Proceedings of SPIE--the International Society for Optical Engineering最新文献

{"title":"Fully Differentiable dMRI Streamline Propagation in PyTorch.","authors":"Jongyeon Yoon, Elyssa M McMaster, Michael E Kim, Gaurav Rudravaram, Kurt G Schilling, Bennett A Landman, Daniel Moyer","doi":"10.1117/12.3086590","DOIUrl":"10.1117/12.3086590","url":null,"abstract":"<p><p>Diffusion MRI (dMRI) provides a distinctive means to probe the microstructural architecture of living tissue, facilitating applications such as brain connectivity analysis, modeling across multiple conditions, and the estimation of macrostructural features. Tractography, which emerged in the final years of the 20th century and accelerated in the early 21st century, is a technique for visualizing white matter pathways in the brain using dMRI. Most diffusion tractography methods rely on procedural streamline propagators or global energy minimization methods. Although recent advancements in deep learning have enabled tasks that were previously challenging, existing tractography approaches are often non-differentiable, limiting their integration in end-to-end learning frameworks. While progress has been made in representing streamlines in differentiable frameworks, no existing method offers fully differentiable propagation. In this work, we propose a fully differentiable solution that retains numerical fidelity with a leading streamline algorithm. The key is that our PyTorch-engineered streamline propagator has no components that block gradient flow, making it fully differentiable. We show that our method matches standard propagators while remaining differentiable. By translating streamline propagation into a differentiable PyTorch framework, we enable deeper integration of tractography into deep learning workflows, laying the foundation for a new category of macrostructural reasoning that is not only computationally robust but also scientifically rigorous.</p>","PeriodicalId":74505,"journal":{"name":"Proceedings of SPIE--the International Society for Optical Engineering","volume":"13925 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13245533/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148213597","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deep learning‑based cell type prediction in lung tissue from brightfield histology using CODEX-derived labels.","authors":"Sumanth Devarasetty, Nicholas Lucarelli, Sayat Mimar, Sanjay Jain, Tarek M El-Achkar, Jeffrey Purkerson, Gloria Pryhuber, Pinaki Sarder","doi":"10.1117/12.3088120","DOIUrl":"10.1117/12.3088120","url":null,"abstract":"<p><p>Identifying cell types present in lung biopsies can provide critical information on pathological processes, tissue organization, and organ health, and are valuable in both clinical and research settings. Knowledge of cell types, their distributions, and spatial relationships can assist pathologists and researchers in diagnosis, prognosis, and mechanistic investigations. Multiplex imaging technologies such as Phenocycler^™, based on co-detection by indexing (CODEX) can provide whole-slide spatial maps of protein expression and can accurately identify cell types; however, CODEX shares limitations including high costs of antibodies and reagents, as well as labor-intensive conjugation and validation steps. In contrast, hematoxylin and eosin (H&E) staining is inexpensive and routinely available, offering the potential for scalable cell-type mapping if robust prediction can be achieved directly from histology. In this work, we develop a deep learning pipeline to automatically detect cell types in H&E-stained lung tissue sections by leveraging ground-truth annotations from paired CODEX images. The dataset comprises over 2.3 million labeled cells, segmented from lung tissue sections obtained from multiple donors, and grouped into five broad classes: epithelial, immune, endothelial, stromal, and contractile. We train a DeepLabV3+ semantic segmentation model with ResNet backbone on patches from these annotated slides and reserve a subset for testing. Our model achieved 51.3% balanced accuracy across five classes, which is ~2.5x the random-guessing baseline of 20%. This framework demonstrates the feasibility of approximating multiplexed imaging-based cell type maps from routine histology.</p>","PeriodicalId":74505,"journal":{"name":"Proceedings of SPIE--the International Society for Optical Engineering","volume":"13932 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13286254/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148310655","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine learning assisted wavefront sensor.","authors":"Conor McFadden, Bingying Chen, Reto Fiolka","doi":"10.1117/12.3084846","DOIUrl":"10.1117/12.3084846","url":null,"abstract":"<p><p>Adaptive optics aims to restore diffraction limited resolution in an imaging system in the presence of optical aberrations. To this end, the wavefront error needs to be measured prior to its compensation. Traditional wavefront sensors typically measure the local gradient of the wavefront emitted from a guide star (or another known point source) using a microlens array in front of a camera. They are typically dedicated devices that need to be integrated into an imaging system. Here we have tested the concept of estimating the wavefront error directly from an image of a guide star by training a machine learning model. We produce a two-photon laser spot in a water fluorescein media, and introduce random, but known aberrations using a deformable mirror to form a large training set. After validation, we integrated the machine learning wavefront sensor in an AO feedback loop with the deformable mirror. We demonstrate proof of principle compensation of sample-introduced optical aberrations with this system.</p>","PeriodicalId":74505,"journal":{"name":"Proceedings of SPIE--the International Society for Optical Engineering","volume":"13860 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13234910/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148201571","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Design and evaluation of a deformable multi-organ abdominal phantom for multimodality imaging applications.","authors":"Mandy Yuan, Nati Nawawithan, James Yu, Chinmay Patwardhan, Suhani Swain, Izabella Chaverra, Baowei Fei","doi":"10.1117/12.3086211","DOIUrl":"https://doi.org/10.1117/12.3086211","url":null,"abstract":"<p><p>A deformable multi-organ abdominal phantom can have many applications in medical imaging and image-guided interventions. It can provide ground truth data for registration evaluation and procedure training. In this study, we designed and fabricated a deformable abdominal phantom for multimodal imaging applications. Computed tomography (CT) images of a human patient were used to segment the abdominal surface, kidneys, spleen, and liver. Scaled down 3D-printed molds were created based on the segmented 3D models of these structures. To allow for deformability, an agar-gelatin gel mixture was used to cast the phantom using the 3D-printed molds with potassium iodide added to the target formulation as a CT contrast agent. Psyllium husk powder was utilized as an ultrasound (US) scattering agent. Simulated tumors or targets were incorporated into the kidney phantom. All components were assembled into an abdominal mold to replicate patient anatomical structures. The phantom was evaluated using CT and ultrasound imaging, with preliminary results demonstrating its ability to provide ground truth for tumor localization. This study presents a reproducible framework for creating a deformable abdominal phantom that can be applied to the development and validation of image-guided interventional procedures.</p>","PeriodicalId":74505,"journal":{"name":"Proceedings of SPIE--the International Society for Optical Engineering","volume":"13927 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13709972/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149290531","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Algorithm Comparison for Oxygen Saturation Estimation in Whole Blood Using Diffuse Reflectance Hyperspectral Imaging.","authors":"Michelle Bryarly, Minh Ha Tran, Arrsh Ali, Ling Ma, Baowei Fei","doi":"10.1117/12.3081303","DOIUrl":"https://doi.org/10.1117/12.3081303","url":null,"abstract":"<p><p>Hyperspectral imaging (HSI) has been used <i>in vivo</i> to estimate the oxygenation level (SO₂) of whole blood inside blood vessels. However, little literature is devoted to validating the estimation. The goal of this research is to compare the performance of three commonly used methods: the Kubelka-Munk algorithm, optical density algorithm, and inverse Monte Carlo method, for the task of estimating SO₂ from diffuse reflectance spectra. Validation was performed using two data types: Monte Carlo simulated data and <i>ex vivo</i> sheep blood that was deoxygenated using sodium dithionite (Na₂S₂O₄). The SO₂ of sheep blood was confirmed using a blood gas analyzer. We measured the diffuse reflectance using two types of HSI cameras: a push-broom and a snapshot HSI camera system. Our results showed that visible wavelengths (400 - 600 nm) are preferred for micro-vessels with diameters less than 100 microns, and near-infrared wavelengths (600 - 900 nm) are preferred for larger vessels. We determined experimentally that sodium dithionite consistently deoxygenated whole blood by 35% for every 1 mg/mL of sodium dithionite to hemoglobin. It was found that the inverse Monte Carlo method performed well in estimating SO₂, with an average estimation error of 5.53% and 6.97% for <i>ex vivo</i> sheep blood and Monte Carlo spectra, respectively. Our findings provide a basis for determining future methods for <i>in vivo</i> SO₂ estimations, including surgical HSI applications that rely on accurate perfusion assessment.</p>","PeriodicalId":74505,"journal":{"name":"Proceedings of SPIE--the International Society for Optical Engineering","volume":"13863 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13699330/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149278453","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Analytic Bounds on GAMLSS Model Variability of Normative White Matter Brain Charts.","authors":"Michael E Kim, Gaurav Rudravaram, Adam Saunders, Chenyu Gao, Karthik Ramadass, Nancy R Newlin, Praitayini Kanakaraj, Sam Bogdanov, Derek Archer, Timothy J Hohman, Angela L Jefferson, Victoria L Morgan, Alexandra Roche, Dario J Englot, Susan M Resnick, Lori L Beason Held, Murat Bilgel, Laurie E Cutting, Laura A Barquero, Micah A D'Archangel, Tin Q Nguyen, Kathryn L Humphreys, Yanbin Niu, Sophia Vinci-Booher, Carissa J Cascio, Kimberly R Pechman, Niranjana Shashikumar, Zhiyuan Li, Simon N Vandekar, Panpan Zhang, John C Gore, Yihao Liu, Lianrui Zuo, Kurt G Schilling, Daniel C Moyer, Bennett A Landman","doi":"10.1117/12.3085643","DOIUrl":"10.1117/12.3085643","url":null,"abstract":"<p><p>Brain charts, or normative models of quantitative neuroimaging measures, can identify trajectories of brain development and abnormalities in groups and individuals by leveraging large populations. Recent work has extended these brain charts to model microstructural and macrostructural features of white matter. Assessments of variance for these brain charts are necessary to determine whether the models being used for these data are stable. We implement an analytic approach to characterize variability of the parameters in previously released brain charts created using the generalized additive models for location, scale, and shape (GAMLSS) framework. Additionally, we empirically validate the accuracy of each analytic model through a comparison to a bootstrapping approach from 0.2 to 90 years of age. We find that across all models, the analytic coefficient of variation (COV) remains below 5% for ages greater than 0.25 years, with the maximum empirical observed COV reaching 7% at 0.2 years of age. Further, the empirical assessment shows high agreement with the analytic assessment, with COV estimates averaged across the lifespan for all models having a Pearson correlation coefficient of 0.776 and a mean difference of 4 x 10^-4. Both methods exhibit volume and surface area as the features with the largest average COV for the majority of tracts. However, the analytic assessment yields axial diffusivity as the feature most frequently having the smallest COV, whereas the corresponding feature for the empirical assessment is average length. These results suggest that the analytic approach overestimates model stability for WM brain charts when the COV is low and that the validation method is suitable for assessing whether GAMLSS models are unstable.</p>","PeriodicalId":74505,"journal":{"name":"Proceedings of SPIE--the International Society for Optical Engineering","volume":"13930 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13245483/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148213614","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Extending Regression Without Truth to Integrate Ground-Truth Measurements for Evaluating Quantitative Imaging Methods with Patient Data.","authors":"Yan Liu, Abhinav K Jha","doi":"10.1117/12.3088141","DOIUrl":"10.1117/12.3088141","url":null,"abstract":"<p><p>Objective evaluation of quantitative imaging (QI) methods with patient data is often hindered by the lack of gold standards. To address this challenge, a class of regression-without-truth (RWT) techniques have been developed. These techniques assume that the true and measured values are linearly related and estimate the linear-relationship parameters without access to true values. However, reliable estimation of these parameters typically requires many patient samples, which can be expensive and time consuming to obtain, and even impossible in settings such as studies with rare diseases or with new clinical imaging procedures. Thus, there is an important need for strategies to perform evaluation of quantitative imaging methods with a small number of patient samples. In this context, we note that datasets with known ground truth, such as physical phantom studies, could be available. In this manuscript, we propose an approach that integrates information from both patient data without ground truth and known-ground-truth datasets to perform objective evaluation of QI methods. We validated the proposed approach using numerical studies, which showed that the proposed approach yielded improved performance in ranking QI methods compared with RWT technique. The results demonstrate the potential of the proposed approach for evaluating QI methods when patient data are limited and motivate further validation with clinically realistic simulation studies and clinical data.</p>","PeriodicalId":74505,"journal":{"name":"Proceedings of SPIE--the International Society for Optical Engineering","volume":"13928 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13372251/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148458024","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Modeling human observer performance with neural network observers in a forced localization task using undersampled MRI images.","authors":"Sandro Amaglobeli, Justine P Prasad, Craig K Abbey, Angel R Pineda","doi":"10.1117/12.3084847","DOIUrl":"10.1117/12.3084847","url":null,"abstract":"<p><p>Modelling human performance in forced localization tasks is a challenge for model observers because of the many possible signal locations along with capturing human perception. This is particularly challenging when using anatomical images and needing generalization to out-of-distribution images such as in varying undersampling in magnetic resonance imaging (MRI). In this study, we modeled human forced localization (FL) performance using images with varied percentage of low frequencies fully collected in 5× MRI undersampling, ranging from 0% (aliasing) to 20% (blurring). We used a modified EfficientNet-B1 architecture trained to directly predict coordinates (FLNet) as well as using a biologically inspired V1Block, which incorporates Gabor filters to mimic early stages of human visual processing as a preprocessing to FLNet (V1FLNet). When trained on images with 20% of low frequencies collected, both neural networks surpassed average human performance but diverged from human-like performance on out-of-distribution data. Training on both 0% and 20% data, leading to interpolation in the generalization and more training data led to better results. Training on all the conditions resulted in the best results which also matched human performance and would lead to the same choice of which low k-space frequencies to collect as the human observer studies. The adaptive models, trained and evaluated across each of the seven conditions, performed better than the model trained on one condition but not as well as the model trained on all conditions. V1FLNet had similar performance as FLNet especially for the model that used all conditions for training.</p>","PeriodicalId":74505,"journal":{"name":"Proceedings of SPIE--the International Society for Optical Engineering","volume":"13928 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13383116/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148610969","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Combining optimal biomarkers and deep learning for multimodal presurgical prediction of incisional hernia.","authors":"Yubing Tong, Lipeng Xie, Ruben U Fernandez, Tiange Liu, Robyn B Broach, Drew A Torigian, John P Fischer, Jayaram K Udupa","doi":"10.1117/12.3086832","DOIUrl":"https://doi.org/10.1117/12.3086832","url":null,"abstract":"<p><p>Despite its significant clinical and economic burden, the underlying pathophysiology of incisional hernia (IH) remains poorly understood. Imaging-derived features have shown promise in differentiating patients with and without IH. Yet, earlier studies primarily relied on conventional statistical analyses or basic machine learning models applied to relatively small, matched case-control cohorts. In this study, we substantially strengthen and expand upon prior research by analyzing abdominopelvic CT images from a larger cohort of 420 patients (157 IH, 263 non-IH). We introduce a hybrid intelligence (HI) framework that integrates natural intelligence (NI) and artificial intelligence (AI). This approach enables the embedding of hand-crafted features into deep learning (DL) models for prediction, referred to as hybrid deep learning (h-DL). The OBM approach utilizes 34 clinically related parameters alongside 1,065 image-based features. The primary objective of this study is to develop an HI strategy that combines OBM's efficient feature selection with DL's predictive power for improved IH prediction. While the DL model alone outperformed OBM, hybrid models-particularly h-DL-achieved even higher accuracy. This versatile methodology is broadly applicable and can be readily adapted to other biomarker-driven clinical prediction tasks.</p>","PeriodicalId":74505,"journal":{"name":"Proceedings of SPIE--the International Society for Optical Engineering","volume":"13927 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13681200/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149240460","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Monocular absolute depth estimation from endoscopy via domain-invariant feature learning and latent consistency.","authors":"Hao Li, Daiwei Lu, Jesse d'Almeida, Dilara Isik, Ehsan Khodapanah Aghdam, Nick DiSanto, Ayberk Acar, Susheela Sharma, Jie Ying Wu, Robert J Webster, Ipek Oguz","doi":"10.1117/12.3086626","DOIUrl":"10.1117/12.3086626","url":null,"abstract":"<p><p>Monocular depth estimation (MDE) is a critical task to guide autonomous medical robots. However, obtaining absolute (metric) depth from an endoscopy camera in surgical scenes is difficult, which limits supervised learning of depth on real endoscopic images. Current image-level unsupervised domain adaptation methods translate synthetic images with known depth maps into the style of real endoscopic frames and train depth networks using these translated images with their corresponding depth maps. However a domain gap often remains between real and translated synthetic images. In this paper, we present a latent feature alignment method to improve absolute depth estimation by reducing this domain gap in the context of endoscopic videos of the central airway. Our methods are agnostic to the image translation process and focus on the depth estimation itself. Specifically, the depth network takes translated synthetic and real endoscopic frames as input and learns latent domain-invariant features via adversarial learning and directional feature consistency. The evaluation is conducted on endoscopic videos of central airway phantoms with manually aligned absolute depth maps. Compared to state-of-the-art MDE methods, our approach achieves superior performance on both absolute and relative depth metrics, and consistently improves results across various backbones and pretrained weights. Our code is available at https://github.com/MedICL-VU/MDE.</p>","PeriodicalId":74505,"journal":{"name":"Proceedings of SPIE--the International Society for Optical Engineering","volume":"13925 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13441291/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148681262","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}