Journal of Medical Imaging最新文献

{"title":"Comparison of chest X-ray radiography AI model to comorbidities for predicting intensive care unit admission for COVID-19.","authors":"Heather M Whitney, Hui Li, Karen Drukker, Maryellen L Giger","doi":"10.1117/1.JMI.13.4.044501","DOIUrl":"10.1117/1.JMI.13.4.044501","url":null,"abstract":"<p><strong>Purpose: </strong>There are practical limitations to using comorbidities alone in predicting the need for admission to the intensive care unit (ICU). We compared the classification performance of a deep learning/artificial intelligence (DL/AI) model on chest X-ray radiography (CXR) for predicting admission of a patient to the ICU to using two common comorbidity indices, using data collected during the COVID-19 pandemic as a use case.</p><p><strong>Approach: </strong>CXR imaging studies and clinical data of patients who tested positive for COVID-19 between February 2020 and January 2022 were retrospectively collected, yielding 8357 CXR imaging studies from 5046 patients. Classification performance by a DL/AI model in the task of predicting ICU admission within 24 h of imaging was compared to (a) the Charlson comorbidity index (CCI) and (b) the age-adjusted version of the Charlson comorbidity index (ACCI) using the area under the receiver operating characteristic curve (AUC). The AUC from each comorbidity index was compared with the DL/AI model, with a Bonferroni-corrected significance level of <math><mrow><mi>P</mi> <mo>=</mo> <mn>0.025</mn></mrow> </math> .</p><p><strong>Results: </strong>The prediction of ICU admission using the DL/AI model (median AUC [95% CI]: 0.78, [0.74, 0.81]) demonstrated statistical superiority to using the CCI and ACCI comorbidity indices with improvements in AUC of 0.16 ( <math><mrow><mi>P</mi> <mo><</mo> <mn>0.001</mn></mrow> </math> ) and 0.12 ( <math><mrow><mi>P</mi> <mo><</mo> <mn>0.001</mn></mrow> </math> ), respectively.</p><p><strong>Conclusions: </strong>A DL/AI model on CXR for predicting ICU admission within 24 h of imaging obtained superior performance compared with two clinical comorbidity indices in a use case. This work serves as a use case to demonstrate the potential for some medical imaging deep learning models to help improve patient care and resource planning for ICU departments.</p>","PeriodicalId":47707,"journal":{"name":"Journal of Medical Imaging","volume":"13 4","pages":"044501"},"PeriodicalIF":2.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13340710/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148406491","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":"CoReg-Net: contrastive registration for structurally consistent one-shot medical image segmentation.","authors":"Ziyi Xia, Yongzhi Huang, Feng Zhou, Hongwei Zhang, Yun Peng, Yong Liu, Liyun Tu","doi":"10.1117/1.JMI.13.4.044002","DOIUrl":"https://doi.org/10.1117/1.JMI.13.4.044002","url":null,"abstract":"<p><strong>Purpose: </strong>In clinical scenarios where only a single labeled reference image is available for the target segmentation task, registration-based one-shot segmentation is a practical solution through anatomical label transfer. However, existing deep learning-based registration methods are often sensitive to appearance variations, which distort anatomical correspondence in the latent feature space and lead to structurally inconsistent deformation fields. This work aims to improve the structural consistency of deformable registration by explicitly aligning inter-subject anatomical representations.</p><p><strong>Approach: </strong>We propose CoReg-Net, a contrastive registration framework that integrates an inter-subject contrastive learning (ISCL) module to enforce anatomically consistent feature representations by matching spatially corresponding regions across subjects. In parallel, a cascaded skip connection (CSC) mechanism is introduced to bridge the semantic gap between hierarchical encoder and decoder features, enabling more coherent reconstruction of fine-grained deformation details. These two components are jointly optimized to guide the network toward structurally meaningful and topologically regular deformation fields for robust label propagation.</p><p><strong>Results: </strong>The proposed method demonstrates robust and consistent one-shot segmentation performance across diverse anatomical structures and imaging modalities on the IXI, OASIS, and BCV datasets. Compared with state-of-the-art registration-based baselines, CoReg-Net generates deformation fields with higher structural similarity to target images, confirming the effectiveness of contrastive feature alignment. Atlas replacement experiments further verify the robustness of the proposed method under atlas variation. Moreover, the framework exhibits adaptive deformability, preserving topological regularity in brain tissues while accurately capturing large non-rigid deformations in abdominal organs.</p><p><strong>Conclusions: </strong>CoReg-Net shows that explicitly enforcing structural consistency through contrastive registration improves structural similarity and maintains segmentation accuracy. The proposed framework provides a reliable and generalizable solution for atlas-guided anatomical analysis across heterogeneous imaging modalities. Code is available at https://github.com/RelMono/CoRegNet.</p>","PeriodicalId":47707,"journal":{"name":"Journal of Medical Imaging","volume":"13 4","pages":"044002"},"PeriodicalIF":2.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13468818/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148762333","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":"OCR-mediated modality dominance in vision-language models: implications for radiology AI trustworthiness.","authors":"Izzet Turkalp Akbasli, Baris Ozturk, Oguzhan Serin, Volkan Dogan, Göksu Bozdereli Berikol, Donnella S Comeau, Leo Anthony Celi, Orhan Ozguner","doi":"10.1117/1.JMI.13.4.045501","DOIUrl":"10.1117/1.JMI.13.4.045501","url":null,"abstract":"<p><strong>Purpose: </strong>Vision-language models (VLMs) are increasingly proposed for radiologic decision support, yet the security implications of deploying optical character recognition (OCR)-capable models in diagnostic workflows remain poorly characterized. When image-embedded text is not treated as untrusted input, the visual channel becomes vulnerable to adversarial manipulation.</p><p><strong>Approach: </strong>Ten VLMs, none validated for clinical diagnosis, were evaluated on 600 brain magnetic resonance imaging studies for binary tumor detection under 5 conditions: clean input, visible radiology report injection, human-imperceptible stealth OCR injection, and multi-stage immune-prompt defense. A total of 27,000 inference calls were analyzed.</p><p><strong>Results: </strong>At baseline, performance was heterogeneous, with a median accuracy of 0.69, a sensitivity of 0.79, and a specificity of 0.59. Visible injection caused universal specificity collapse to 0.00 across all models with a false-positive rate (FPR) of 1.00 and a median attack success rate (ASR) of 0.97. Stealth injection, despite being imperceptible to human reviewers, drove substantial degradation with a median accuracy of 0.43, an ASR of 0.57, and an FPR of 0.84. Immune prompting achieved only partial mitigation: under stealth injection, median ASR decreased to 0.44, and accuracy improved to 0.56, yet residual overcalling persisted with a median FPR of 0.67, and three models maintained an FPR of 1.00.</p><p><strong>Conclusions: </strong>Commercial VLMs exhibit a deployment-critical failure mode in radiology-like scenarios: OCR-readable text embedded in images can override pixel-level evidence, even under stealth conditions that evade human inspection. Prompt-level defenses provide insufficient protection. Any clinical integration of VLMs must be governed by system-level safeguards, including OCR-aware input handling, provenance controls, and enforced human verification, before deployment in safety-sensitive environments.</p>","PeriodicalId":47707,"journal":{"name":"Journal of Medical Imaging","volume":"13 4","pages":"045501"},"PeriodicalIF":2.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13387900/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148551011","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":"Strengthening the JMI-SPIE Medical Imaging Connection.","authors":"Bennett A Landman, Ivana Išgum","doi":"10.1117/1.JMI.13.4.040101","DOIUrl":"10.1117/1.JMI.13.4.040101","url":null,"abstract":"<p><p>The editorial announces two pilot initiatives linking SPIE Medical Imaging and the Journal of Medical Imaging: a journal-first conference presentation pathway for recently published JMI papers and a streamlined route for selected conference papers to progress to journal publication while preserving rigorous peer review.</p>","PeriodicalId":47707,"journal":{"name":"Journal of Medical Imaging","volume":"13 4","pages":"040101"},"PeriodicalIF":2.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13452402/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148698170","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":"Beyond accuracy: a systematic scoping review of algorithmic bias measurement and mitigation in peer-reviewed publications in medical imaging AI development and evaluation for classification.","authors":"Sara C Chaker, Sahil Sethi, Heather M Whitney","doi":"10.1117/1.JMI.13.4.040901","DOIUrl":"10.1117/1.JMI.13.4.040901","url":null,"abstract":"<p><strong>Purpose: </strong>The rapid development of medical imaging artificial intelligence (AI) models has raised concerns over ensuring that the outputs of these models, when deployed in clinical environments, provide unbiased classifications across different patient populations or operational factors. This scoping review aimed to investigate how the current literature evaluates algorithmic bias in medical imaging AI models using quantitative fairness metrics.</p><p><strong>Approach: </strong>Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR), PubMed and Embase were systematically searched for peer-reviewed literature published between January 2015 and May 2025. Studies that evaluated classification medical imaging AI models for algorithmic bias and reported at least one quantitative measure of fairness were included. Articles that investigated fairness in nonimaging AI models used qualitative bias mitigation methods, examined technical bias (e.g., systematic measurement or preprocessing), or were conference proceedings were excluded.</p><p><strong>Results: </strong>After removing duplicates, 2557 studies were included for title and abstract screening. Following initial screening, 143 articles were retrieved for full-text review. Of these, eight studies were eligible for inclusion. Half of the articles focused on bias detection, whereas the other half investigated mitigation techniques. There was considerable heterogeneity in the bias metrics used across the studies. Nearly all studies ( <math><mrow><mn>7</mn> <mo>/</mo> <mn>8</mn></mrow> </math> ) used a chest X-ray radiography classifier for evaluation and examined algorithmic bias with respect to gender, age, and race.</p><p><strong>Conclusions: </strong>The relative paucity of existing literature and variability in fairness evaluation methods underscore the need for more standardized frameworks to evaluate and ensure fairness in imaging AI systems.</p>","PeriodicalId":47707,"journal":{"name":"Journal of Medical Imaging","volume":"13 4","pages":"040901"},"PeriodicalIF":2.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13537098/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148882119","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":"Fairness-aware deep learning for predicting visual field loss from optical coherence tomography.","authors":"Shreeya Pandey, Leila Gheisi, Yu Huang, Maohua Lin, Hao Zheng, Min Shi","doi":"10.1117/1.JMI.13.4.044502","DOIUrl":"10.1117/1.JMI.13.4.044502","url":null,"abstract":"<p><strong>Purpose: </strong>Standard automated perimetry is the clinical standard for measuring visual field (VF) loss in glaucoma, but it is subjective, time-consuming, and variable across patient visits. In contrast, optical coherence tomography (OCT) provides fast and reproducible structural measurements, creating an opportunity to estimate functional visual field loss directly from imaging. We developed and evaluated a fairness-aware volumetric deep learning framework that integrates an adaptive fairness feedback (AFF) mechanism with three-dimensional (3D) OCT modeling to jointly improve predictive accuracy and demographic equity for 52-point visual field prediction.</p><p><strong>Approach: </strong>Using 3300 paired OCT-VF samples from the Harvard-GF dataset with a patient-level split, we trained a 3D ResNet-18 model incorporating demographic embeddings and an AFF mechanism that dynamically adjusts learning rates based on training subgroup error disparities. Model selection was performed using an internal validation subset of 300 samples. Performance was evaluated using mean absolute error (MAE), root mean squared error (RMSE), <math> <mrow><msup><mi>R</mi> <mn>2</mn></msup> </mrow> </math> , equity-scaled metrics, subgroup disparity analysis, pointwise spatial evaluation, and bootstrap resampling.</p><p><strong>Results: </strong>The fairness-aware model with demographic embeddings and AFF achieved RMSE 5.21 dB, MAE 3.32 dB, and <math> <mrow><msup><mi>R</mi> <mn>2</mn></msup> </mrow> </math> 0.56 on the held-out test set. Compared with a 3D baseline without demographic embedding or AFF, the full model reduced worst-group MAE from 3.90 to 3.53 dB and MAE disparity from 0.54 to 0.34 dB across racial subgroups, corresponding to a 37% reduction in racial MAE disparity while simultaneously improving overall accuracy.</p><p><strong>Conclusions: </strong>Fairness-aware volumetric modeling with AFF improved OCT-based visual field prediction accuracy and reduced racial performance disparities. Ethnicity-based results should be interpreted cautiously because the Hispanic subgroup was small.</p>","PeriodicalId":47707,"journal":{"name":"Journal of Medical Imaging","volume":"13 4","pages":"044502"},"PeriodicalIF":2.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13355556/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148425259","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":"Evaluating federated learning approaches for mammography under breast density heterogeneity.","authors":"Gonzalo Iñaki Quintana, Franco Martin Di Maria, Laurence Vancamberg","doi":"10.1117/1.JMI.13.4.044504","DOIUrl":"https://doi.org/10.1117/1.JMI.13.4.044504","url":null,"abstract":"<p><strong>Purpose: </strong>Breast density is a key factor that influences mammography interpretation and is a major source of heterogeneity in multicenter datasets. Such heterogeneity poses challenges for collaborative machine learning across institutions, particularly in federated learning (FL). We aim to evaluate the impact of breast density-induced heterogeneity on FL for mammography image classification and to assess the robustness of common FL algorithms in realistic clinical settings.</p><p><strong>Approach: </strong>We conducted experiments on mammography datasets under two scenarios: (1) a strongly heterogeneous setting where each participating site contributed exclusively low- or high-density cases, based on the BI-RADS density score, and (2) a population-based setting simulating breast density distributions observed in White and Asian populations. For the strongly heterogeneous setting, we evaluated two configurations: one with two clients, where the cases were grouped as BI-RADS A-B and C-D, and one with four clients, where each site contained cases of a single BI-RADS density. We compared three FL methods (i.e., FedAvg, FedProx, and SCAFFOLD) against centralized training, local-only training, and naïve aggregation approaches, including model ensembling and weight averaging.</p><p><strong>Results: </strong>Across both scenarios, FL consistently achieved performance comparable to centralized training, whereas local models and naïve aggregation approaches underperformed in the presence of strong heterogeneity. Notably, FedAvg achieved accuracy on par with or exceeding centralized training, demonstrating resilience to breast density-induced data imbalance without requiring specialized heterogeneity mitigation algorithms.</p><p><strong>Conclusions: </strong>These findings show that FL can effectively address breast density-related heterogeneity, supporting its feasibility for real-world mammography workflows. The demonstrated robustness of FedAvg underscores the potential for broad clinical deployment of FL, enabling collaborative model development while maintaining data privacy.</p>","PeriodicalId":47707,"journal":{"name":"Journal of Medical Imaging","volume":"13 4","pages":"044504"},"PeriodicalIF":2.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13481055/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148799751","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":"Multi-view and multi-frame voting-based stenosis detection and quantification in X-ray coronary angiography.","authors":"Antonia Popp, Alaa Abd El Al, Marie Hoffmann, Ann Laube, Jörg Kempfert, Anja Hennemuth, Alexander Meyer","doi":"10.1117/1.JMI.13.4.044505","DOIUrl":"10.1117/1.JMI.13.4.044505","url":null,"abstract":"<p><strong>Purpose: </strong>The visual assessment of X-ray coronary angiography (XCA) videos is challenging due to the complex and dynamic structure of the coronary tree. Automated interpretation can improve confidence in identifying stenoses and estimating their severity, thereby supporting clinical decision-making. However, most existing approaches rely on static images or single-view videos, limiting the use of spatial and temporal information.</p><p><strong>Approach: </strong>We propose a multi-step pipeline for automated stenosis detection, localization, and severity estimation in XCA videos. By aggregating findings across multi-view videos, the method classifies stenotic segments throughout the coronary tree and estimates the stenosis severity. The approach is evaluated at the patient level using a dataset of 219 patients. We further compare the severity estimation performance between single- and multi-view projections to assess the importance of multi-view integration.</p><p><strong>Results: </strong>Our approach achieves an <math><mrow><mi>F</mi> <mn>1</mn></mrow> </math> -score of 64.54% for segment-wise stenosis detection and 68.09% for severity classification of correctly detected stenoses. The severity classification improves when stenoses are detected in at least two views. Compared with a state-of-the-art single-view approach, our method demonstrates superior performance in patient-level stenosis detection and severity estimation.</p><p><strong>Conclusions: </strong>The proposed pipeline enables robust stenosis detection, localization, and severity assessment in XCA examinations. The results highlight the importance of incorporating both temporal and multi-view information for reliable patient-level predictions.</p>","PeriodicalId":47707,"journal":{"name":"Journal of Medical Imaging","volume":"13 4","pages":"044505"},"PeriodicalIF":2.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13496233/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148799773","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":"Mask R-CNN-based carotid plaque localization in ultrasound images.","authors":"Maxwell J Kiernan, Rashid Al Mukaddim, Carol C Mitchell, Jenna Maybock, Stephanie M Wilbrand, Robert J Dempsey, Tomy Varghese","doi":"10.1117/1.JMI.13.4.044001","DOIUrl":"10.1117/1.JMI.13.4.044001","url":null,"abstract":"<p><strong>Purpose: </strong>In this study, we present the refinement of a Mask R-CNN model initially designed for carotid lumen detection to automatically generate bounding boxes (BB) enclosing atherosclerotic plaque. Although the model also produces segmentation masks within these bounding boxes, this study primarily evaluates bounding box detection performance as the key step to support segmentation in our ultrasound elastography workflow.</p><p><strong>Approach: </strong>We utilize a PyTorch torchvision implementation of Mask R-CNN for carotid plaque detection and BB placement. Our dataset consists of 118 severe stenotic carotid plaques from patients clinically indicated for carotid endarterectomy. Due to variability plaque presentation, different R-CNN models showed varying results based on the allowed number of prediction regions. An overview analysis of shared predictions from these models showed slight improvement over individual model results.</p><p><strong>Results: </strong>Using bounding box detection as the primary endpoint, we achieved a maximum Dice similarity coefficient (DSC) of 0.74 and intersection over union (IoU) of 0.61 for the best-performing model, with a filtered multimodel approach improving DSC to 0.76. The corresponding plaque mask segmentation performance was lower ( <math><mrow><mi>DSC</mi> <mo>=</mo> <mn>0.64</mn></mrow> </math> , <math><mrow><mi>IoU</mi> <mo>=</mo> <mn>0.49</mn></mrow> </math> ), reflecting the increased difficulty of precise plaque delineation.</p><p><strong>Conclusion: </strong>Due to significant variation in plaque presentation and types among patients, the accuracy of the Plaque Mask R-CNN network would benefit from incorporating additional patient datasets to increase variation in the training data.</p>","PeriodicalId":47707,"journal":{"name":"Journal of Medical Imaging","volume":"13 4","pages":"044001"},"PeriodicalIF":2.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13446570/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148686339","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":"Charge-cloud-based micrometer resolution in deep-silicon photon-counting CT.","authors":"Rickard Brunskog, Mats Persson, Moa Yveborg Tamm, Umberto Follo, Mats Danielsson","doi":"10.1117/1.JMI.13.4.043501","DOIUrl":"10.1117/1.JMI.13.4.043501","url":null,"abstract":"<p><strong>Purpose: </strong>We are developing a monolithic deep-silicon photon-counting sensor targeting spatial resolution on the order of <math><mrow><mn>1</mn> <mtext>  </mtext> <mi>μ</mi> <mi>m</mi></mrow> </math> . This work investigates how pixel pitch, noise level, threshold number, and threshold placement affect the achievable tangential and wafer-thickness resolution to guide the CMOS electronics and sensor design.</p><p><strong>Approach: </strong>Allpix Squared simulations were used to evaluate pixel pitches with two noise levels and two threshold-placement schemes over varying threshold numbers. Interaction position was estimated in both dimensions using lookup tables trained on the thresholded pixel output, and performance was quantified using the MTF.</p><p><strong>Results: </strong>Decreasing pixel pitch strongly improved tangential resolution for Compton interactions, whereas photoelectric interactions showed a weaker dependence on pixel pitch and threshold placement. Equal-counting thresholds approached the achievable resolution with fewer thresholds than equidistant thresholds. For eight equal-counting thresholds, the 10% MTF across the studied pixel sizes ranged from <math><mrow><mtext>1284  </mtext> <mrow><mi>lp</mi></mrow> <mo>/</mo> <mi>cm</mi></mrow> </math> to <math><mrow><mtext>2452  </mtext> <mrow><mi>lp</mi></mrow> <mo>/</mo> <mi>cm</mi></mrow> </math> in the tangential direction and from 15 to <math><mrow><mn>71</mn> <mtext>  </mtext> <mrow><mi>lp</mi></mrow> <mo>/</mo> <mi>cm</mi></mrow> </math> in the wafer-thickness direction for Compton interactions, compared with 409 to <math><mrow><mn>507</mn> <mtext>  </mtext> <mrow><mi>lp</mi></mrow> <mo>/</mo> <mi>cm</mi></mrow> </math> and 18 to <math><mrow><mn>79</mn> <mtext>  </mtext> <mrow><mi>lp</mi></mrow> <mo>/</mo> <mi>cm</mi></mrow> </math> , respectively, for photoelectric interactions. RMSE analysis further showed that some interactions for the <math><mrow><mn>25</mn> <mtext>  </mtext> <mi>μ</mi> <mi>m</mi></mrow> </math> pitch reached below <math><mrow><mn>2</mn> <mtext>  </mtext> <mi>μ</mi> <mi>m</mi></mrow> </math> tangentially and a few tens of micrometers in the wafer-thickness direction.</p><p><strong>Conclusions: </strong>Spatial resolution improved with a threshold number up to about 8 to 10, beyond which additional programmable thresholds gave only modest gain. The achieved tangential resolution also suggests that analyzer-free phase-contrast imaging is feasible for realistic geometries and expected interference patterns.</p>","PeriodicalId":47707,"journal":{"name":"Journal of Medical Imaging","volume":"13 4","pages":"043501"},"PeriodicalIF":2.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13446572/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148686389","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}