Development and external multicentric validation of a deep learning-based clinical target volume segmentation model for whole-breast radiotherapy

Physics and Imaging in Radiation Oncology, Journal Year: 2025, Volume and Issue: unknown, P. 100749 - 100749

Published: March 1, 2025

Language: Английский

---

Open‐source deep‐learning models for segmentation of normal structures for prostatic and gynecological high‐dose‐rate brachytherapy: Comparison of architectures

Journal of Applied Clinical Medical Physics, Journal Year: 2025, Volume and Issue: unknown

Published: April 5, 2025

Abstract Background The use of deep learning‐based auto‐contouring algorithms in various treatment planning services is increasingly common. There a notable deficit commercially or publicly available models trained on large diverse datasets containing high‐dose‐rate (HDR) brachytherapy scans, leading to poor performance images that include HDR implants. Purpose To implement and evaluate automatic organs‐at‐risk (OARs) segmentation for prostatic‐and‐gynecological computed tomography (CT)‐guided planning. Methods materials 1316 (CT) scans corresponding files from 1105 prostatic‐or‐gynecological patients treated at our institution 2017 2024 were used model training. Data sources comprised six CT scanners including mobile unit with previously reported susceptibility image streaking artifacts. Two UNet‐derived architectures, UNet++ nnU‐Net, investigated bladder rectum tested 100 clinically‐used 62 patients, disjoint the training set, collected 2024. Performance was evaluated using Dice‐Similarity‐Coefficient (DSC) between predicted contours slices common Clinical‐Target‐Volume (CTV). Additionally, blinded evaluation ten random test cases conducted by three experienced planners. Results Median (interquartile range) 3D DSC CTV‐containing 0.95 (0.04) 0.87 (0.09) models, respectively, 0.96 (0.03) 0.88 (0.10) nnU‐Net. rank‐sum did not reveal statistically significant differences these ( p = 0.15 0.27, respectively). scored higher than contours. Conclusion Both architectures perform similarly are adequately accurate reduce contouring time review‐and‐edit context during chosen implementation due lower computing hardware requirements routine clinical use.

Language: Английский

---

Clinical adoption of deep learning target auto-segmentation for radiation therapy: challenges, clinical risks, and mitigation strategies

Deleted Journal, Journal Year: 2024, Volume and Issue: 1(1)

Published: Jan. 1, 2024

Abstract Radiation therapy is a localized cancer treatment that relies on precise delineation of the target to be treated and healthy tissues guarantee optimal effect. This step, known as contouring or segmentation, involves identifying both volumes organs at risk imaging modalities like CT, PET, MRI guide radiation delivery. Manual however, time-consuming highly subjective, despite presence guidelines. In recent years, automated segmentation methods, particularly deep learning models, have shown promise in addressing this task. However, challenges persist their clinical use, including need for robust quality assurance (QA) processes risks associated with use models. review examines considerations adoption auto-segmentation radiotherapy, focused volume. We discuss potential (eg, over- under-segmentation, automation bias, appropriate trust), mitigation strategies human oversight, uncertainty quantification, education professionals), we highlight importance expanding QA include geometric, dose-volume, outcome-based performance monitoring. While offers significant benefits, careful attention rigorous measures are essential its successful integration practice.

Language: Английский

---

Automated segmentation in planning-CT for breast cancer radiotherapy: A review of recent advances

Radiotherapy and Oncology, Journal Year: 2024, Volume and Issue: 202, P. 110615 - 110615

Published: Nov. 1, 2024

Postoperative radiotherapy (RT) has been shown to effectively reduce disease recurrence and mortality in breast cancer (BC) treatment. A critical step the planning workflow is accurate delineation of clinical target volumes (CTV) organs-at-risk (OAR). This literature review evaluates recent advancements deep-learning (DL) atlas-based auto-contouring techniques for CTVs OARs BC planning-CT images RT. It examines their performance regarding geometrical dosimetric accuracy, inter-observer variability, time efficiency. Our findings indicate that both DL- methods generally show comparable across CTVs, with DL slightly outperforming consistency accuracy. Auto-segmentation most achieved robust results segmentation quality planning. However, lymph node levels (LNLs) presented greatest challenge auto-segmentation significant impact on The translation these into practice limited by geometric metrics lack dose evaluation studies. Additionally, algorithms showed diverse structure sets, while training datasets varied size, origin, patient positioning imaging protocols, affecting model sensitivity. Guideline inconsistencies varying definitions ground truth led substantial suggesting a need reliable consensus dataset. Finally, our highlights popularity U-Net architecture. In conclusion, automated contouring proven efficient many breast-CTV, further improvements are necessary LNL delineation, analysis, building.

Language: Английский

---

Automated Organ Segmentation for Radiation Therapy: A Comparative Analysis of AI-Based Tools Versus Manual Contouring in Korean Cancer Patients

Cancers, Journal Year: 2024, Volume and Issue: 16(21), P. 3670 - 3670

Published: Oct. 30, 2024

Accurate delineation of tumors and organs at risk (OARs) is crucial for intensity-modulated radiation therapy. This study aimed to evaluate the performance OncoStudio, an AI-based auto-segmentation tool developed Korean patients, compared with Protégé AI, a globally that uses data from cancer patients.

Language: Английский

---