Journal of Medical Imaging and Radiation Sciences最新文献

{"title":"Evaluating the Dosimetric Impacts of Translational Corrections without Rotational Adjustment during Patient Setup for Lung Stereotactic Body Radiation Therapy Accuracy","authors":"Yuan Te Cheng","doi":"10.1016/j.jmir.2026.102523","DOIUrl":"10.1016/j.jmir.2026.102523","url":null,"abstract":"<div><h3>Background</h3><div>Stereotactic body radiation therapy (SBRT) delivers high-dose, precise radiotherapy, and provides, among many tumor types, excellent local control for early‑stage non‑small cell lung cancer and pulmonary metastases. Accurate patient positioning with 6 degrees-of-freedom (DoF) (3 translational and 3 rotational) is ideal for SBRT due to the steep dose gradients. However, rotational corrections are not always available due to cost or system design (e.g. some ring gantry systems). This study evaluates the dosimetric impact of delivering SBRT on lung cancer patients without rotational corrections.</div></div><div>Stereotactic body radiation therapy (SBRT) delivers high-dose, precise radiotherapy, and provides, among many tumor types, excellent local control for early‑stage non‑small cell lung cancer and pulmonary metastases. Accurate patient positioning with 6 degrees-of-freedom (DoF) (3 translational and 3 rotational) is ideal for SBRT due to the steep dose gradients. However, rotational corrections are not always available due to cost or system design (e.g. some ring gantry systems). This study evaluates the dosimetric impact of delivering SBRT on lung cancer patients without rotational corrections.</div><div><h3>Methods</h3><div>The daily setup corrections for patients treated with SBRT for the lung lesion(s) under breath-hold at our hospital were retrieved. The planning CTs were “back-rotated” and then aligned with the original planning CT to mimic the patient positions without rotational corrections. Dose to the target volumes and organ-at-risk (OAR) in such treatment position was recalculated and compared with the original treatment plan to assess the dosimetric impacts. The rotational errors with our manual patient setup were also analyzed.</div></div><div>The daily setup corrections for patients treated with SBRT for the lung lesion(s) under breath-hold at our hospital were retrieved. The planning CTs were “back-rotated” and then aligned with the original planning CT to mimic the patient positions without rotational corrections. Dose to the target volumes and organ-at-risk (OAR) in such treatment position was recalculated and compared with the original treatment plan to assess the dosimetric impacts. The rotational errors with our manual patient setup were also analyzed.</div><div><h3>Results</h3><div>The mean D<sub>95%</sub> and D<sub>99%</sub> of PTV are 49.92 Gy and 48.16 Gy respectively, compared with the 50.04 Gy and 48.39 Gy of the original plans. D<sub>95%</sub> of 59% of the cases was below the required 50 Gy, and D<sub>99%</sub> of one case was less than the 45 Gy requirement. OAR analysis revealed only limited significance, with one case has D<sub>max</sub> to the rib (52.88 Gy) exceeding the protocol threshold of 52.5 Gy. The rotational errors were relatively small, measuring 0.64° ± 0.43° for pitch, 0.81° ± 0.32° for yaw, and 0.8° ± 0.45° for roll.</div></div><div>The mean D<sub>95%</sub> and D<sub>99%<","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102523"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149003886","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":"Utilities of ¹¹C-MTO Breath-Hold PET/CT for Clinical Assessment of Adrenocortical Lesion in Primary Aldosteronism","authors":"Yuen Kwan Tiffany Chan, Yu Ching Lau, Sirong Chen, Yuk Lau Leung, Chi Lai Ho","doi":"10.1016/j.jmir.2026.102524","DOIUrl":"10.1016/j.jmir.2026.102524","url":null,"abstract":"<div><h3>Background</h3><div>Clinically, lateralization and lesion characterization of adrenocortical lesions using ¹¹C-metomidate (¹¹C-MTO) PET/CT rely on image localization and lesion-to-contralateral SUVmax ratio which are prone to respiratory artefact on routine free-breathing (FB) PET and image noise. This study aimed to investigate the utilities of breath-hold (BH) PET/CT by comparing with FB-PET/CT.</div></div><div>Clinically, lateralization and lesion characterization of adrenocortical lesions using ¹¹C-metomidate (¹¹C-MTO) PET/CT rely on image localization and lesion-to-contralateral SUVmax ratio which are prone to respiratory artefact on routine free-breathing (FB) PET and image noise. This study aimed to investigate the utilities of breath-hold (BH) PET/CT by comparing with FB-PET/CT.</div><div><h3>Methods</h3><div>¹¹C-MTO PET/CT scans performed in our hospital from June 2024 to January 2026 were reviewed. PET started 30 minutes post ¹¹C-MTO injection (0.2mCi/kg body weight). Only patients who were diagnosed with active adrenocortical lesions on FB-PET (10-min acquisition) and could hold single deep inspiration for CT (10 seconds) and then PET acquisition (15-30 seconds) were considered. Lesions were contoured (MIM 6.9.6) on BH-PET, BH-CT, FB-PET and FB-CT images reconstructed in 2mm thickness. For each lesion, Dice similarity coefficient (DSC) between PET and CT contours were determined for BH, and FB. Using lesion volume on BH-CT as the standard reference, absolute percentage difference was calculated for lesion volumes on BH-PET, and FB-PET. Lesion-to-contralateral SUVmax ratio was computed. Paired t-tests were conducted for statistical analyses.</div></div><div>¹¹C-MTO PET/CT scans performed in our hospital from June 2024 to January 2026 were reviewed. PET started 30 minutes post ¹¹C-MTO injection (0.2mCi/kg body weight). Only patients who were diagnosed with active adrenocortical lesions on FB-PET (10-min acquisition) and could hold single deep inspiration for CT (10 seconds) and then PET acquisition (15-30 seconds) were considered. Lesions were contoured (MIM 6.9.6) on BH-PET, BH-CT, FB-PET and FB-CT images reconstructed in 2mm thickness. For each lesion, Dice similarity coefficient (DSC) between PET and CT contours were determined for BH, and FB. Using lesion volume on BH-CT as the standard reference, absolute percentage difference was calculated for lesion volumes on BH-PET, and FB-PET. Lesion-to-contralateral SUVmax ratio was computed. Paired t-tests were conducted for statistical analyses.</div><div><h3>Results</h3><div>28 lesions in 26 patients (mean age 55.6, range 28-78; mean injected dose 15.0mCi, range 10.7-19.7) were analysed. Mean DSC on BH-PET/CT and FB-PET/CT were 0.63±0.14 and 0.47±0.18 respectively (<em>p</em><0.05). Mean lesion size was 19.9±12.6mm LD and 5.84±13.38ml on BH-CT. Mean absolute percentage difference of lesion volume for BH-PET and FB-PET were 18.9%±16.3 and 45.2%±41.5 respectively (<em>p</em><0","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102524"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149003923","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 Review of Local Practice: Functional Seated Scoliosis Imaging for Non-Ambulatory Paediatric Patients at Hong Kong Children's Hospital","authors":"Kwing Chung Lai, Tsan Pun Lam","doi":"10.1016/j.jmir.2026.102519","DOIUrl":"10.1016/j.jmir.2026.102519","url":null,"abstract":"<div><h3>Background</h3><div>At Hong Kong Children's Hospital (HKCH), specialized assessments are provided for paediatric patients with complex mobility impairments. These evaluations optimize seating, bracing, and posture in non-ambulatory children through multidisciplinary collaboration. The radiology department supports this process by acquiring a standardized series of seated and supine X-ray images to provide information for clinical decision-making.</div></div><div>At Hong Kong Children's Hospital (HKCH), specialized assessments are provided for paediatric patients with complex mobility impairments. These evaluations optimize seating, bracing, and posture in non-ambulatory children through multidisciplinary collaboration. The radiology department supports this process by acquiring a standardized series of seated and supine X-ray images to provide information for clinical decision-making.</div><div><h3>Patient type</h3><div>Patients referred for these assessments present with severe neuromuscular or neurological disorders. Analysis of more than 30 cases since 2024 shows that approximately 40% involve cerebral palsy variants, typically classified as GMFCS IV–V, indicating profound motor dependency. Around 25% have hereditary neuromuscular conditions (e.g., HMSN, FSHD, SMA type 2, MLD). Another 20–25% are affected by genetic syndromes (e.g., Schinzel-Giedion syndrome, KAT6A syndrome, Dravet syndrome, CDKL5 deficiency disorder, CPS I deficiency). These children are typically wheelchair-bound sitters requiring head and truncal support, with many using TLSO braces and customized wheelchairs. Seated imaging is essential to evaluate real-world posture, pelvic stability, brace efficacy, and scoliosis progression.</div></div><div>Patients referred for these assessments present with severe neuromuscular or neurological disorders. Analysis of more than 30 cases since 2024 shows that approximately 40% involve cerebral palsy variants, typically classified as GMFCS IV–V, indicating profound motor dependency. Around 25% have hereditary neuromuscular conditions (e.g., HMSN, FSHD, SMA type 2, MLD). Another 20–25% are affected by genetic syndromes (e.g., Schinzel-Giedion syndrome, KAT6A syndrome, Dravet syndrome, CDKL5 deficiency disorder, CPS I deficiency). These children are typically wheelchair-bound sitters requiring head and truncal support, with many using TLSO braces and customized wheelchairs. Seated imaging is essential to evaluate real-world posture, pelvic stability, brace efficacy, and scoliosis progression.</div><div><h3>Multidisciplinary collaboration</h3><div>Pre-assessment will be done by occupational therapists, physiotherapists, and prosthetic and orthotic specialists who perform evaluations and adjustments of the customised wheelchair. The patient will then be transferred to the radiology department for x-ray examinations. Orthopaedic specialists will lead follow-up consultation to review findings and finalize treatment plans.</div></div><d","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102519"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149003926","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":"Mechanisms and preventive strategies for radiation-induced endothelial dysfunction: A systematic review","authors":"Shing Yau Tam, Ka Hei Man, Yuanxi Li","doi":"10.1016/j.jmir.2026.102507","DOIUrl":"10.1016/j.jmir.2026.102507","url":null,"abstract":"<div><h3>Background</h3><div>20 million new cancer cases were diagnosed worldwide annually and over 50% require radiotherapy. Despite technological advancements, radiation-induced endothelial damage remains a leading cause of morbidity and mortality among cancer survivors. The underlying radiobiological mechanisms are complex and not fully elucidated. This systematic review aims to synthesizes recent evidence on the mechanisms and preventive strategies for radiation-induced endothelial dysfunction and identify future research directions.</div></div><div>20 million new cancer cases were diagnosed worldwide annually and over 50% require radiotherapy. Despite technological advancements, radiation-induced endothelial damage remains a leading cause of morbidity and mortality among cancer survivors. The underlying radiobiological mechanisms are complex and not fully elucidated. This systematic review aims to synthesizes recent evidence on the mechanisms and preventive strategies for radiation-induced endothelial dysfunction and identify future research directions.</div><div><h3>Methods</h3><div>A systematic review was conducted in late 2025 in accordance with the PRISMA guideline across 3 databases (PubMed, Scopus and Embase). Records from the past 5 years were searched using the keywords (radiotherapy [Title/Abstract]) AND (endothelial [Title/Abstract]) AND (damage [Title/Abstract] OR dysfunction [Title/Abstract]). Studies were screened independently, and the methodological quality of included articles was assessed using the MINORS approach.</div></div><div>A systematic review was conducted in late 2025 in accordance with the PRISMA guideline across 3 databases (PubMed, Scopus and Embase). Records from the past 5 years were searched using the keywords (radiotherapy [Title/Abstract]) AND (endothelial [Title/Abstract]) AND (damage [Title/Abstract] OR dysfunction [Title/Abstract]). Studies were screened independently, and the methodological quality of included articles was assessed using the MINORS approach.</div><div><h3>Results</h3><div>Of 145 identified records, 19 studies were included (3 clinical studies with sample size of 15-43 patients, and 16 non-clinical studies). All articles were of “good” quality (MINORS score ≥10). Radiation doses exceeding 2 Gy induce endothelial damage, with more severe dysfunction observed above 10 Gy, typically manifesting 1-8 weeks post-irradiation. The predominant mechanism was increased oxidative stress and reactive oxygen species (n=10, 53%). Other significant causative factors included the release of pro-inflammatory mediators (n=4, 21%) and the activation of endothelium-to-mesenchymal transition (n=3, 16%). Preventive strategies identified were normal tissue dose reduction and the application of statins. Notably, herbal extractions including puerarin and chlorogenic acid demonstrated protective potential. Common study limitations were small sample sizes and the use of non-human models (n=6, 30%). Moreover, there w","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102507"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149003933","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":"Performance comparison of two Asia-developed AI auto-contouring systems in prostate radiotherapy: Validation in complex clinical scenarios","authors":"Keisuke Yasui, Yuya Nagake, Ayami Sugiyama, Yasunori Saito, Hidetoshi Shimizu, Haruka Uezono, Naoki Hayashi","doi":"10.1016/j.jmir.2026.102508","DOIUrl":"10.1016/j.jmir.2026.102508","url":null,"abstract":"<div><h3>Background</h3><div>Accurate organ delineation in radiotherapy treatment planning directly influences dose distribution and treatment-related toxicity, yet inter-observer variability remains a persistent challenge. AI-based auto-segmentation systems have been introduced to reduce contouring workload and improve consistency; however, most validation studies focus on standard anatomical cases, leaving performance in clinically challenging edge cases poorly characterized. This study compares two Asia-developed AI auto-contouring systems—RatoGuide (Japan) and OncoStudio (Korea)—in prostate radiotherapy planning CT, with systematic evaluation under edge-case conditions.</div></div><div>Accurate organ delineation in radiotherapy treatment planning directly influences dose distribution and treatment-related toxicity, yet inter-observer variability remains a persistent challenge. AI-based auto-segmentation systems have been introduced to reduce contouring workload and improve consistency; however, most validation studies focus on standard anatomical cases, leaving performance in clinically challenging edge cases poorly characterized. This study compares two Asia-developed AI auto-contouring systems—RatoGuide (Japan) and OncoStudio (Korea)—in prostate radiotherapy planning CT, with systematic evaluation under edge-case conditions.</div><div><h3>Methods</h3><div>A retrospective validation study analyzed 45 prostate cancer patients stratified into four groups: Normal (n=8, standard cases without implants), Spacer (n=5, rectal spacers only), Seed/Spacer (n=16, combined brachytherapy seeds and spacers), and GM/Spacer (n=16, gold fiducial markers and spacers). Expert radiation oncologists and medical physicists manually contoured five anatomical structures (prostate, rectum, bladder, seminal vesicles, femoral heads) according to JCOG2011 protocol, serving as reference contours. AI-generated auto-contours were compared using Sørensen–Dice similarity coefficient (DSC) and 95% Hausdorff distance (HD). Statistical analyses employed Wilcoxon signed-rank tests for inter-software comparisons and Mann–Whitney U tests for edge-case analysis.</div></div><div>A retrospective validation study analyzed 45 prostate cancer patients stratified into four groups: Normal (n=8, standard cases without implants), Spacer (n=5, rectal spacers only), Seed/Spacer (n=16, combined brachytherapy seeds and spacers), and GM/Spacer (n=16, gold fiducial markers and spacers). Expert radiation oncologists and medical physicists manually contoured five anatomical structures (prostate, rectum, bladder, seminal vesicles, femoral heads) according to JCOG2011 protocol, serving as reference contours. AI-generated auto-contours were compared using Sørensen–Dice similarity coefficient (DSC) and 95% Hausdorff distance (HD). Statistical analyses employed Wilcoxon signed-rank tests for inter-software comparisons and Mann–Whitney U tests for edge-case analysis.</div><div><h3>Results</h3><div>RatoG","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102508"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149003999","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":"Optimizing contrast injection dose for Cardiac Computed Tomography imaging for paediatrics","authors":"Wing Tak Lai","doi":"10.1016/j.jmir.2026.102512","DOIUrl":"10.1016/j.jmir.2026.102512","url":null,"abstract":"<div><h3>Introduction</h3><div>Cardiac Computed Tomography (CT) imaging is often regarded as a complex imaging technique in diagnostic imaging. Pediatric cardiac CT is even more challenging, as most related diseases are congenital, vary in nature, and commonly affect multiple structures of the heart. Furthermore, the techniques used for pediatric cardiac CT vary significantly with patient age and pathology. It is also difficult or impossible to control the breathing or heart rate of an infant. Finally, patient body weight and the size of the intravenous access often limit the amount of contrast that can be delivered to enhance the body structures. Since 2020, the Department of Radiology at the Hong Kong Children’s Hospital has conducted research to determine the optimal timing and contrast delivery regime for patients ranging from 0 to 18 years of age. The objective of this research is to minimize the contrast dose and related kVp used for Pediatric cardiac CT imaging<sup>3</sup>.</div></div><div>Cardiac Computed Tomography (CT) imaging is often regarded as a complex imaging technique in diagnostic imaging. Pediatric cardiac CT is even more challenging, as most related diseases are congenital, vary in nature, and commonly affect multiple structures of the heart. Furthermore, the techniques used for pediatric cardiac CT vary significantly with patient age and pathology. It is also difficult or impossible to control the breathing or heart rate of an infant. Finally, patient body weight and the size of the intravenous access often limit the amount of contrast that can be delivered to enhance the body structures. Since 2020, the Department of Radiology at the Hong Kong Children’s Hospital has conducted research to determine the optimal timing and contrast delivery regime for patients ranging from 0 to 18 years of age. The objective of this research is to minimize the contrast dose and related kVp used for Pediatric cardiac CT imaging<sup>3</sup>.</div><div><h3>Materials and Methods</h3><div>Approximately 200 patients were included in the study. The general principle was to estimate the required injection duration and match it with the maximum permissible injection rate, ensuring the total contrast volume remained below the maximum dose limit calculated by body weight. A multi-disciplinary team including medical physicists, nurses, radiographers, and radiologists was involved in the research program. Phantom studies were used to define the optimal contrast injection duration for achieving approximately 400 Hounsfield Units (HU) during imaging. Once the injection duration was confirmed, data on different body weights, injection rates, kVp settings, and the resulting HU values were recorded for analysis.</div></div><div>Approximately 200 patients were included in the study. The general principle was to estimate the required injection duration and match it with the maximum permissible injection rate, ensuring the total contrast volume remained below the ","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102512"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149004002","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":"Engineering a Safety-First Agentic Retrieval-Augmented Generation Architecture for Pediatric Interventional Radiology","authors":"Chun Wai Will Lee, Pak Lun Lam, Elaine Yee Ling Kan, Kevin Kin Fen Fung","doi":"10.1016/j.jmir.2026.102527","DOIUrl":"10.1016/j.jmir.2026.102527","url":null,"abstract":"<div><h3>Background / Purpose</h3><div>As interventional radiology transitions toward a direct clinical care model, the demand for reliable, localized patient education tools has surged. However, generic large language models pose unacceptable hallucination risks in high-stakes pediatric settings. The purpose of this study is to detail the development of a bilingual artificial intelligence assistant, engineered with an agentic retrieval-augmented generation architecture, to safely deliver institution-specific pediatric guidance.</div></div><div>As interventional radiology transitions toward a direct clinical care model, the demand for reliable, localized patient education tools has surged. However, generic large language models pose unacceptable hallucination risks in high-stakes pediatric settings. The purpose of this study is to detail the development of a bilingual artificial intelligence assistant, engineered with an agentic retrieval-augmented generation architecture, to safely deliver institution-specific pediatric guidance.</div><div><h3>Methods</h3><div>The system was built using a graph-based multi-agent framework to move beyond standard black-box retrieval, prioritizing a five-phase safety-first pipeline. First, an emergency detection layer uses algorithmic screening to redirect urgent medical queries to emergency care. Second, an orchestration agent translates layman queries into clinical, conceptual, and layman search variants to maximize vector retrieval recall. Third, a retrieval and self-correction loop evaluates retrieved institutional documents, automatically rewriting queries if initial relevance is low. Fourth, clinical synthesis generates structured outputs capturing confidence scores, reasoning, and cited sources. Finally, a simulated senior review sub-agent systematically audits drafted answers for factual correctness and clinical safety before delivery.</div></div><div>The system was built using a graph-based multi-agent framework to move beyond standard black-box retrieval, prioritizing a five-phase safety-first pipeline. First, an emergency detection layer uses algorithmic screening to redirect urgent medical queries to emergency care. Second, an orchestration agent translates layman queries into clinical, conceptual, and layman search variants to maximize vector retrieval recall. Third, a retrieval and self-correction loop evaluates retrieved institutional documents, automatically rewriting queries if initial relevance is low. Fourth, clinical synthesis generates structured outputs capturing confidence scores, reasoning, and cited sources. Finally, a simulated senior review sub-agent systematically audits drafted answers for factual correctness and clinical safety before delivery.</div><div><h3>Results</h3><div>An automated expert-agent evaluation framework successfully assessed the pipeline across edge-case clinical queries. Furthermore, the system demonstrated hospital deployment feasibility by comparing a secure, loca","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102527"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149004455","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":"Beyond the 5-Year Milestone: Evaluating the surveillance mammography program for late breast cancer recurrence in Prince of Wales Hospital","authors":"Miu Yung Sze, Lok Ching Chan, Nga Yee Choi","doi":"10.1016/j.jmir.2026.102528","DOIUrl":"10.1016/j.jmir.2026.102528","url":null,"abstract":"<div><h3>Background</h3><div>The optimal imaging strategy for post-mastectomy patients beyond five years remains variable. Prince of Wales Hospital provides a surveillance program with mammography every 18 months to patients with mastectomy performed more than 5 years till 10 year, then clinical surveillance till 15 year. This study evaluates the specific yield of our mammographic surveillance program in detecting late recurrences for women with mastectomy done.</div></div><div>The optimal imaging strategy for post-mastectomy patients beyond five years remains variable. Prince of Wales Hospital provides a surveillance program with mammography every 18 months to patients with mastectomy performed more than 5 years till 10 year, then clinical surveillance till 15 year. This study evaluates the specific yield of our mammographic surveillance program in detecting late recurrences for women with mastectomy done.</div><div><h3>Methods</h3><div>A retrospective cohort of 233 post-mastectomy patients was monitored over 15 years. Program effectiveness was measured by the Cancer Detection Rate (CDR), recall rate and sensitivity of institutional imaging versus external or symptomatic detection.</div></div><div>A retrospective cohort of 233 post-mastectomy patients was monitored over 15 years. Program effectiveness was measured by the Cancer Detection Rate (CDR), recall rate and sensitivity of institutional imaging versus external or symptomatic detection.</div><div><h3>Results</h3><div>There were 8 cases of recurrence (3.4%) identified. All the recurrences were found after 10 years post-surgery. One contralateral breast cancer recurrence (12.5%; sensitivity) was detected by our surveillance mammography program confirmed by ultrasound-guided biopsy after suspicious recall. The CDR was 4.29 per 1,000, which aligns with the America College of Radiology benchmark for subsequent screening (2.5 – 5.1 per 1,000). The program maintained a low recall rate of 1.28% and a PPV of 33.3%.</div><div>The remaining 7 cases (87.5%; sensitivity) were identified externally or through other modalities.</div></div><div>There were 8 cases of recurrence (3.4%) identified. All the recurrences were found after 10 years post-surgery. One contralateral breast cancer recurrence (12.5%; sensitivity) was detected by our surveillance mammography program confirmed by ultrasound-guided biopsy after suspicious recall. The CDR was 4.29 per 1,000, which aligns with the America College of Radiology benchmark for subsequent screening (2.5 – 5.1 per 1,000). The program maintained a low recall rate of 1.28% and a PPV of 33.3%.</div><div>The remaining 7 cases (87.5%; sensitivity) were identified externally or through other modalities.</div><div><h3>Conclusion</h3><div>The surveillance program of Prince of Wales Hospital demonstrated effectiveness in detecting contralateral breast cancer recurrence with a cancer detection rate of 4.29 per 1000 and a low recall rate of 1.28%. However most late recurre","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102528"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149004456","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":"Performance trends and analysis of a cardiac-dedicated SPECT from a daily quality assurance program: experience sharing to optimize machine performance and to explore artificial application to the results","authors":"Lik Wai Kwok, KH SIN, Monica Chu, BF Yeung, Kevin So, KS Ng, Cathy Cheung, KK Kwan, Martin Law, TK Chow, WH Ma","doi":"10.1016/j.jmir.2026.102525","DOIUrl":"10.1016/j.jmir.2026.102525","url":null,"abstract":"<div><h3>Background</h3><div>To describe the daily quality assurance (DQA) program for a cardiac-dedicated system, to interpret the DQA images before patient service, to use time series of measurable parameters to predict possible machine performance and to explore such time trend for possible artificial intelligence (AI) application to make the DQA program more fruitful.</div></div><div>To describe the daily quality assurance (DQA) program for a cardiac-dedicated system, to interpret the DQA images before patient service, to use time series of measurable parameters to predict possible machine performance and to explore such time trend for possible artificial intelligence (AI) application to make the DQA program more fruitful.</div><div><h3>Methods</h3><div>The cardiac-dedicated SPECT system, equipped with cadmium-zinc-telluride (CZT)-based detectors, has demonstrated myocardial perfusion imaging with image quality superior to that of conventional SPECT or SPECT/CT while reducing patient and personnel radiation exposure and decreasing imaging time. DQA is important to maintain the optimized performance of the system, in which a 57Co rod source of nominal radioactivity 10 mCi placed vertically at a pre-defined position in front of the system. The DQA program tabulates measurable parameters of energy resolution, uniformity and sensitivity. The reconstructed image of the 57Co rod source can be visualized as a straight image.</div></div><div>The cardiac-dedicated SPECT system, equipped with cadmium-zinc-telluride (CZT)-based detectors, has demonstrated myocardial perfusion imaging with image quality superior to that of conventional SPECT or SPECT/CT while reducing patient and personnel radiation exposure and decreasing imaging time. DQA is important to maintain the optimized performance of the system, in which a 57Co rod source of nominal radioactivity 10 mCi placed vertically at a pre-defined position in front of the system. The DQA program tabulates measurable parameters of energy resolution, uniformity and sensitivity. The reconstructed image of the 57Co rod source can be visualized as a straight image.</div><div><h3>Results</h3><div>Daily results of measurable parameters have been collected for the year 2025, within which the average energy resolution is 5.8±0.1% (specification:<6%), tomographic sensitivity is 2346±9 cpm/μCi(specification:>1900 cpm/μCi), image uniformity is 96±1% (specification:>85%). Time trends for the measurable parameters show that consistent system performance can be maintained using a simple yet clinically applicable DQA program. The reconstructed 57Co rod source should be free from image artifact before daily patient service, all of which will be presented and discussed for a better and more efficient DQA program for the future.</div></div><div>Daily results of measurable parameters have been collected for the year 2025, within which the average energy resolution is 5.8±0.1% (specification:<6%), tomographic sen","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102525"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149003924","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":"Artificial Intelligence in Healthcare: Transforming Medical Practice and Patient Outcomes","authors":"Yuen Nee Yvonne Loh","doi":"10.1016/j.jmir.2026.102531","DOIUrl":"10.1016/j.jmir.2026.102531","url":null,"abstract":"<div><div>Artificial Intelligence (AI) is revolutionizing healthcare delivery by improving diagnostic precision, refining treatment protocols, and enhancing operational efficiency across medical specialties. AI technologies address critical contemporary healthcare challenges, including lengthy procedures, inadequate treatment adjustments, and resource allocation limitations. AI-enhanced adaptive planning systems in radiation therapy exemplify this transformation. Traditional radiation therapy planning is labor-intensive and frequently fails to optimally adjust to anatomical changes during treatment. AI-driven systems streamline Adaptive Radiation Therapy (ART) planning, enabling real-time modifications based on tumor size variations and patient anatomy changes. These systems employ sophisticated imaging acquisition, AI-optimized adaptive planning algorithms, and automated treatment delivery protocols to enhance therapeutic effectiveness while minimizing healthy tissue exposure. Beyond oncology, AI applications span diverse healthcare sectors, including medical imaging analysis, predictive analytics for patient risk assessment, drug discovery acceleration, and clinical decision support systems. Machine learning algorithms analyze extensive datasets to uncover patterns imperceptible to human observers, facilitating earlier disease detection and personalized treatment strategies. Natural language processing improves electronic health record management, while robotic surgery systems enhance procedural accuracy. AI implementation in healthcare delivers substantial benefits: reduced staff workload, shortened patient waiting times, increased treatment precision, and optimized resource utilization. However, challenges persist, including data privacy concerns, algorithmic bias mitigation, regulatory compliance, and requirements for thorough clinical validation. As healthcare systems globally confront rising patient volumes and increasing complexity, AI technologies offer scalable solutions that augment rather than replace human expertise. The future of healthcare lies in collaborative integration of AI capabilities with clinical knowledge, promising more efficient, precise, and accessible medical care for diverse patient populations while preserving essential human elements of compassionate healthcare delivery.</div></div><div>Artificial Intelligence (AI) is revolutionizing healthcare delivery by improving diagnostic precision, refining treatment protocols, and enhancing operational efficiency across medical specialties. AI technologies address critical contemporary healthcare challenges, including lengthy procedures, inadequate treatment adjustments, and resource allocation limitations. AI-enhanced adaptive planning systems in radiation therapy exemplify this transformation. Traditional radiation therapy planning is labor-intensive and frequently fails to optimally adjust to anatomical changes during treatment. AI-driven systems streamline Adaptive Radia","PeriodicalId":46420,"journal":{"name":"Journal of Medical Imaging and Radiation Sciences","volume":"57 5","pages":"Article 102531"},"PeriodicalIF":2.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149003688","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}