discriminate cutaneous neurofibromas from physiological tissues with a sensitivity of 100%, specificity of 97.3%, and overall classification accuracy of 97.6%. The results enable the presented optical, non-invasive technique in combination with artificial intelligence as a promising candidate to ameliorate both, diagnosis and treatment of patients affected by cutaneous neurofibroma and NF1.

Combined Unet and CNN image classification model for COVID disease detection using CXR/CT imaging

Accurate SARS-CoV-2 screening is made possible by automated Computer-Aided Diagnosis (CAD) which reduces the stress on healthcare systems. Since Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) is highly contagious, the transition chain can be broken through an early diagnosis by clinical knowledge and Artificial Intelligence (AI). Manual findings are time and labor-intensive. Even if Reverse Transcription-Polymerase Chain Reaction (RT-PCR) delivers quick findings, Chest X-ray (CXR) imaging is still a more trustworthy tool for disease classification and assessment. Several studies have been conducted using Deep Learning (DL) algorithms for COVID-19 detection. One of the biggest challenges in modernizing healthcare is extracting useful data from high-dimensional, heterogeneous, and complex biological data. Intending to introduce an automated COVID-19 diagnosis model, this paper develops a proficient optimization model that enhances the classification performance with better accuracy. The input images are initially pre-processed with an image filtering approach for noise removal and data augmentation to extend the dataset. Secondly, the images are segmented via U-Net and are given to classification using the Fused U-Net Convolutional Neural Network (FUCNN) model. Here, the performance of U-Net is enhanced through the modified Moth Flame Optimization (MFO) algorithm named Chaotic System-based MFO (CSMFO) by optimizing the weights of U-Net. The significance of the implemented model is confirmed over a comparative evaluation with the state-of-the-art models. Specifically, the proposed CSMFO-FUCNN attained 98.45% of accuracy, 98.63% of sensitivity, 98.98% of specificity, and 98.98% of precision.

Using U-Net convolutional neural network to model pixel-based electrostatic potential distributions in GaN power MIS-HEMTs

This study demonstrates a novel use of the U-Net convolutional neural network (CNN) for modeling pixel-based electrostatic potential distributions in GaN metal–insulator-semiconductor high-electron mobility transistors (MIS-HEMTs) with various gate and source field plate designs and drain voltages. The pixel-based images of the potential distribution are successfully modeled from the developed U-Net CNN with an error of less than 1% error relative to a TCAD simulated reference of a 500-V electrostatic potential distribution in the AlGaN/GaN interface. Furthermore, the modeling time of potential distributions by U-Net takes about 80 ms. Therefore, the U-Net CNN is a promising approach to efficiently model the pixel-based distributions characteristics in GaN power devices.

Application of deep learning on mammographies to discriminate between low and high-risk DCIS for patient participation in active surveillance trials

BackgroundDuctal Carcinoma In Situ (DCIS) can progress to invasive breast cancer, but most DCIS lesions never will. Therefore, four clinical trials (COMET, LORIS, LORETTA, AND LORD) test whether active surveillance for women with low-risk Ductal carcinoma In Situ is safe (E. S. Hwang et al., BMJ Open, 9: e026797, 2019, A. Francis et al., Eur J Cancer. 51: 2296–2303, 2015, Chizuko Kanbayashi et al. The international collaboration of active surveillance trials for low-risk DCIS (LORIS, LORD, COMET, LORETTA), L. E. Elshof et al., Eur J Cancer, 51, 1497–510, 2015). Low-risk is defined as grade I or II DCIS. Because DCIS grade is a major eligibility criteria in these trials, it would be very helpful to assess DCIS grade on mammography, informed by grade assessed on DCIS histopathology in pre-surgery biopsies, since surgery will not be performed on a significant number of patients participating in these trials.ObjectiveTo assess the performance and clinical utility of a convolutional neural network (CNN) in discriminating high-risk (grade III) DCIS and/or Invasive Breast Cancer (IBC) from low-risk (grade I/II) DCIS based on mammographic features. We explored whether the CNN could be used as a decision support tool, from excluding high-risk patients for active surveillance.MethodsIn this single centre retrospective study, 464 patients diagnosed with DCIS based on pre-surgery biopsy between 2000 and 2014 were included. The collection of mammography images was partitioned on a patient-level into two subsets, one for training containing 80% of cases (371 cases, 681 images) and 20% (93 cases, 173 images) for testing. A deep learning model based on the U-Net CNN was trained and validated on 681 two-dimensional mammograms. Classification performance was assessed with the Area Under the Curve (AUC) receiver operating characteristic and predictive values on the test set for predicting high risk DCIS-and high-risk DCIS and/ or IBC from low-risk DCIS.ResultsWhen classifying DCIS as high-risk, the deep learning network achieved a Positive Predictive Value (PPV) of 0.40, Negative Predictive Value (NPV) of 0.91 and an AUC of 0.72 on the test dataset. For distinguishing high-risk and/or upstaged DCIS (occult invasive breast cancer) from low-risk DCIS a PPV of 0.80, a NPV of 0.84 and an AUC of 0.76 were achieved.ConclusionFor both scenarios (DCIS grade I/II vs. III, DCIS grade I/II vs. III and/or IBC) AUCs were high, 0.72 and 0.76, respectively, concluding that our convolutional neural network can discriminate low-grade from high-grade DCIS.

Hemodynamic property incorporated brain tumor segmentation by deep learning and density-based analysis of dynamic susceptibility contrast-enhanced magnetic resonance imaging (MRI).

Magnetic resonance imaging (MRI) is a primary non-invasive imaging modality for tumor segmentation, leveraging its exceptional soft tissue contrast and high resolution. Current segmentation methods typically focus on structural MRI, such as T1-weighted post-contrast-enhanced or fluid-attenuated inversion recovery (FLAIR) sequences. However, these methods overlook the blood perfusion and hemodynamic properties of tumors, readily derived from dynamic susceptibility contrast (DSC) enhanced MRI. This study introduces a novel hybrid method combining density-based analysis of hemodynamic properties in time-dependent perfusion imaging with deep learning spatial segmentation techniques to enhance tumor segmentation. First, a U-Net convolutional neural network (CNN) is employed on structural images to delineate a region of interest (ROI). Subsequently, Hierarchical Density-Based Scans (HDBScan) are employed within the ROI to augment segmentation by exploring intratumoral hemodynamic heterogeneity through the investigation of tumor time course profiles unveiled in DSC MRI. The approach was tested and evaluated using a cohort of 513 patients from the open-source University of Pennsylvania glioblastoma database (UPENN-GBM) dataset, achieving a 74.83% Intersection over Union (IoU) score when compared to structural-only segmentation. The algorithm also exhibited increased precision and localized predictions of heightened segmentation boundary complexity, resulting in a 146.92% increase in contour complexity (ICC) compared to the reference standard provided by the UPENN-GBM dataset. Importantly, segmenting tumors with the developed new approach uncovered a negative correlation of the tumor volume with the scores in the Karnofsky Performance Scale (KPS) clinically used for assessing the functional status of patients (-0.309), which is not observed with the prevailing segmentation standard. This work demonstrated that including hemodynamic properties of tissues from DSC MRI can improve existing structural or morphological feature-based tumor segmentation techniques with additional information on tumor biology and physiology. This approach can also be applied to other clinical indications that use perfusion MRI for diagnosis or treatment monitoring.

A comprehensive deep learning geometric shape optimization framework with field prediction surrogate and reinforcement learning

The optimization of aerodynamic components' geometric shapes demands a novel technical approach for adaptive and efficient exploration and decision-making within the design space. In this study, we introduce an innovative shape optimization framework that leverages deep reinforcement learning with neural network surrogate models. The field prediction surrogate, realized by two distinct U-net architectures, can efficiently generate holistic field solutions based on the transformed mesh coordinates. Subsequently, an inference engine dynamically calculates the key metric of the flow fields, serving as the objective function for the subsequent geometry-aware Deep Q network (DQN)-based optimization. The framework's efficacy is validated using a rocket nozzle as an illustrative example. During surrogate validation, under both friction and frictionless conditions, the l1 errors of the entire flow field of both the U-net vision transformer (ViT) and U-net convolutional neural network (CNN) architectures are less than 0.4%. The proposed U-net ViT consistently outperforms U-net CNN, and the superiority is particularly evident in complex flow areas, outlet sections, and vacuum thrust prediction. Following training, the DQN model is employed to explore the design variable space. The B-spline defining profile successfully is optimized to a final expanding segment shape with improved thrust. Under frictionless conditions, it closely approaches the theoretical optimum. In the practical condition considering friction, the optimized shape gains a 2.96% thrust improvement. The results demonstrate that the proposed framework, especially when coupled with U-net ViT, exhibits enhanced accuracy and adaptability for shape optimization tasks.

An Efficient Segmentation and Classification of Brain Tumor Detection using Deep Learning

In medical diagnostics, the identification of brain tumors is a crucial endeavour that frequently necessitates the careful review of intricate imaging data. With the development of deep learning methods, especially convolutional neural networks (CNNs), automated brain tumor detection has become a viable way to help radiologists diagnose patients accurately and quickly. This study provides an extensive overview of current developments in deep learning approaches for brain tumor identification. The suggested techniques usually start with preprocessing medical images, like CT or MRI scans, and then use CNN architectures to extract features. Different CNN architectures, such as U-Net and conventional CNNs, have been used to extract discriminative characteristics from brain pictures. The performance of brain tumor detection systems has also been improved using transfer learning techniques, which make use of pre-trained models on sizable datasets. These results are especially encouraging when there is a shortage of training data. Additionally, research has been done on using ensemble learning approaches to strengthen the generality and durability of brain tumor detection models. By combining several base models into one prediction, these methods improve overall performance and lower the likelihood of overfitting. Furthermore, the incorporation of sophisticated regularisation methods such batch normalisation and dropout has enhanced the deep learning models' capacity for generalisation in brain tumor identification tasks. Moreover, the use of deep learning models in practical clinical environments demands the resolution of issues of uncertainty estimation and model interpretability

Abstract 230: Deep learning embedding-based segmentation for morphological analysis in organoids

Abstract Background: The organoid model is a useful tool for modeling the cellular microenvironment of the organ from which they are derived. Organoids recapitulate the self-organization of heterogenous cell types and the microenvironment. Quantifying the morphological features of organoids can provide valuable insights into cellular organizational defects and growth characteristics, which can facilitate drug discovery. Measurement of these characteristics in live organoids can be performed quickly and easily directly in culture using widefield microscopy. The current state-of-the-art method to detect and identify the shape of individual organoids in brightfield images uses the U-Net Convolutional Neural Network (CNN) with a watershed transform to label pixels. However, this method yields jagged shape proposals and cannot detect overlapping regions. We propose the use of an embedding-based segmentation network based on the Branched ERF-Net to solve these issues. Methods: Organoids were established from the proximal colon of BrafV600E heterozygous mice, and 50 phase-contrast images of the organoids were acquired at 2.5x and 4x magnification. Organoids present in acquired images were labeled manually. The network was trained on this collection of manually labeled images and validated on a separate validation dataset. Results: The network accurately labels organoids. Overlapping organoids are segmented correctly and shape proposals are smoother. Detection of overlapping organoids with the Branched ERF-Net architecture yields an accurate organoid count, verified against manual counting. Smoother shape proposals also enable the use of convexity defects to measure organoid budding. Conclusions: Deep-learning analysis of widefield images enables the rapid assessment of morphological characteristics, such as size, count, and budding, which are key to understanding proliferation and differentiation changes. The modified Branched ERF-Net architecture we propose for organoid segmentation is a robust and versatile method to quantify organoid morphology. Citation Format: Akash Sureshkumar, Shilpa Bisht, Hariharan Easwaran. Deep learning embedding-based segmentation for morphological analysis in organoids [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 230.

Abstract 2319: Deep learning-enabled dynamic infiltration and response to NK therapies in solid tumor organoids

Abstract Patient-derived organoid co-culture models have become a state of the art platform to evaluate the effectiveness of immunotherapeutic agents in preclinical studies. However, preclinical development as well as quantification of therapeutic potency and resistance require differential labeling of cells with fluorescent dyes or transgenic fluorescent proteins, which may dysregulate the effector cell function or produce undesired cytotoxicity. Deep learning-based image analysis allows the use of brightfield images to co-localize tumor organoids (TOs) with effector cells, and measure TO-specific responses to novel candidate immunotherapies in a label-free manner; this can accelerate their evaluation as therapeutic candidates in cancer patients and shed light on tumor-immune interaction mechanisms. Here we record multi-day time-lapse confocal microscopy images of TOs co-cultured with NK cells at increasing concentrations of target to effector cells. Two pre-trained U-Net convolutional neural networks are used to separately segment tumor and immune cells from the brightfield channel, and quantify immune cell infiltration over time. TO segmentation masks are registered with a vital dye (caspase 3/7) channel to selectively quantify TO cell death and correlate it with estimated immune cell infiltration across a large cohort of patient-derived TO models (including breast, colorectal, endometrial, gastric, head and neck, liver, lung and pancreatic cancer), at increasing concentrations of 6 different NK cell types (identified as A through F). We find that the estimated density of infiltrating immune cells is highly correlated with TO death, as quantified by fluorescence intensity of caspase over time (median Pearson’s r=0.89, P<0.001). Higher ratios of effector to target cells lead to higher activation of NK cells as measured by infiltrating cell density. Differential infiltration dynamics are observed across TOs and immune cell lines, as peak infiltration density is affected by co-culture time, ratio of target to effector cells, TO line, and NK cell type. These findings highlight that the paired segmentation models are able to measure varying degrees of infiltration across different types of immune cells and TO models. The approach described here is highly scalable and only requires capturing brightfield images of the assay, therefore eliminating the need to label cells and avoiding phototoxic effects or alteration of the immune cells activity. Time series measurements enable quantification of patterns of immune cell activation, including infiltration, migration and co-localization dynamics, which provide insights into the pharmacokinetics and mechanisms for specific immune therapies. Overall, this methodology enables high throughput screening of many therapeutic candidates across dozens to hundreds of unique TO-models, thereby facilitating targeted precision therapy. Citation Format: Luca Lonini, Sonal Khare, Timothy Don Lopez, Stanislaw Szydlo, Michael Streit, Olga A. Karginova, Mary K. Flaherty, Ameen Salhaudeen, Andre Kunert, Anna Oja, Kate Sasser, Martin C. Stumpe, Chi-Sing Ho. Deep learning-enabled dynamic infiltration and response to NK therapies in solid tumor organoids [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2319.

Deep Learning Powered Identification of Differentiated Early Mesoderm Cells from Pluripotent Stem Cells.

Pluripotent stem cells can be differentiated into all three germ-layers including ecto-, endo-, and mesoderm in vitro. However, the early identification and rapid characterization of each germ-layer in response to chemical and physical induction of differentiation is limited. This is a long-standing issue for rapid and high-throughput screening to determine lineage specification efficiency. Here, we present deep learning (DL) methodologies for predicting and classifying early mesoderm cells differentiated from embryoid bodies (EBs) based on cellular and nuclear morphologies. Using a transgenic murine embryonic stem cell (mESC) line, namely OGTR1, we validated the upregulation of mesodermal genes (Brachyury (T): DsRed) in cells derived from EBs for the deep learning model training. Cells were classified into mesodermal and non-mesodermal (representing endo- and ectoderm) classes using a convolutional neural network (CNN) model called InceptionV3 which achieved a very high classification accuracy of 97% for phase images and 90% for nuclei images. In addition, we also performed image segmentation using an Attention U-Net CNN and obtained a mean intersection over union of 61% and 69% for phase-contrast and nuclear images, respectively. This work highlights the potential of integrating cell culture, imaging technologies, and deep learning methodologies in identifying lineage specification, thus contributing to the advancements in regenerative medicine. Collectively, our trained deep learning models can predict the mesoderm cells with high accuracy based on cellular and nuclear morphologies.

Depth_ASPP_UNet++: a novel 2D ESPI phase unwrapping based on convolutional neural network

Phase unwrapping is a key step in optical metrology and physical optics to obtain accurate phase distributions. In practice, phase images obtained from electronic speckle pattern interferometry (ESPI) exhibit diverse and complex morphology, with significant shape variations and non-uniform densities among different individuals. This takes challenges for accurately extracting phase information and unwrapping the phase. With the progress of deep learning technology in optical image processing, real-time performance and accuracy have become concerned issues. In this paper, an ESPI phase unwrapping method based on convolutional neural network UNet++ is proposed. The proposed network combines the depthwise separable convolution (DSC), atrous spatial pyramid pooling (ASPP), defined as Depth_ASPP_UNet++. In this model, the use of DSC improves network computational efficiency and provides better feature representation capability. In addition, ASPP is introduced to pay more attention to the phase information of the phase image, and then obtain better phase unwrapping results. The experimental results show that our proposed method can obtain excellent results, especially with various of variable density, different noise levels, and different speckle sizes.

Efficient Mapping Between Void Shapes and Stress Fields Using Deep Convolutional Neural Networks With Sparse Data

Abstract Establishing fast and accurate structure-to-property relationships is an important component in the design and discovery of advanced materials. Physics-based simulation models like the finite element method (FEM) are often used to predict deformation, stress, and strain fields as a function of material microstructure in material and structural systems. Such models may be computationally expensive and time intensive if the underlying physics of the system is complex. This limits their application to solve inverse design problems and identify structures that maximize performance. In such scenarios, surrogate models are employed to make the forward mapping computationally efficient to evaluate. However, the high dimensionality of the input microstructure and the output field of interest often renders such surrogate models inefficient, especially when dealing with sparse data. Deep convolutional neural network (CNN) based surrogate models have shown great promise in handling such high-dimensional problems. In this paper, a single ellipsoidal void structure under a uniaxial tensile load represented by a linear elastic, high-dimensional and expensive-to-query, FEM model. We consider two deep CNN architectures, a modified convolutional autoencoder framework with a fully connected bottleneck and a UNet CNN, and compare their accuracy in predicting the von Mises stress field for any given input void shape in the FEM model. Additionally, a sensitivity analysis study is performed using the two approaches, where the variation in the prediction accuracy on unseen test data is studied through numerical experiments by varying the number of training samples from 20 to 100.

Full-length radiograph based automatic musculoskeletal modeling using convolutional neural network

Full-length radiographs contain information from which many anatomical parameters of the pelvis, femur, and tibia may be derived, but only a few anatomical parameters are used for musculoskeletal modeling. This study aimed to develop a fully automatic algorithm to extract anatomical parameters from full-length radiograph to generate a musculoskeletal model that is more accurate than linear scaled one. A U-Net convolutional neural network was trained to segment the pelvis, femur, and tibia from the full-length radiograph. Eight anatomic parameters (six for length and width, two for angles) were automatically extracted from the bone segmentation masks and used to generate the musculoskeletal model. Sørensen-Dice coefficient was used to quantify the consistency of automatic bone segmentation masks with manually segmented labels. Maximum distance error, root mean square (RMS) distance error and Jaccard index (JI) were used to evaluate the geometric accuracy of the automatically generated pelvis, femur and tibia models versus CT bone models. Mean Sørensen-Dice coefficients for the pelvis, femur and tibia 2D segmentation masks were 0.9898, 0.9822 and 0.9786, respectively. The algorithm-driven bone models were closer to the 3D CT bone models than the scaled generic models in geometry, with significantly lower maximum distance error (28.3 % average decrease from 24.35 mm) and RMS distance error (28.9 % average decrease from 9.55 mm) and higher JI (17.2 % average increase from 0.46) (P < 0.001). The algorithm-driven musculoskeletal modeling (107.15 ± 10.24 s) was faster than the manual process (870.07 ± 44.79 s) for the same full-length radiograph. This algorithm provides a fully automatic way to generate a musculoskeletal model from full-length radiograph that achieves an approximately 30 % reduction in distance errors, which could enable personalized musculoskeletal simulation based on full-length radiograph for large scale OA populations.