Journal of Medical Imaging and Radiation Sciences最新文献

{"title":"Using Deep Learning-based Optical Character Recognition (OCR) AI method to automatically, effectively and accurately de-identify burned-in Protected Health Information (PHI) pixel text in DICOM images","authors":"Hoi Ning LEUNG, Kai Man TO, Wei Kin LI, Wai Ming HUI","doi":"10.1016/j.jmir.2026.102509","DOIUrl":"10.1016/j.jmir.2026.102509","url":null,"abstract":"<div><h3>Objectives</h3><div>Burned-in patient PHI often appears in secondary captured DICOM images, particularly Ultrasound. This project aims to develop and evaluate a deep learning OCR-based tool to automatically remove such information.</div></div><div>Burned-in patient PHI often appears in secondary captured DICOM images, particularly Ultrasound. This project aims to develop and evaluate a deep learning OCR-based tool to automatically remove such information.</div><div><h3>Background</h3><div>De-identifying PHI in DICOM images is essential for clinical research, teaching, and AI training while ensuring compliance with regulations such as HIPAA. Earlier methods combined OCR with whitelists to avoid excessive removal, but fixed databases cannot cover every clinically relevant label or measurement, so valuable information may still be removed. Tsui et al. reported impractical processing times (∼60 s/image), while Macdonald et al. achieved faster speeds (1–2 s/image) but required high-end hardware (RTX 3090). Our project introduces a header-matching approach powered by the latest open-source OCR models to balance accuracy and efficiency.</div></div><div>De-identifying PHI in DICOM images is essential for clinical research, teaching, and AI training while ensuring compliance with regulations such as HIPAA. Earlier methods combined OCR with whitelists to avoid excessive removal, but fixed databases cannot cover every clinically relevant label or measurement, so valuable information may still be removed. Tsui et al. reported impractical processing times (∼60 s/image), while Macdonald et al. achieved faster speeds (1–2 s/image) but required high-end hardware (RTX 3090). Our project introduces a header-matching approach powered by the latest open-source OCR models to balance accuracy and efficiency.</div><div><h3>Methodology</h3><div>The DICOM OCR De-identifier (D_OD) Tool was developed in Visual Studio Code using Python 3.11.9. It employs cnOCR 2.3.2.1 with PaddlePaddle-trained models (<em>en_PP-OCRv3_det_infer.onnx</em> and <em>en_PP-OCRv3_rec_infer.onnx</em>) executed via ONNX Runtime 1.22.1. PyDICOM extracts patient-related text from DICOM headers and pixel data. After preprocessing, header information is compared with OCR outputs; exact matches trigger masking of the corresponding burned-in text by rendering the pixel area black.</div></div><div>The DICOM OCR De-identifier (D_OD) Tool was developed in Visual Studio Code using Python 3.11.9. It employs cnOCR 2.3.2.1 with PaddlePaddle-trained models (<em>en_PP-OCRv3_det_infer.onnx</em> and <em>en_PP-OCRv3_rec_infer.onnx</em>) executed via ONNX Runtime 1.22.1. PyDICOM extracts patient-related text from DICOM headers and pixel data. After preprocessing, header information is compared with OCR outputs; exact matches trigger masking of the corresponding burned-in text by rendering the pixel area black.</div><div><h3>Results</h3><div>Evaluation was conducted using a limited set of dummy patient data ac","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102509"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149004000","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Prognostic biomarker potential of carcinoembryonic antigen for assessing response to neoadjuvant chemoradiotherapy in colorectal cancer","authors":"Shu-Han Yang","doi":"10.1016/j.jmir.2026.102515","DOIUrl":"10.1016/j.jmir.2026.102515","url":null,"abstract":"<div><h3>Background / Purpose</h3><div>Neoadjuvant concurrent chemoradiotherapy (NACRT) is the standard strategy for locally advanced colorectal cancer (CRC), aiming to downstage the tumor and increase resectability. However, some patients endure the toxicity of ineffective treatment. Post-treatment prognosis heavily relies on the degree of tumor response. This study aimed to analyze the association of pre-NACRT Carcinoembryonic Antigen (CEA) and glycemic indices (FPG, HbA1c) with the post-operative Tumor Regression Grade (TRG). Our goal was to identify pre-treatment markers to guide clinical decision-making and reduce surgical risk.</div></div><div>Neoadjuvant concurrent chemoradiotherapy (NACRT) is the standard strategy for locally advanced colorectal cancer (CRC), aiming to downstage the tumor and increase resectability. However, some patients endure the toxicity of ineffective treatment. Post-treatment prognosis heavily relies on the degree of tumor response. This study aimed to analyze the association of pre-NACRT Carcinoembryonic Antigen (CEA) and glycemic indices (FPG, HbA1c) with the post-operative Tumor Regression Grade (TRG). Our goal was to identify pre-treatment markers to guide clinical decision-making and reduce surgical risk.</div><div><h3>Methods</h3><div>This retrospective study, approved by the KMUHIRB (KMUHIRB-E(I)-20240367), included a significant cohort of patients with locally advanced CRC receiving NACRT at our hospital. All patients underwent surgery 6–8 weeks post-NACRT. Patients were grouped by TRG scores (TRG 0, TRG 1, and combined TRG 2+3 for poor response). Clinical variables (CEA, HbA1c, FPG) were collected for statistical correlation analysis.</div></div><div>This retrospective study, approved by the KMUHIRB (KMUHIRB-E(I)-20240367), included a significant cohort of patients with locally advanced CRC receiving NACRT at our hospital. All patients underwent surgery 6–8 weeks post-NACRT. Patients were grouped by TRG scores (TRG 0, TRG 1, and combined TRG 2+3 for poor response). Clinical variables (CEA, HbA1c, FPG) were collected for statistical correlation analysis.</div><div><h3>Results</h3><div>Statistical analysis revealed a strong, significant, and consistent positive correlation between pre-NACRT CEA levels and TRG scores. Higher pre-treatment CEA was associated with a significantly increased likelihood of a poorer therapeutic response (higher TRG score). Conversely, no significant association was found between glycemic indices (FPG, HbA1c) and TRG scores, thus not supporting the hypothesis that hyperglycemia impairs treatment response.</div></div><div>Statistical analysis revealed a strong, significant, and consistent positive correlation between pre-NACRT CEA levels and TRG scores. Higher pre-treatment CEA was associated with a significantly increased likelihood of a poorer therapeutic response (higher TRG score). Conversely, no significant association was found between glycemic indices (FPG, HbA1c) and TRG score","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102515"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149004005","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Evaluation of Automated Script-based Treatment Planning for Stereotactic Body Radiation Therapy of Prostate Cancer","authors":"Ming Kit Joyce Leung","doi":"10.1016/j.jmir.2026.102529","DOIUrl":"10.1016/j.jmir.2026.102529","url":null,"abstract":"<div><h3>Background and Purpose</h3><div>Manual treatment planning in stereotactic body radiation therapy (SBRT) requires significant expertise and human intervention in the iterative optimization cycles, posing different challenges including inter-planner and intra-planner variation, time-consuming and labour-intensive nature, and suboptimal plan quality. An in-house script-based approach has been proposed to automate the optimization process without manual intervention in the treatment planning for SBRT of prostate cancer. This study evaluates this approach to assess its dosimetric acceptability and deliverability for clinical implementation.</div></div><div>Manual treatment planning in stereotactic body radiation therapy (SBRT) requires significant expertise and human intervention in the iterative optimization cycles, posing different challenges including inter-planner and intra-planner variation, time-consuming and labour-intensive nature, and suboptimal plan quality. An in-house script-based approach has been proposed to automate the optimization process without manual intervention in the treatment planning for SBRT of prostate cancer. This study evaluates this approach to assess its dosimetric acceptability and deliverability for clinical implementation.</div><div><h3>Methods</h3><div>Twenty-six patients who received SBRT for prostate cancer were retrospectively selected. Script-based plans were generated with the proposed script, and were assessed and compared with manual plans. Quantitatively, plan quality was evaluated in terms of planning target volume (PTV) coverage and sparing of organs-at-risk (OARs). Plan deliverability was assessed by 3D gamma passing rates and monitor units. Plan consistency was also compared between different planning approaches. Qualitatively, a blind pairwise comparison by experienced radiation dosimetrists was conducted to select the clinically preferred plans for each case.</div></div><div>Twenty-six patients who received SBRT for prostate cancer were retrospectively selected. Script-based plans were generated with the proposed script, and were assessed and compared with manual plans. Quantitatively, plan quality was evaluated in terms of planning target volume (PTV) coverage and sparing of organs-at-risk (OARs). Plan deliverability was assessed by 3D gamma passing rates and monitor units. Plan consistency was also compared between different planning approaches. Qualitatively, a blind pairwise comparison by experienced radiation dosimetrists was conducted to select the clinically preferred plans for each case.</div><div><h3>Results</h3><div>All script-based plans fulfilled PTV and critical OARs (rectum, sigmoid and bladder) dose constraints, and passed 3D gamma analysis. Compared with manual plans, script-based plans demonstrated superiority in PTV conformity, dose spillage, sparing of bladder, femurs and urethra, plan consistency, delivery accuracy and efficiency, but inferiority in PTV coverage and sparing ","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102529"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149004304","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Fluoroscopy in Action: Transforming Paediatric Cardiac Interventions at the Hong Kong Children’s Hospital","authors":"Ying Kit Tse, Tsan Pun Lam","doi":"10.1016/j.jmir.2026.102521","DOIUrl":"10.1016/j.jmir.2026.102521","url":null,"abstract":"<div><h3>Background</h3><div>Service of paediatric cardiac interventions has been provided by the Hong Kong Children's Hospital (HKCH), a tertiary medical center providing specialized cardiology services, since its commencement over the past five years. During the minimally invasive cardiac interventions, x-ray fluoroscopy has emerged as a vital imaging modality to facilitate the patient safety, procedural accuracy and effectiveness. In this presentation, audience could gain insights of the values of x-ray fluoroscopy throughout the paediatric cardiac interventions.</div></div><div>Service of paediatric cardiac interventions has been provided by the Hong Kong Children's Hospital (HKCH), a tertiary medical center providing specialized cardiology services, since its commencement over the past five years. During the minimally invasive cardiac interventions, x-ray fluoroscopy has emerged as a vital imaging modality to facilitate the patient safety, procedural accuracy and effectiveness. In this presentation, audience could gain insights of the values of x-ray fluoroscopy throughout the paediatric cardiac interventions.</div><div><h3>Purpose</h3><div>We will closely examine the technical aspects of fluoroscopic imaging, emphasizing how real-time visualization aids in precise catheter navigation and device placement. The common paediatric cardiac interventions in the HKCH are mainly of four categories: (1) balloon angioplasty for thoracic structures, such as aorta and cardiac valves; (2) stent placements in the thoracic vessels; (3) closures of defects, such as patent ductus arteriosus (PDA), atrial septal defects (ASD), and ventricular septal defects (VSD); and (4) pacemaker implantation.</div></div><div>We will closely examine the technical aspects of fluoroscopic imaging, emphasizing how real-time visualization aids in precise catheter navigation and device placement. The common paediatric cardiac interventions in the HKCH are mainly of four categories: (1) balloon angioplasty for thoracic structures, such as aorta and cardiac valves; (2) stent placements in the thoracic vessels; (3) closures of defects, such as patent ductus arteriosus (PDA), atrial septal defects (ASD), and ventricular septal defects (VSD); and (4) pacemaker implantation.</div><div><h3>Outcome</h3><div>Continuous imaging during interventions allows immediate assessment of procedural progress, enhances clinician confidence, and minimizes complications. The use of fluoroscopy in these procedures offers numerous benefits, including decreased morbidity, shorter recovery times, and improved prognoses for young patients.</div></div><div>Continuous imaging during interventions allows immediate assessment of procedural progress, enhances clinician confidence, and minimizes complications. The use of fluoroscopy in these procedures offers numerous benefits, including decreased morbidity, shorter recovery times, and improved prognoses for young patients.</div><div><h3>Conclusion</h3><div>By ","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102521"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149003884","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Quality assurance program for Dual Energy X-ray Bone Densitometry (DXA): Long-term results from a single institution","authors":"Monica S.Y. Chu, B.F. Yeung, Keith Kwan, Cathy Cheung, Martin Law, Charles Chow, Vivian Ma","doi":"10.1016/j.jmir.2026.102506","DOIUrl":"10.1016/j.jmir.2026.102506","url":null,"abstract":"<div><h3>Background</h3><div>Dual-energy X-ray absorptiometry (DXA) is the gold standard to diagnose osteoporosis and monitor long term treatment response in adults. DXA applications in pediatric patients include primary bone disorders, prolonged immobility or cerebral palsy and endocrine disorders. Periodic follow up scans are required that demand a robust quality assurance (QA) program on bone densitometer. We are going to describe the QA program for a DXA scanner capable of performing whole body scan (WBS) and to discuss the long term QA results.</div></div><div>Dual-energy X-ray absorptiometry (DXA) is the gold standard to diagnose osteoporosis and monitor long term treatment response in adults. DXA applications in pediatric patients include primary bone disorders, prolonged immobility or cerebral palsy and endocrine disorders. Periodic follow up scans are required that demand a robust quality assurance (QA) program on bone densitometer. We are going to describe the QA program for a DXA scanner capable of performing whole body scan (WBS) and to discuss the long term QA results.</div><div><h3>Methods</h3><div>Daily spine phantom QA results of bone mineral density (BMD) and weekly body composition calibration have been collected by using the vendor program since machine installation. Since the year 2025, WBS has been performed to measure patient table top uniformity. Weekly WBS was performed on a 5-year old humanoid phantom to obtain BMD results over the previous three months.</div></div><div>Daily spine phantom QA results of bone mineral density (BMD) and weekly body composition calibration have been collected by using the vendor program since machine installation. Since the year 2025, WBS has been performed to measure patient table top uniformity. Weekly WBS was performed on a 5-year old humanoid phantom to obtain BMD results over the previous three months.</div><div><h3>Results</h3><div>The average spine phantom BMD is obtained as 0.884±0.003 g/cm<sup>2</sup>. (specification: 0.88 g/cm<sup>2</sup>), showing consistent long term machine performance. Image uniformity, in terms of average count ± SD of the patient table top image, has been measured as 523.0±0.1 counts in the year 2025. Time trends for these measurable parameters show that consistent performance of the system can be maintained by using simple and yet clinical applicable QA phantoms.</div></div><div>The average spine phantom BMD is obtained as 0.884±0.003 g/cm<sup>2</sup>. (specification: 0.88 g/cm<sup>2</sup>), showing consistent long term machine performance. Image uniformity, in terms of average count ± SD of the patient table top image, has been measured as 523.0±0.1 counts in the year 2025. Time trends for these measurable parameters show that consistent performance of the system can be maintained by using simple and yet clinical applicable QA phantoms.</div><div><h3>Conclusion</h3><div>Time series of measurable parameters are useful to predict and maintain an optimized syst","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102506"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149004273","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Reframing inclusive pregnancy status check in radiography after the UK Supreme Court ruling on the Equality Act 2010","authors":"Kathy Pang","doi":"10.1016/j.jmir.2026.102502","DOIUrl":"10.1016/j.jmir.2026.102502","url":null,"abstract":"<div><h3>Background</h3><div>Traditionally, all female patients of childbearing age (12 -55 years) are asked about the possibility of pregnancy and their last menstrual dates in pelvic, abdominal and lumbar radiographic procedures to avoid unnecessary radiation exposure to the unborn fetus. However, this gender-based screening excludes individuals capable of pregnancy but are not recorded as female such as transmen and non-binary individuals. Electronic patient systems in many countries, including the United Kingdom and Asian countries record patient gender from identification documents which leads to a loophole in patient safety practice.</div></div><div>Traditionally, all female patients of childbearing age (12 -55 years) are asked about the possibility of pregnancy and their last menstrual dates in pelvic, abdominal and lumbar radiographic procedures to avoid unnecessary radiation exposure to the unborn fetus. However, this gender-based screening excludes individuals capable of pregnancy but are not recorded as female such as transmen and non-binary individuals. Electronic patient systems in many countries, including the United Kingdom and Asian countries record patient gender from identification documents which leads to a loophole in patient safety practice.</div><div><h3>Background</h3><div>In 2021, the Society of Radiographers (SoR) published guidelines for the Inclusive Pregnancy Status (IPS) check, which all patients aged 12-55 are asked about potential pregnancy irrespective of their gender and variations in sex characteristics. This safer approach ensures all individuals have provided critical information and disclosure for informed clinical decision making. However, its implementation is inconsistent across clinical settings due to its non-statutory nature.</div></div><div>In 2021, the Society of Radiographers (SoR) published guidelines for the Inclusive Pregnancy Status (IPS) check, which all patients aged 12-55 are asked about potential pregnancy irrespective of their gender and variations in sex characteristics. This safer approach ensures all individuals have provided critical information and disclosure for informed clinical decision making. However, its implementation is inconsistent across clinical settings due to its non-statutory nature.</div><div><h3>Results</h3><div>In April 2025, the UK Supreme Court ruled that the definitions of “sex”, “man” and “woman” in the Equality Act 2010 refer to biological rather than certified sex. This legal clarification reinforces the implementation of anatomy-based IPS screening to safeguard patient safety and confidentiality.</div></div><div>In April 2025, the UK Supreme Court ruled that the definitions of “sex”, “man” and “woman” in the Equality Act 2010 refer to biological rather than certified sex. This legal clarification reinforces the implementation of anatomy-based IPS screening to safeguard patient safety and confidentiality.</div><div><h3>Conclusion</h3><div>This proposal adopts a sta","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102502"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149004362","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Quantitative Assessment of Ultra-High Resolution Photon-Counting CT on Deep Learning Contouring","authors":"Sze Ting Wong, Ka Fai Cheng, Wing Ki Fung, Sau Yi Candy Fong, George Chiu","doi":"10.1016/j.jmir.2026.102533","DOIUrl":"10.1016/j.jmir.2026.102533","url":null,"abstract":"<div><h3>Background / Purpose</h3><div>Photon-counting CT (PCCT) has emerged as a transformative imaging technology, with growing interest for integration into radiotherapy workflows. While PCCT with ultra-high resolution (UHR) provides enhanced image detail, its impact on auto-contouring for radiotherapy planning remains underexplored. This study investigates whether UHR significantly alters deep learning (DL) contouring performance across thoracic organs-at-risk (OARs) compared to standard resolution (SR) input.</div></div><div>Photon-counting CT (PCCT) has emerged as a transformative imaging technology, with growing interest for integration into radiotherapy workflows. While PCCT with ultra-high resolution (UHR) provides enhanced image detail, its impact on auto-contouring for radiotherapy planning remains underexplored. This study investigates whether UHR significantly alters deep learning (DL) contouring performance across thoracic organs-at-risk (OARs) compared to standard resolution (SR) input.</div><div><h3>Methods</h3><div>Ten thoracic PCCT scans were acquired and reconstructed at SR (512 × 512) and UHR (1024 × 1024). For both resolutions, DL-generated and manual expert contours were created for 16 OARs encompassing diverse anatomical characteristics. Geometric agreement was assessed using Dice Similarity Coefficient (DSC), Mean Distance to Agreement (MDA), and 95th percentile Hausdorff Distance (HD95). Auto-versus-manual contours and SR-manual versus UHR-manual contours were compared across resolutions.</div></div><div>Ten thoracic PCCT scans were acquired and reconstructed at SR (512 × 512) and UHR (1024 × 1024). For both resolutions, DL-generated and manual expert contours were created for 16 OARs encompassing diverse anatomical characteristics. Geometric agreement was assessed using Dice Similarity Coefficient (DSC), Mean Distance to Agreement (MDA), and 95th percentile Hausdorff Distance (HD95). Auto-versus-manual contours and SR-manual versus UHR-manual contours were compared across resolutions.</div><div><h3>Results</h3><div>Organ-specific sensitivity to resolution was observed. Large homogeneous structures—such as lungs and liver—demonstrated near-identical contours across resolutions (mean DSC >0.995; MDA <0.126mm). Small or anatomically complex OARs revealed pronounced differences: the left anterior descending artery (LAD) and proximal bronchial tree (PBT) showed degraded agreement (DSC: 0.926±0.038 and 0.942±0.106; HD95: 10.6mm and 78.3mm).</div><div>For DL-versus-manual comparisons, the heart, LAD, PBT, stomach and trachea consistently showed mean HD95 >5mm across both resolutions (range: 8.2–86.5mm), with suboptimal DSC (0.4–0.8) and MDA (2.9–16.9mm). No statistically significant improvement in DL contouring was observed with UHR images.</div></div><div>Organ-specific sensitivity to resolution was observed. Large homogeneous structures—such as lungs and liver—demonstrated near-identical contours across resolutions (","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102533"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149004453","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A five-year retrospective review of lateral elbow radiograph image quality in AED X-ray at United Christian Hospital","authors":"Ka Hei Shek, Man Hin Lee","doi":"10.1016/j.jmir.2026.102516","DOIUrl":"10.1016/j.jmir.2026.102516","url":null,"abstract":"<div><h3>Background</h3><div>A true lateral elbow radiograph is crucial for diagnosis of trauma. A recurring issue of suboptimal lateral elbow image quality in the Accident & Emergency Department (AED) was raised. Multiple factors are identified and primarily attributed to the limited height range of the examination couch. This limitation compromised the positioning of true lateral view, reducing image quality and diagnostic confidence. Custom-built soft pads were introduced in February 2022 to facilitate positioning.</div></div><div>A true lateral elbow radiograph is crucial for diagnosis of trauma. A recurring issue of suboptimal lateral elbow image quality in the Accident & Emergency Department (AED) was raised. Multiple factors are identified and primarily attributed to the limited height range of the examination couch. This limitation compromised the positioning of true lateral view, reducing image quality and diagnostic confidence. Custom-built soft pads were introduced in February 2022 to facilitate positioning.</div><div><h3>Purpose</h3><div>To evaluate the image quality of lateral elbow radiographs in AED X-ray at United Christian Hospital.</div></div><div>To evaluate the image quality of lateral elbow radiographs in AED X-ray at United Christian Hospital.</div><div><h3>Methods</h3><div>A retrospective review of lateral elbow radiographs from AED of UCH (2020-2024) was conducted. Annually, 170 randomly selected cases (total n=850) were independently reviewed by two radiographers against four criteria for a true lateral projection (90° flexion, superimposition of epicondyles, alignment of radial head and coronoid process, & technical exposure). Pass/fail rates were calculated for overall positioning and each criterion.</div></div><div>A retrospective review of lateral elbow radiographs from AED of UCH (2020-2024) was conducted. Annually, 170 randomly selected cases (total n=850) were independently reviewed by two radiographers against four criteria for a true lateral projection (90° flexion, superimposition of epicondyles, alignment of radial head and coronoid process, & technical exposure). Pass/fail rates were calculated for overall positioning and each criterion.</div><div><h3>Results</h3><div>The overall pass rate for true lateral positioning demonstrated a marked and sustained increase, rising from 29.9% in 2021 to 46.2% following the 2022 intervention, and further to 68.8% in 2024. Improvements in criteria were also observed.</div></div><div>The overall pass rate for true lateral positioning demonstrated a marked and sustained increase, rising from 29.9% in 2021 to 46.2% following the 2022 intervention, and further to 68.8% in 2024. Improvements in criteria were also observed.</div><div><h3>Conclusion</h3><div>The introduction of custom-built soft pads significantly improved the lateral elbow image quality in a challenging AED environment. While patient-related factors remain, this intervention effectively minimized ","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102516"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149004505","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Implementing quantitative MRI: RTT requirements for successful translation from research to clinic","authors":"Lisa McDaid, Michael Dubec, Catherine Parry, Benedict Dobby, Claire Nelder, Rebecca Riding, James Tallon, Ananya Choudhury, Robert Chuter, Peter Hoskin, Cynthia L Eccles","doi":"10.1016/j.jmir.2026.102385","DOIUrl":"10.1016/j.jmir.2026.102385","url":null,"abstract":"<div><h3>Purpose/Aim</h3><div>Successful translation of quantitative MRI (qMRI) into real-time radiotherapy requires standardised, repeatable MRI pulse sequences and highly consistent imaging practice, as quantitative imaging biomarkers are sensitive to protocol variation. This work aimed to identify training, competency and governance requirements for therapeutic radiographers acquiring qMRI on both an MR-guided linear accelerator (MR-Linac) and an MR Imaging system used exclusively for treatment planning (MR Sim).</div></div><div><h3>Methods/Process</h3><div>To define the training, competency and governance requirements needed for reliable qMRI acquisition, deviations from the standardised qMRI protocol were first identified by reviewing any notes to file. All qMRI imaging is performed as part of a clinical trial at our institute, and it is a good clinical practice requirement that protocol deviations are recorded formally. These deviations included incomplete or incorrect lesion coverage and mis-timed dynamic acquisitions of intravenous Gadolinium-based contrast agents (GBCA). Semi-structured focus groups were then conducted with seven therapeutic radiographers routinely involved in qMRI acquisition, and data were analysed using Braun and Clarke’s thematic analysis framework to identify key learning needs and barriers to implementation(1). Four core themes emerged: foundational knowledge, technical skills, workflow integration and clinical governance. Gaps included understanding of qMRI sequence parameters and sources of variability that may affect biomarker reliability, safe delivery of GBCA and hyperoxic gas, and confidence in escalating image-quality concerns. Workflow challenges related to pathway alignment, role clarity and interdisciplinary communication, with limited therapeutic radiographer involvement in early sequence and pathway development reported as a recurring variable. Findings informed revision of the therapeutic radiographer competency framework and establishment of closer collaboration with MR clinical scientists to support bidirectional learning.</div></div><div><h3>Results or Benefits/Challenges</h3><div>The small focus groups created a safe environment for radiographers to share experiences and perceptions, enabling richer exploration of learning needs and perceived barriers. However, potential bias from group dynamics was recognised, and deliberate use of small groups was intended to mitigate this. Enhancing knowledge and communication is expected to increase therapeutic radiographer autonomy and meaningful involvement in future data synthesis.</div></div><div><h3>Conclusions/Impact</h3><div>Therapeutic radiographer competence is a critical determinant of qMRI reliability and validation on both the MR-Linac and MR Sim. A reflexive, systematic approach has produced targeted training aligned with known sources of imaging biomarker variability and includes synchronous and asynchronous learning. It has strengthened interdisc","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 4","pages":"Article 102385"},"PeriodicalIF":2.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148565125","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"New ‘Gizmos’: Use of an AI Tool as an Academic Resource for Learners in the Medical Radiation Sciences","authors":"Natalie Rozanec, Kieng Tan","doi":"10.1016/j.jmir.2026.102362","DOIUrl":"10.1016/j.jmir.2026.102362","url":null,"abstract":"<div><h3>Purpose/Aim</h3><div>As use of artificial intelligence (AI) continues to grow, it is increasingly important that new health professional graduates are equipped to use AI effectively. One recommended way to introduce AI into educational spaces is to use a scaffolded approach, where early learners are directed on how to use AI, gradually developing independence and greater skills as they progress. This initiative introduced a free AI-based study application tool into a second-entry program’s university-level physics course as a resource for student success, to examine the feasibility on how this can lay the foundation for further scaffolded use of AI in an educational program.</div></div><div><h3>Methods/Process</h3><div>A variety of AI learning tools were reviewed and tested. One tool was selected for use, as it has excellent free functionalities, a spaced repetition system to optimize memory retention grounded in scientific learning principles, the ability to generate flashcards, quizzes and lesson plans from a variety of file formats, gamification features, and the ability to create a private class study group. The tool allows learners to try active recall quizzes, work through breakdowns of definitions/concepts, and challenge peers to a game on specific study decks. Additional positive online reviews were noted by students enrolled in allied health professional programs and medical schools. A private class study group, “The ‘Curie-ous’ Corner” was created. A link to join was provided to students through the course web page and was optional. The course instructor created study decks based on all lecture and tutorial materials. These decks were available to students for quiz/exam preparation. The course instructor asked students for informal, anecdotal feedback on the AI tool at multiple timepoints during the course.</div></div><div><h3>Results or Benefits/Challenges</h3><div>By the end of the course, 52/57 enrolled students voluntarily joined “The Curie-ous Corner”. From the instructor perspective, uploading study decks was fast and easy, and the wide variety of formats accepted helped to create study resources from many different types of course materials. The tool provided notifications to the instructor when used by students, which helped monitor usage. Informal anecdotal feedback from students was mostly positive. Students reported the tool was very user-friendly, saving significant time in preparing flashcards/study aids. Several students reported they had started using it to help study for other courses. The only reported challenges with the tool involved a brief period (less than one day) of server issues, during which the tool was inaccessible.</div></div><div><h3>Conclusions/Impact</h3><div>Student engagement with the AI tool was significant, with the instructor receiving usage notifications prior to every quiz and test. While use of this AI tool may have contributed to better overall academic results, a formal analysis could n","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 4","pages":"Article 102362"},"PeriodicalIF":2.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148565538","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}