Structure–Function Relationships in Geographic Atrophy Based on Mesopic Microperimetry, Fundus Autofluorescence, and Optical Coherence Tomography

Translational Vision Science & Technology, Год журнала: 2025, Номер 14(2), С. 7 - 7

Опубликована: Фев. 5, 2025

Purpose: To examine relationships between retinal structure and visual function in geographic atrophy (GA) by analyzing spatial agreement absolute scotomas macular structure, focusing on (1) choroidal hypertransmission, a key feature of complete pigment epithelium outer (cRORA), (2) fundus autofluorescence (FAF)–defined GA. Methods: Mesopic microperimetry (using novel T-shaped pattern) multimodal imaging were recorded longitudinally phase II GA trial. Horizontal vertical optical coherence tomography (OCT) line scans (corresponding to the T axes) graded for hypertransmission; FAF images Spatial concordance zones scotoma was quantified Dice similarity coefficient (DSC). Results: The analysis population comprised 24 participants (mean follow-up 26.8 months). For estimated mean DSC 0.70 (95% confidence interval [CI], 0.64–0.77). This significantly higher than FAF-defined (0.67; 95% CI, 0.61–0.74; difference = 0.03, 0.02–0.05, P < 0.001). Mean OCT reflectivity strongly associated with likelihood severity scotoma. Conclusions: structural features is moderately high slightly hypertransmission supports cRORA feature, as an outcome measure interventional trials. provides more information explain alone. However, given some discordance both features, performing alongside remains important. Translational Relevance: These findings provide insights into complex relationship contribute nuanced understanding measures.

Язык: Английский

---

Real-world validation of Artificial Intelligence-based Computed Tomography auto-contouring for prostate cancer radiotherapy planning

Physics and Imaging in Radiation Oncology, Год журнала: 2023, Номер 28, С. 100501 - 100501

Опубликована: Окт. 1, 2023

Background and purposeArtificial Intelligence (AI)-based auto-contouring for treatment planning in radiotherapy needs extensive clinical validation, including the impact of editing after automatic segmentation. The aims this study were to assess performance a commercial system Clinical Target Volumes (CTVs) (prostate/seminal vesicles) selected Organs at Risk (OARs) (rectum/bladder/femoral heads+femurs), evaluating also inter-observer variability (manual vs automatic+editing) reduction contouring time.Materials methodsTwo expert observers contoured CTVs/OARs 20 patients our Treatment Planning System (TPS). Computed Tomography (CT) images sent workstation: contours generated back TPS, where could edit them if necessary. Inter- intra-observer consistency was estimated using Dice Similarity Coefficients (DSC). Radiation oncologists asked score quality contours, ranging from 1 (complete re-contouring) 5 (no editing). Contouring times automatic+edit) compared.ResultsDSCs only) consistent with (between 0.65 seminal vesicles 0.94 bladder); further improved performances (range: 0.76-0.94). median 4 (little editing) it <4 3/2 two respectively. Inter-observer automatic+editing significantly, being lower than manual (e.g.: vesicles: 0.83vs0.73; prostate: 0.86vs0.83; rectum: 0.96vs0.81). Oncologist time reduced 17-24 minutes 3-7 (p<0.01).ConclusionAutomatic AI-based followed by can replace contouring, resulting significantly segmentation better between operators.

Язык: Английский

---

Evaluating ChatGPT’s competency in radiation oncology: A comprehensive assessment across clinical scenarios

Radiotherapy and Oncology, Год журнала: 2024, Номер 202, С. 110645 - 110645

Опубликована: Ноя. 19, 2024

Язык: Английский

---

Guidance on selecting and evaluating AI auto-segmentation systems in clinical radiotherapy: insights from a six-vendor analysis

Physical and Engineering Sciences in Medicine, Год журнала: 2025, Номер unknown

Опубликована: Янв. 13, 2025

Abstract Artificial Intelligence (AI) based auto-segmentation has demonstrated numerous benefits to clinical radiotherapy workflows. However, the rapidly changing regulatory, research, and market environment presents challenges around selecting evaluating most suitable solution. To support adoption of AI systems, Selection Criteria recommendations were developed enable a holistic evaluation vendors, considering not only raw performance but associated risks uniquely related deployment AI. In-house experience key bodies work on ethics, standards, best practices for in Radiation Oncology reviewed inform selection criteria strategies. A retrospective analysis using was performed across six including quantitative assessment five metrics (Dice, Hausdorff Distance, Average Surface Dice, Added Path Length) 20 head neck, thoracic, 19 male pelvis patients models as March 2023. total 47 identified seven categories. showed that overall no vendor exceedingly well, with systematically poor Data Security & Responsibility, Vendor Support Tools, Transparency Ethics. In terms performance, vendors varied widely from excellent poor. As new regulations come into force scope systems adapt needs, continued interest ensuring safe, fair, transparent will persist. The framework provided herein aims promote user confidence by exploring breadth clinically relevant factors informed decision-making.

Язык: Английский

---

Head and Neck Gross Tumor Volume Automatic Segmentation Using PocketNet

Lecture notes in computer science, Год журнала: 2025, Номер unknown, С. 241 - 249

Опубликована: Янв. 1, 2025

Язык: Английский

---

Geometric and dosimetric evaluation of a commercial AI auto‐contouring tool on multiple anatomical sites in CT scans

Journal of Applied Clinical Medical Physics, Год журнала: 2025, Номер unknown

Опубликована: Март 17, 2025

Abstract Current radiotherapy practices rely on manual contouring of CT scans, which is time‐consuming, prone to variability, and requires highly trained experts. There a need for more efficient consistent methods. This study evaluated the performance Varian Ethos AI auto‐contouring tool assess its potential integration into clinical workflows. retrospective included 223 patients with treatment sites in pelvis, abdomen, thorax, head neck regions. The generated auto‐contours each patients’ pre‐treatment planning CT, 45 unique structures were across cohort. Multiple measures geometric similarity computed, including surface Dice Similarity Coefficient (sDSC) mean distance agreement (MDA). Dosimetric concordance was by comparing dose maximum 2 cm 3 (D cc ) between contours. demonstrated high accuracy well‐defined like bladder, lungs, femoral heads. Smaller those less defined boundaries, such as optic nerves duodenum, showed lower agreement. Over 70% sDSC > 0.8, 74% had MDA < 2.5 mm. Geometric generally correlated dosimetric concordance, however differences contour definitions did result some exhibiting deviations. offers promising reliability many anatomical structures, supporting use Auto‐contouring errors, although rare, highlight importance ongoing QA expert oversight.

Язык: Английский

---

Proposal for a Method for Assessing the Quality of an Updated Deep Learning-Based Automatic Segmentation Program

Cureus, Год журнала: 2025, Номер unknown

Опубликована: Март 27, 2025

This study aimed to verify whether a commercial deep learning-based automatic segmentation (DLS) method can maintain contour geometric accuracy post-update and propose streamlined validation that minimizes the burden on clinical workflows. included 109 participants. Radiation oncologists used computed tomography (CT) imaging identify 28 organs located in head neck, chest, abdomen, pelvic regions. Contours were delineated CT images using AI-Rad Companion Organs RT (AIRC; Siemens Healthineers, Erlangen, Germany) versions VA30, VA50, VA50. The Dice similarity coefficient, maximum Hausdorff distance, mean distance agreement calculated contours with significant differences among versions. To evaluate identified contours, ground truth was defined as by radiation oncologists, indices for VA60 recalculated. Statistical analysis performed between each version. Among evaluated, nine did not satisfy established criteria. revealed brain, rectum, bladder differed substantially across AIRC In particular, pre-update rectum had (range) of 0.76 (0.40-1.16), whereas exhibited lower quality, 1.13 (0.24-5.68). Therefore, DLS methods undergo regular updates must be reassessed quality region interest. proposed help reduce workflows while appropriately evaluating performance.

Язык: Английский

---

The Worrisome Impact of an Inter-rater Bias on Neural Network Training

Lecture notes in electrical engineering, Год журнала: 2025, Номер unknown, С. 463 - 473

Опубликована: Янв. 1, 2025

Язык: Английский

---

Contouring in transition: perceptions of AI-based autocontouring by radiation oncologists and medical physicists in German-speaking countries

Strahlentherapie und Onkologie, Год журнала: 2025, Номер unknown

Опубликована: Апрель 28, 2025

Язык: Английский

---

Modeling inter‐reader variability in clinical target volume delineation for soft tissue sarcomas using diffusion model

Medical Physics, Год журнала: 2025, Номер unknown

Опубликована: Май 3, 2025

Abstract Background Accurate delineation of the clinical target volume (CTV) is essential in radiotherapy treatment soft tissue sarcomas. However, this process subject to inter‐reader variability due need for assessment risk and extent potential microscopic spread. This can lead inconsistencies planning, potentially impacting outcomes. Most existing automatic CTV methods do not account only generate a single each case. Purpose study aims develop deep learning‐based technique multiple contours case, simulating practice. Methods We employed publicly available dataset consisting fluorodeoxyglucose positron emission tomography (FDG‐PET), x‐ray computed (CT), pre‐contrast T1‐weighted magnetic resonance imaging (MRI) scans from 51 patients with sarcoma, along an independent validation set containing five additional patients. An experienced reader drew contour gross tumor (GTV) patient based on multi‐modality images. Subsequently, two readers, together first one, were responsible contouring three CTVs total GTV. developed diffusion model‐based learning method that capable generating arbitrary number different plausible mimic delineation. The proposed model incorporates separate encoder extract features GTV masks, leveraging critical role information accurate Results demonstrated superior performance highest Dice Index (0.902 compared values below 0.881 state‐of‐the‐art models) best generalized energy distance (GED) (0.209 exceeding 0.221 models). It also achieved second‐highest recall precision metrics among ambiguous image segmentation models. both datasets exhibited consistent trends, reinforcing reliability our findings. Additionally, ablation studies exploring structures input configurations highlighted significance incorporating prior Conclusions successfully generates sarcomas, effectively capturing

Язык: Английский

---