Evaluation of deep learning-based autosegmentation in breast cancer radiotherapy

Radiation Oncology, Journal Year: 2021, Volume and Issue: 16(1)

Published: Oct. 14, 2021

Abstract Purpose To study the performance of a proposed deep learning-based autocontouring system in delineating organs at risk (OARs) breast radiotherapy with group experts. Methods Eleven experts from two institutions delineated nine OARs 10 cases adjuvant after breast-conserving surgery. Autocontours were then provided to for correction. Overall, 110 manual contours, corrected autocontours, and autocontours each type OAR analyzed. The Dice similarity coefficient (DSC) Hausdorff distance (HD) used compare degree agreement between best contour (chosen by an independent expert committee) autocontour, contour. Higher DSCs lower HDs indicated better geometric overlap. amount time reduction using was examined. User satisfaction evaluated survey. Results Manual had similar accuracy average DSC value (0.88 vs. 0.90 0.90). ranked second place, based on DSCs, first among contours. Interphysician variations reduced compared contours (DSC: 0.89–0.90 0.87–0.90; HD: 4.3–5.8 mm 5.3–7.6 mm). Among delineations, largest variations, which improved most significantly system. total mean times 37 min 6 autocontours. results survey revealed good user satisfaction. Conclusions as that experts’ contouring. This can be valuable improving quality reducing interphysician variability clinical practice.

Language: Английский

---

Assessing Interobserver Variability in the Delineation of Structures in Radiation Oncology: A Systematic Review

International Journal of Radiation Oncology*Biology*Physics, Journal Year: 2022, Volume and Issue: 115(5), P. 1047 - 1060

Published: Nov. 22, 2022

Language: Английский

---

Artificial intelligence in breast cancer imaging: risk stratification, lesion detection and classification, treatment planning and prognosis—a narrative review

Exploration of Targeted Anti-tumor Therapy, Journal Year: 2022, Volume and Issue: unknown, P. 795 - 816

Published: Dec. 27, 2022

The advent of artificial intelligence (AI) represents a real game changer in today's landscape breast cancer imaging. Several innovative AI-based tools have been developed and validated recent years that promise to accelerate the goal patient-tailored management. Numerous studies confirm proper integration AI into existing clinical workflows could bring significant benefits women, radiologists, healthcare systems. approach has proved particularly useful for developing new risk prediction models integrate multi-data streams planning individualized screening protocols. Furthermore, help radiologists pre-screening lesion detection phase, increasing diagnostic accuracy, while reducing workload complications related overdiagnosis. Radiomics radiogenomics approaches extrapolate so-called imaging signature tumor plan targeted treatment. main challenges development are huge amounts high-quality data required train validate these need multidisciplinary team with solid machine-learning skills. purpose this article is present summary most important applications imaging, analyzing possible perspectives widespread adoption tools.

Language: Английский

---

Experience of Implementing Deep Learning-Based Automatic Contouring in Breast Radiation Therapy Planning: Insights From Over 2000 Cases

International Journal of Radiation Oncology*Biology*Physics, Journal Year: 2024, Volume and Issue: 119(5), P. 1579 - 1589

Published: Feb. 29, 2024

Language: Английский

---

Clinical implementation of deep-learning based auto-contouring tools–Experience of three French radiotherapy centers

Cancer/Radiothérapie, Journal Year: 2021, Volume and Issue: 25(6-7), P. 607 - 616

Published: Aug. 11, 2021

Language: Английский

---

Comprehensive clinical evaluation of deep learning-based auto-segmentation for radiotherapy in patients with cervical cancer

Frontiers in Oncology, Journal Year: 2023, Volume and Issue: 13

Published: April 28, 2023

Deep learning-based models have been actively investigated for various aspects of radiotherapy. However, cervical cancer, only a few studies dealing with the auto-segmentation organs-at-risk (OARs) and clinical target volumes (CTVs) exist. This study aimed to train deep model OAR/CTVs patients cancer undergoing radiotherapy evaluate model's feasibility efficacy not geometric indices but also comprehensive evaluation.A total 180 abdominopelvic computed tomography images were included (training set, 165; validation 15). Geometric such as Dice similarity coefficient (DSC) 95% Hausdorff distance (HD) analyzed. A Turing test was performed physicians from other institutions asked delineate contours without using auto-segmented assess inter-physician heterogeneity contouring time.The correlation between manual acceptable anorectum, bladder, spinal cord, cauda equina, right left femoral heads, bowel bag, uterocervix, liver, kidneys (DSC greater than 0.80). The stomach duodenum showed DSCs 0.67 0.73, respectively. CTVs 0.75 0.80. results favorable most OARs CTVs. No had large, obvious errors. median overall satisfaction score participating 7 out 10. Auto-segmentation reduced shortened time by 30 min among radiation oncologists different institutions. Most participants favored auto-contouring system.The proposed may be an efficient tool Although current completely replace humans, it can serve useful in real-world clinics.

Language: Английский

---

Clinical evaluation of a deep learning segmentation model including manual adjustments afterwards for locally advanced breast cancer

Technical Innovations & Patient Support in Radiation Oncology, Journal Year: 2023, Volume and Issue: 26, P. 100211 - 100211

Published: May 15, 2023

Deep learning (DL) models are increasingly developed for auto-segmentation in radiotherapy. Qualitative analysis is of great importance clinical implementation, next to quantitative. This study evaluates a DL segmentation model left- and right-sided locally advanced breast cancer both quantitatively qualitatively.For each side was trained, including primary CTV (CTVp), lymph node levels 1-4, heart, lungs, humeral head, thyroid esophagus. For evaluation, automatic segmentation, correction contours when needed, manual delineation performed processes were timed. Quantitative scoring with dice-similarity coefficient (DSC), 95% Hausdorff Distance (95%HD) surface DSC (sDSC) used compare the (not-corrected) corrected contours. by five radiotherapy technologists radiation oncologists using 3-point Likert scale.Time reduction achieved cases, correction. The time (mean ± std) 42.4% 26.5% 58.5% 19.1% OARs CTVs, respectively, corresponding an absolute mean (hh:mm:ss) 00:08:51 00:25:38. Good quantitative results before correction, e.g. CTVp 0.92 0.06, whereas statistically significantly improved this contour only 0.02 0.05, respectively. In 92% auto-contours scored as clinically acceptable, or without corrections.A trained shown be time-efficient way generate acceptable cancer.

Language: Английский

---

A qualitative, quantitative and dosimetric evaluation of a machine learning-based automatic segmentation method in treatment planning for gastric cancer

Physica Medica, Journal Year: 2025, Volume and Issue: 130, P. 104896 - 104896

Published: Jan. 7, 2025

Language: Английский

---

Automated contouring for breast cancer radiotherapy in the isocentric lateral decubitus position: a neural network-based solution for enhanced precision and efficiency

Strahlentherapie und Onkologie, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 3, 2025

Language: 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, Journal Year: 2025, Volume and Issue: unknown

Published: March 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.

Language: Английский

---