The current landscape of artificial intelligence in oral and maxillofacial surgery– a narrative review

Oral and Maxillofacial Surgery, Год журнала: 2025, Номер 29(1)

Опубликована: Янв. 17, 2025

Язык: Английский

---

Development of a deep learning-based model to evaluate changes during radiotherapy using cervical cancer digital pathology

Journal of Radiation Research, Год журнала: 2025, Номер unknown

Опубликована: Март 7, 2025

This study aims to create a deep learning-based classification model for cervical cancer biopsy before and during radiotherapy, visualize the results on whole slide images (WSIs), explore clinical significance of obtained features. included 95 patients with who received radiotherapy between April 2013 December 2020. Hematoxylin-eosin stained biopsies were digitized WSIs divided into small tiles. Our adopted feature extractor DenseNet121 classifier support vector machine. About 12 400 tiles used training 6000 testing. The performance was assessed per-tile per-WSI basis. resultant probability defined as status (RSP) its color map visualized WSIs. Survival analysis performed examine RSP. In test set, trained had an area under receiver operating characteristic curve 0.76 0.95 per-WSI. visualization, focused viable tumor components stroma in biopsies. While survival failed show prognostic impact RSP treatment, cases low at diagnosis prolonged overall compared those high (P = 0.045). conclusion, we successfully developed classify result images. Low treatment better prognosis, suggesting that morphologic features using may be useful predicting prognosis.

Язык: Английский

---

Fully automated brain resection cavity delineation for radiation target volume definition in glioblastoma patients using deep learning

Radiation Oncology, Год журнала: 2020, Номер 15(1)

Опубликована: Май 6, 2020

Abstract Background Automated brain tumor segmentation methods are computational algorithms that yield delineation from, in this case, multimodal magnetic resonance imaging (MRI). We present an automated method and its results for resection cavity (RC) glioblastoma multiforme (GBM) patients using deep learning (DL) technologies. Methods Post-operative, T1w with without contrast, T2w fluid attenuated inversion recovery MRI studies of 30 GBM were included. Three radiation oncologists manually delineated the RC to obtain a reference segmentation. developed DL method, which utilizes all four sequences learn perform delineations. evaluated terms Dice coefficient (DC) estimated volume measurements. Results Median DC three oncologist 0.85 (interquartile range [IQR]: 0.08), 0.84 (IQR: 0.07), 0.86 0.07). The automatic compared different raters 0.83 0.14), 0.81 0.12), 0.13) was significantly lower among (chi-square = 11.63, p 0.04). did not detect statistically significant difference measured volumes (Kruskal-Wallis test: chi-square 1.46, 0.69). main sources error due signal inhomogeneity similar intensity patterns between tissues. Conclusions proposed approach yields promising proof concept study. Compared human experts, still subpar.

Язык: Английский

---

Machine learning in dental, oral and craniofacial imaging: a review of recent progress

PeerJ, Год журнала: 2021, Номер 9, С. e11451 - e11451

Опубликована: Май 17, 2021

Artificial intelligence has been emerging as an increasingly important aspect of our daily lives and is widely applied in medical science. One major application artificial science imaging. As a component intelligence, many machine learning models are diagnosis treatment with the advancement technology imaging facilities. The popularity convolutional neural network dental, oral craniofacial heightening, it continually to broader spectrum scientific studies. Our manuscript reviews fundamental principles rationales behind learning, summarizes its research progress recent applications specifically It also problems that remain be resolved evaluates prospect future development this field study.

Язык: Английский

---

Artificial Intelligence in radiotherapy: state of the art and future directions

Medical Oncology, Год журнала: 2020, Номер 37(6)

Опубликована: Апрель 22, 2020

Язык: Английский

---

Validation of clinical acceptability of deep-learning-based automated segmentation of organs-at-risk for head-and-neck radiotherapy treatment planning

Frontiers in Oncology, Год журнала: 2023, Номер 13

Опубликована: Апрель 6, 2023

Organ-at-risk segmentation for head and neck cancer radiation therapy is a complex time-consuming process (requiring up to 42 individual structure, may delay start of treatment or even limit access function-preserving care. Feasibility using deep learning (DL) based autosegmentation model reduce contouring time without compromising contour accuracy assessed through blinded randomized trial oncologists (ROs) retrospective, de-identified patient data.

Язык: Английский

---

Using Auto-Segmentation to Reduce Contouring and Dose Inconsistency in Clinical Trials: The Simulated Impact on RTOG 0617

International Journal of Radiation Oncology*Biology*Physics, Год журнала: 2020, Номер 109(5), С. 1619 - 1626

Опубликована: Ноя. 13, 2020

Язык: Английский

---

External validation of deep learning-based contouring of head and neck organs at risk

Physics and Imaging in Radiation Oncology, Год журнала: 2020, Номер 15, С. 8 - 15

Опубликована: Июль 1, 2020

Background and purposeHead neck (HN) radiotherapy can benefit from automatic delineation of tumor surrounding organs because the complex anatomy regular need for adaptation. The aim this study was to assess performance a commercially available deep learning contouring (DLC) model on an external validation set.Materials methodsThe CT-based DLC model, trained at University Medical Center Groningen (UMCG), applied independent set 58 patients Radboud (RUMC). results were compared RUMC manual reference using Dice similarity coefficient (DSC) 95th percentile Hausdorff distance (HD95). Craniocaudal spatial information added by calculating binned measures. In addition, qualitative evaluation acceptance contours in both groups observers.ResultsGood correspondence shown mandible (DSC 0.90; HD95 3.6 mm). Performance reasonable glandular OARs, brainstem oral cavity 0.78–0.85, 3.7–7.3 other aerodigestive tract OARs showed only moderate agreement 0.53–0.65, around 9 measures displayed largest deviations caudally and/or cranially.ConclusionsThis demonstrates that provide starting point when patient cohort. did not reveal large differences interpretation guidelines between UMCG observers.

Язык: Английский

---

Reinventing radiation therapy with machine learning and imaging bio-markers (radiomics): State-of-the-art, challenges and perspectives

Methods, Год журнала: 2020, Номер 188, С. 44 - 60

Опубликована: Июль 19, 2020

Язык: Английский

---

Automatic clinical target volume delineation for cervical cancer in CT images using deep learning

Medical Physics, Год журнала: 2021, Номер 48(7), С. 3968 - 3981

Опубликована: Апрель 27, 2021

Purpose Accurately delineating clinical target volumes (CTV) is essential for completing radiotherapy plans but time‐consuming, labor‐intensive, and prone to inter‐observer variation. Automating CTV delineation has the benefits of both speeding up contouring process improving quality contours. Recently, auto‐segmentation approaches based on deep learning have achieved some improvements. However, unlike organ segmentation, contains potential tumor spread tissues or subclinical disease tissues, resulting in poorly defined margin interface irregular shape. It not reasonable directly apply segmentation algorithms tasks without considering unique characteristics shape margin. In this work, we propose a novel automatic algorithm addressing challenges. Methods Our method, called RA‐CTVNet, segments from cervical cancer CT images. RA‐CTVNet denotes our with A rea‐aware reweight strategy R ecursive refinement strategy. (1) order whole‐volume images delineate all CTVs one shot, method built upon popular 3D Unet architecture. We further extend it robust residual squeeze‐and‐excitation blocks better feature representation. (2) area‐aware which assigns different weights slices. The core adjusting model’s attention each slice. (3) terms trade‐off between providing performance improvements meeting limitations GPU memory, exploit new recursive address challenge. Results This retrospective study included 462 patients diagnosed who received June 2017 May 2019. Extensive experiments were conducted evaluate RA‐CTVNet. First, compared network architectures, Dice similarity coefficient (DSC). Second, ablation study. results showed that backbone, increased DSC by 3.3% average 1.6% average. Then, three human experts. performed than two experts while comparably third expert. Finally, multicenter evaluation was verify accuracy generalizability. Conclusions findings show able offer an efficient framework delineation. tailored can improve contours, great reducing burden increasing future, if more training data, are possible, bringing approach closer real practice.

Язык: Английский

---