Prediction of toxicity outcomes following radiotherapy using deep learning-based models: A systematic review

PurposeThis study aims to perform a comprehensive systematic review of deep learning (DL) models in predicting RT-induced toxicity.Materials and methodsA literature review was performed following the PRISMA guidelines. Systematic searches were performed in PubMed, Scopus, Cochrane and Embase databases from the earliest record up to September 2022. Related studies on deep learning models for radiotherapy toxicity prediction were selected based on predefined PICOS criteria.ResultsFourteen studies of radiotherapy-treated patients on different types of cancer [prostate (n = 2), HNC (n = 4), liver (n = 2), lung (n = 4), cervical (n = 1), and oesophagus (n = 1)] were eligible for inclusion in the systematic review. Information regarding patient characteristics and model development was summarized. Several approaches, such as ensemble learning, data augmentation, and transfer learning, that were utilized by selected studies were discussed.ConclusionDeep learning techniques are able to produce a consistent performance for toxicity prediction. Future research using large and diverse datasets and standardization of the study methodologies are required to improve the consistency of the research output.     

Automatic segmentation of thoracic CT images using three deep learning models

PurposeDeep learning (DL) techniques are widely used in medical imaging and in particular for segmentation. Indeed, manual segmentation of organs at risk (OARs) is time-consuming and suffers from inter- and intra-observer segmentation variability. Image segmentation using DL has given very promising results. In this work, we present and compare the results of segmentation of OARs and a clinical target volume (CTV) in thoracic CT images using three DL models.Materials and methodsWe used CT images of 52 patients with breast cancer from a public dataset. Automatic segmentation of the lungs, the heart and a CTV was performed using three models based on the U-Net architecture. Three metrics were used to quantify and compare the segmentation results obtained with these models: the Dice similarity coefficient (DSC), the Jaccard coefficient (J) and the Hausdorff distance (HD).ResultsThe obtained values of DSC, J and HD were presented for each segmented organ and for the three models. Examples of automatic segmentation were presented and compared to the corresponding ground truth delineations. Our values were also compared to recent results obtained by other authors.ConclusionThe performance of three DL models was evaluated for the delineation of the lungs, the heart and a CTV. This study showed clearly that these 2D models based on the U-Net architecture can be used to delineate organs in CT images with a good performance compared to other models. Generally, the three models present similar performances. Using a dataset with more CT images, the three models should give better results.     

Different radiation techniques to deliver therapeutic dose to the axilla in patients with sentinel lymph node-positive breast cancer: Doses,techniques challenges and clinical considerations

Purpose

To evaluate the coverage of different levels of axillary lymph nodes and organs at risk according to the field design of AMAROS study (levels I–II–III–IV), breast tangents with supraclavicular and infraclavicular fields (levels II–III–IV) and high tangent fields to the breast after breast-conserving surgery.

鼻咽癌调强放疗中靶区及危及器官的体积和剂量变化

目前调强放疗是鼻咽癌的主要治疗手段.在6～7周的疗程中,临床大部分医师均使用初始制定的放疗计划.研究表明,由于原发肿瘤和转移淋巴结的缩小以及体质量下降导致的正常组织和体表轮廓的收缩,这些有意义的变化将影响靶区及危及器官的剂量分布,从而影响部分患者的治疗疗效.鼻咽癌调强放疗期间复位并进行靶区修改很有必要.     

Applicators used for vaginal high dose rate brachytherapy: Effect of type and shape on dose distribution and toxicity,a literature review

The aim of this article is to represent a literature review on the applicators used for vaginal high dose rate brachytherapy for patients with endometrial cancer. The different types and shapes of the clinically used applicator as well as the effects of their characteristics on dose distribution, target coverage and dose received by organs at risk are discussed in detail.     

IMRT中OAR和靶区相交体积与其吸收剂量间关系研究

目的 探讨IMRT中OAR与靶区相交体积(V_overlap)和平均剂量(D_mOAR)之间的相关关系。方法 随机选取本院接受鼻咽癌IMRT的患者与宫颈癌根治性IMRT的患者各50例,以此为模板分析IMRT中V_overlap与其D_mOAR之间的关系,并利用matlab软件拟合出不同OAR上述二者之间的相关公式。结果 鼻咽癌与宫颈癌患者OAR和靶区相交体积因人而异,V_overlap与其本身体积(V_OAR)的比值与D_mOAR和处方剂量(D_p)比值之间存在着显著正相关性(P均=0.01),并得到出二者之间相关公式。结论 IMRT过程中患者靶区大小各异,不同分期和OAR充盈程度及大小不一等因素也会导致不同V_overlap/V_OAR。在制作放疗计划前根据不同OAR拟合出相关公式,在给定处方条件下通过V_overlap/V_OAR提前计算出各自D_mOAR,作为计划验收标准,从而减少放疗计划优化过程中主观因素造成影响。     

Brain dose-sparing radiotherapy techniques for localized intracranial germinoma: Case report and literature review of modern irradiation

We examined the effects of intensity-modulated radiation therapy with dose-sparing and avoidance technique on a pediatric patient with localized intracranial germinoma. We also reviewed the literature regarding modern irradiation techniques in relation to late neurocognitive sequelae. A patient with a localized intracranial germinoma in the third ventricle anterior to the pineal gland received a dose-sparing intensity-modulated radiation therapy. The planning was compared to the radiation oncologist's guide of organs at risk and dose constraints for dosimetric analyses. The patient received radiation therapy alone. The total dose was 54 Gy delivered in 2.0 Gy fractions to the primary tumour and 37 Gy in 1.4 Gy fractions to whole ventricles using a dose-sculpting plan. Dosimetry analyses showed that dose-sparing intensity-modulated radiation therapy delivered reduced doses to the whole brain, temporal lobes, hippocampi, cochleae, and optic nerves. With a follow-up of 22 months, failure-free survival was 100% for the patient and no adverse events during radiation treatment process. Intensity-modulated radiation therapy with dose sparing and avoidance technique can spare the limbic circuit, central nervous system, and hippocampus for pineal germ cell tumours. This technique reduces the integral dose delivered to the uninvolved normal brain tissues and may reduce late neurocognitive sequelae caused by cranial radiotherapy.     

Impact of the method chosen for the analysis of recurrences after radiotherapy for head and neck cancers: volume-based,point-based and combined methods

Intensity modulated radiation therapy for head and neck is a complex technique. Inappropriate delineation and/or dose distribution can lead to recurrences. Analysis of these recurrences should lead to improve clinical practice. For several years, different methods of analysis have been described. The purpose of this review is to describe these different methods and to discuss their advantages and limitations. The first published methods used a volume-based approach studying the entire volume of recurrence according to initial target volumes, or dose distribution. The main limitation of these methods was that the volume of recurrence studied was dependent on the delay in diagnosis of that recurrence. Subsequently, other methods used point-based approaches, conceptualizing recurrence either as a spherical expansion from a core of radioresistant cells (center of mass of recurrence volume) or using a more clinical approach, taking into account tumor expansion pathways. More recently, more precise combined methods have been described, combining the different approaches. The choice of method is decisive for conclusions on the origin of recurrence.     

人工智能技术在鼻咽癌放疗危及器官自动勾画中的应用研究

目的 评估基于人工智能技术的放疗危及器官自动勾画软件在临床应用中的价值.方法 选取2019年4月—2020年4月在自贡市第一人民医院肿瘤科治疗的30例鼻咽癌患者进行研究.首先,使用AccuContour软件勾画30例鼻咽癌患者的危及器官;然后使用瓦里安Eclipse v13.6计划系统人为手工勾画危及器官;最后将自动勾...     

Delineation of organs at risk involved in swallowing for radiotherapy treatment planning

Background and purpose

Radiotherapy, alone or combined with chemotherapy, is a treatment modality used frequently in head and neck cancer. In order to report, compare and interpret the sequelae of radiation treatment adequately, it is important to delineate organs at risk (OARs) according to well-defined and uniform guidelines. The aim of this paper was to present our institutional Computed Tomography (CT)-based delineation guidelines for organs in the head and neck at risk for radiation-induced swallowing dysfunction (SWOARs).

Material and methods

After analyses of the human anatomy of the head and neck area and literature review, CT-based guidelines for delineation of the most relevant SWOARs were described by a panel of experts.

Results and conclusions

This paper described institutional guidelines for the delineation of potential SWOARs, accompanied by CT-based illustrations presenting examples of the delineated structures and their corresponding anatomic borders. This paper is essential to ensure adequate interpretation of future reports on the relationship between dose distribution in these SWOARs and different aspects of post-treatment swallowing dysfunction.     

Contouring workload in adjuvant breast cancer radiotherapy

Purpose

To measure the impact of contouring on worktime in the adjuvant radiation treatment of breast cancer, and to identify factors that might affect the measurements.

Clinical evaluation of atlas and deep learning based automatic contouring for lung cancer

Background and purpose

Contouring of organs at risk (OARs) is an important but time consuming part of radiotherapy treatment planning. The aim of this study was to investigate whether using institutional created software-generated contouring will save time if used as a starting point for manual OAR contouring for lung cancer patients.

Évaluation d’un logiciel pour la délinéation automatique des organes à risques et des volumes cibles ganglionnaires chez des patientes prises en charge pour un cancer du sein

Purpose

Intensity-modulated radiotherapy needs the strict delineation of target volumes as well as organs at risk and the time used for this procedure is long. The purpose of this study was to evaluate the Workflow Box system (Mirada Medical, UK) for automatic delineation and segmentation for everyday use of organs at risk and lymph nodes delineation in patients treated for early stage breast cancer.

Comparison of passive-beam proton therapy,helical tomotherapy and 3D conformal radiation therapy in Hodgkin's lymphoma female patients receiving involved-field or involved site radiation therapy

PurposeSecond cancers and cardiovascular toxicities are long term radiation toxicity in locally advanced Hodgkin's lymphomas. In this study, we evaluate the potential reduction of dose to normal tissue with helical tomotherapy and proton therapy for Hodgkin's lymphoma involved-field or involved-site irradiation compared to standard 3D conformal radiation therapy.Patients and methodsFourteen female patients with supradiaphragmatic Hodgkin's lymphoma were treated at our institution with 3D conformal radiation therapy or helical tomotherapy to a dose of 30 Gy in 15 fractions. A planning comparison was achieved including proton therapy with anterior/posterior passive scattered beams weighted 20 Gy/10 Gy.ResultsMean doses to breasts, lung tissue and heart with proton therapy were significantly lower compared to helical tomotherapy and to 3D conformal radiation therapy. Helical tomotherapy assured the best protection of lungs from doses above 15 Gy with the V_20 Gy equal to 16.4%, compared to 19.7% for proton therapy (P = 0.01) or 22.4% with 3D conformal radiation therapy (P < 0.01). Volumes of lung receiving doses below 15 Gy were significantly larger for helical tomotherapy than for proton therapy or 3D conformal radiation therapy, with respective lung doses V_10 Gy = 37.2%, 24.6% and 27.4%. Also, in the domain of low doses, the volumes of breast that received more than 10 Gy or more than 4 Gy with helical tomotherapy were double the corresponding volumes for proton therapy, with V_4 Gy representing more than a third of one breast volume with helical tomotherapy.ConclusionsHelical tomotherapy achieved a better protection to the lungs for doses above 15 Gy than passive proton therapy or 3D conformal radiation therapy. However, dose distributions could generally be improved by using protons even with our current passive-beam technology, especially allowing less low dose spreading and better breast tissue sparing, which is an important factor to consider when treating Hodgkin's lymphomas in female patients. Prospective clinical study is needed to evaluate the tolerance and confirm these findings.     

鼻咽癌IMRT中OAR勾画差异所致剂量不确定性研究

目的 观察鼻咽癌IMRT中低年资与高年资医师勾画OAR的差异和剂量学差异,评估针对性的培训对改善差异的作用。方法 选择初治鼻咽癌患者 16例,组织低年资和高年资医师各3位分别勾画患者OAR并与参考OAR比较几何差异和剂量学差异。针对差异最大的2个OAR,组织两组医师培训后重新勾画并再次比较差异。差异比较行配对t检验。结果 低年资、高年资医师勾画OAR的 D_max差异分别为(2.33±12.06)%(－48.06%~137.82%)、(0.09±4.72)%(－49.54%~42.96%)(P=0.039),其中视交叉的差异最大[(5.85±19.63)%∶(1.36±4.64)%,P=0.042];D_mean差异分别为(3.10±8.07)%(－46.76%~59.76%)、(－0.93±2.03)%(－45.54%~35.69%)(P=0.021),其中腮腺的差异最大[(13.23±13.39)%∶(3.20±6.71)%,P=0.002]。培训并重新勾画后低年资、高年资医师勾画视交叉的 D_max差异分别为(1.68±3.34)%、(1.50±1.87)%(P=0.841),其中低年资医师的差异较培训前明显减少[(1.68±3.34)%∶(5.85±19.63)%,P=0.048];腮腺的 D_mean差异分别为(2.46±3.06)%、(1.35±3.00)%(P=0.274),分别较培训前的差异明显减少[(2.46±3.06)%∶(13.23±13.39)%,P=0.002;(1.35±3.00)%∶(3.20±6.71)%,P=0.033]。 结论 鼻咽癌IMRT中OAR的勾画差异可引起明显的剂量不确定性,针对性的培训可提高勾画精准性。     

Recommendation for a contouring method and atlas of organs at risk in nasopharyngeal carcinoma patients receiving intensity-modulated radiotherapy

Background and purpose

To recommend contouring methods and atlas of organs at risk (OARs) for nasopharyngeal carcinoma (NPC) patients receiving intensity-modulated radiotherapy, in order to help reach a consensus on interpretations of OARs delineation.

Methods and materials

Two to four contouring methods for the middle ear, inner ear, temporal lobe, parotid gland and spinal cord were identified via systematic literature review; their volumes and dosimetric parameters were compared in 41 patients. Areas under the receiver operating characteristic curves for temporal lobe contouring were compared in 21 patients with unilateral temporal lobe necrosis (TLN).

Results

Various contouring methods for the temporal lobe, middle ear, inner ear, parotid gland and spinal cord lead to different volumes and dosimetric parameters (P < 0.05). For TLN, D1 of PRV was the most relevant dosimetric parameter and 64 Gy was the critical point. We suggest contouring for the temporal lobe, middle ear, inner ear, parotid gland and spinal cord. A CT–MRI fusion atlas comprising 33 OARs was developed.

Conclusions

Different dosimetric parameters may hinder the dosimetric research. The present recommendation and atlas, may help reach a consensus on subjective interpretation of OARs delineation to reduce inter-institutional differences in NPC patients.     

基于多任务学习的轻量级卷积神经网络在肺部危及器官自动分割中的应用研究

目的:研究基于多任务学习轻量级卷积神经网络(MTLW-CNN)在肺部危及器官(OAR)自动分割中的应用。方法:MTLW-CNN包含多个特征共享层和OAR的对应勾画分支。收集497例肺癌病例,以同时包含肺、心脏和脊髓的CT及医生勾画的轮廓为样本,随机划分300例用于训练与验证,197例用于测试。在测试集应用MTLW-CN...     

基于深度学习的宫颈癌放疗靶区及危及器官自动勾画研究

目的:构建基于深度学习（deep learning,DL）的卷积神经网络模型,实现宫颈癌患者放射治疗计划的临床靶区体积（clinical target volume,CTV）和危及器官（organ at risks,OARs）自动勾画。方法:回顾性分析在福建省肿瘤医院行放射治疗的宫颈癌患者99例。对患者CT图像进行预处理,作为模型输入。设计一种基于DL的自动勾画模型,使用组合损失函数训练该模型。以医师手动勾画为度量标准,计算DL自动勾画模型下CTV靶区和膀胱、直肠、乙状结肠、左右骨髓、左右股骨头的准确率,并与基于图谱的自动勾画方法（atlas-based auto segmentation,ABAS）相比较。结果:DL模型在CTV靶区和7种危及器官（膀胱、直肠、乙状结肠、左右骨髓、左右股骨头）的戴斯系数分别为（0.85±0.02、0.94±0.04、0.87±0.03、0.67±0.14、0.85±0.03、0.87±0.03、0.87±0.06和0.87±0.06）,95%豪斯多夫距离(mm)分别为（3.22±0.56、1.37±0.37、1.41±0.34、27.39±35.63、1.40±0.17、1.36±0.22、6.78±7.89和6.45±7.44）,平均表面距离(mm)分别为（0.25±0.05、0.12±0.06、0.19±0.05、2.29±2.71、0.16±0.04、0.15±0.03、0.36±0.33和0.38±0.37）。DL勾画模型的戴斯系数均高于ABAS勾画模型。除乙状结肠外,DL勾画模型的95%豪斯多夫距离和平均表面距离均小于ABAS勾画模型。结论:提出的DL模型能较好地实现宫颈癌放疗临床靶区和危及器官的自动勾画,可为临床医师勾画提供初步参考,节省临床靶区和危及器官勾画的时间。