Deep learning-based auto-delineation of gross tumour volumes and involved nodes in PET/CT images of head and neck cancer patients.
Ontology highlight
ABSTRACT: Purpose
Identification and delineation of the gross tumour and malignant nodal volume (GTV) in medical images are vital in radiotherapy. We assessed the applicability of convolutional neural networks (CNNs) for fully automatic delineation of the GTV from FDG-PET/CT images of patients with head and neck cancer (HNC). CNN models were compared to manual GTV delineations made by experienced specialists. New structure-based performance metrics were introduced to enable in-depth assessment of auto-delineation of multiple malignant structures in individual patients.Methods
U-Net CNN models were trained and evaluated on images and manual GTV delineations from 197 HNC patients. The dataset was split into training, validation and test cohorts (n= 142, n = 15 and n = 40, respectively). The Dice score, surface distance metrics and the new structure-based metrics were used for model evaluation. Additionally, auto-delineations were manually assessed by an oncologist for 15 randomly selected patients in the test cohort.Results
The mean Dice scores of the auto-delineations were 55%, 69% and 71% for the CT-based, PET-based and PET/CT-based CNN models, respectively. The PET signal was essential for delineating all structures. Models based on PET/CT images identified 86% of the true GTV structures, whereas models built solely on CT images identified only 55% of the true structures. The oncologist reported very high-quality auto-delineations for 14 out of the 15 randomly selected patients.Conclusions
CNNs provided high-quality auto-delineations for HNC using multimodality PET/CT. The introduced structure-wise evaluation metrics provided valuable information on CNN model strengths and weaknesses for multi-structure auto-delineation.
SUBMITTER: Moe YM
PROVIDER: S-EPMC8263429 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:BackgroundGallium-68 lung perfusion PET/CT is an emerging imaging modality for the assessment of regional lung function, especially to optimise radiotherapy (RT) planning. A key step of lung functional avoidance RT is the delineation of lung functional volumes (LFVs) to be integrated into radiation plans. However, there is currently no consistent and reproducible delineation method for LFVs. The aim of this study was to develop and evaluate an automated delineation threshold method based on total lung function for LFVs delineation with Gallium-68 MAA lung PET/CT imaging.Material and methodPatients prospectively enrolled in the PEGASUS trial-a pilot study assessing the feasibility of lung functional avoidance using perfusion PET/CT imaging for lung stereotactic body radiotherapy (SBRT) of primary or secondary lesion-were analysed. Patients underwent lung perfusion MAA-68Ga PET/CT imaging and pulmonary function tests (PFTs) as part of pre-treatment evaluation. LFVs were delineated using two methods: the commonly used relative to the maximal pixel value threshold method (pmax threshold method, X%pmax volumes) and a new approach based on a relative to whole lung function threshold method (WLF threshold method, FVX% volumes) using a dedicated iterative algorithm. For both methods, LFVs were expressed in terms of % of the anatomical lung volume (AV) and of % of the total lung activity. Functional volumes were compared for patients with normal PFTs and pre-existing airway disease.Results60 patients were analysed. Among the 48 patients who had PFTs, 31 (65%) had pre-existing lung disease. The pmax and WLF threshold methods clearly provided different functional volumes with a wide range of relative lung function for a given pmax volume, and conversely, a wide range of corresponding pmax values for a given WLF volume. The WLF threshold method provided more reliable and consistent volumes with much lower dispersion of LFVs as compared to the pmax method, especially in patients with normal PFTs.ConclusionsWe developed a relative to whole lung function threshold segmentation method to delineate lung functional volumes on perfusion PET/CT imaging. The automated algorithm allows for reproducible contouring. This new approach, relatively unaffected by the presence of hot spots, provides reliable and consistent functional volumes, and is clinically meaningful for clinicians.
| S-EPMC10093378 | biostudies-literature
Project description:BackgroundThe delineation of intraprostatic lesions is vital for correct delivery of focal radiotherapy boost in patients with prostate cancer (PC). Errors in the delineation could translate into reduced tumour control and potentially increase the side effects. The purpose of this study is to compare PET-based delineation methods with histopathology.Materials and methodsThe study population consisted of 15 patients with confirmed high-risk PC intended for prostatectomy. [68Ga]-PSMA-PET/MR was performed prior to surgery. Prostate lesions identified in histopathology were transferred to the in vivo [68Ga]-PSMA-PET/MR coordinate system. Four radiation oncologists manually delineated intraprostatic lesions based on PET data. Various semi-automatic segmentation methods were employed, including absolute and relative thresholds, adaptive threshold, and multi-level Otsu threshold.ResultsThe gross tumour volumes (GTVs) delineated by the oncologists showed a moderate level of interobserver agreement with Dice similarity coefficient (DSC) of 0.68. In comparison with histopathology, manual delineations exhibited the highest median DSC and the lowest false discovery rate (FDR) among all approaches. Among semi-automatic approaches, GTVs generated using standardized uptake value (SUV) thresholds above 4 (SUV > 4) demonstrated the highest median DSC (0.41), with 0.51 median lesion coverage ratio, FDR of 0.66 and the 95th percentile of the Hausdorff distance (HD95%) of 8.22 mm.InterpretationManual delineations showed a moderate level of interobserver agreement. Compared to histopathology, manual delineations and SUV > 4 exhibited the highest DSC and the lowest HD95% values. The methods that resulted in a high lesion coverage were associated with a large overestimation of the size of the lesions.
| S-EPMC11332483 | biostudies-literature
Project description:ObjectiveTo localize the distribution of regional nodes in recurrent/advanced breast cancer patients based on 18-fludeoxyglucose (FDG) positron emission tomography/CT (PET/CT) images and validate the coverage of clinical target volumes (CTVs) for regional nodes with current contouring guidelines.MethodsWe enrolled 154 recurrent/advanced breast cancer patients with FDG-avid regional nodes who underwent PET/CT between January 2018 and June 2020. Involvement of lymph node regions including axillary lymph node level I-III (ALN-I, ALN-II, ALN-III), Rotter's nodes (RN), medial supraclavicular (SC-M), lateral supraclavicular (SC-L) and internal mammary nodes (IMN) was recorded respectively. Coverage of the CTVs in different atlases and the locations of out-of-field were evaluated.ResultsA total of 348 lymph node regions containing disease were identified, including ALN-I 109, ALN-II 46, ALN-III 36, RN 17, SC-M 68, SC-L 36 and IMN 36. Recurrent ALNs mainly located cranially and ventrally to the axillary vein (AV). Ipsilateral cervical nodes were simultaneously affected in 33/76 SC positive patients. RADCOMP (306/348) and RUIJIN (291/348) guidelines had higher coverage compared with RTOG (205/348) and ESTRO (202/348) guidelines (p < 0.001, respectively). In primary non-metastastic and recurrent patients, major missings located in SC-L (7/7, 17/17) and IMN (7/10, 15/19) for RTOG guideline while SC-L (7/7, 17/17) for ESTRO guideline (p < 0.001, respectively). Among recurrent patients, SC-M (22/31) was another major missing area for ESTRO guideline (p < 0.001).ConclusionThe current guidelines effectively cover most regional nodes in postoperative breast cancer patients. SC-L and IMN were the major missing regions. Recurrent ALNs were most often seen in cranial and ventral to the AV. The CTV of patients with clinically positive SC was recommended to extend up to the hyoid level. The CTVs should be adjusted based on risks of recurrence individually.Advances in knowledgeThe difference of regional nodes delineation between current guidelines mainly located in SC and IMN regions. High axilla including subclavicular nodes and the RN above AV for recurrent patients and the region between cricoid and hyoid for positive SC patients should be meticulously contoured.
| S-EPMC10996945 | biostudies-literature
Project description:PurposeDelineation of tumour boundaries is important for quantification of [(18)F]fluoro-2-deoxy-D-glucose (FDG) positron emission tomography (PET) studies and for definition of biological target volumes in radiotherapy. Several (semi-)automatic tumour delineation methods have been proposed, but these methods differ substantially in estimating tumour volume and their performance may be affected by imaging parameters. The main purpose of this study was to explore the performance dependence of various (semi-)automatic tumour delineation methods on different imaging parameters, i.e. reconstruction parameters, noise levels and tumour characteristics, and thereby the need for standardization or inter-institute calibration.MethodsSix different types of delineation methods were evaluated by assessing accuracy and precision in estimating tumour volume from simulations and phantom experiments. The evaluated conditions were various tumour sizes, iterative reconstruction algorithm settings and image filtering, tumour to background ratios (TBR), noise levels and region growing initializations.ResultsThe accuracy of all automatic delineation methods was influenced when imaging parameters were varied. The performance of all tumour delineation methods depends on variation of TBR, image resolution and image noise level, and to a lesser extent on number of iterations during image reconstruction or the initialization method of the region generation. For sphere sizes larger than 20 mm diameter a contrast-oriented method provided the most accurate results, on average, over all simulated conditions. For threshold-based methods the accuracy of tumour delineation improved after image denoising/filtering.ConclusionThe accuracy and precision of all studied tumour delineation methods was affected by physiological and imaging parameters. The latter illustrates the need for optimizing imaging parameters and/or for careful calibration and optimization of delineation methods.
| S-EPMC3228515 | biostudies-literature
Project description: Not available
| S-EPMC5368524 | biostudies-literature
Project description:Computed tomographic (CT) images are widely used for the identification of abnormal brain tissue following infarct and hemorrhage in stroke. Manual lesion delineation is currently the standard approach, but is both time-consuming and operator-dependent. To address these issues, we present a method that can automatically delineate infarct and hemorrhage in stroke CT images. The key elements of this method are the accurate normalization of CT images from stroke patients into template space and the subsequent voxelwise comparison with a group of control CT images for defining areas with hypo- or hyper-intense signals. Our validation, using simulated and actual lesions, shows that our approach is effective in reconstructing lesions resulting from both infarct and hemorrhage and yields lesion maps spatially consistent with those produced manually by expert operators. A limitation is that, relative to manual delineation, there is reduced sensitivity of the automated method in regions close to the ventricles and the brain contours. However, the automated method presents a number of benefits in terms of offering significant time savings and the elimination of the inter-operator differences inherent to manual tracing approaches. These factors are relevant for the creation of large-scale lesion databases for neuropsychological research. The automated delineation of stroke lesions from CT scans may also enable longitudinal studies to quantify changes in damaged tissue in an objective and reproducible manner.
| S-EPMC3984449 | biostudies-literature
Project description:BackgroundThis study investigates the patterns of PET-positive lymph nodes (LNs) in anal cancer. The aim was to provide information that could inform future anal cancer radiotherapy contouring guidelines.MethodsThe baseline [18F]-FDG PET-CTs of 190 consecutive anal cancer patients were retrospectively assessed. LNs with a Deauville score (DS) of ≥3 were defined as PET-positive. Each PET-positive LN was allocated to a LN region and a LN sub-region; they were then mapped on a standard anatomy reference CT. The association between primary tumor localization and PET-positive LNs in different regions were analyzed.ResultsPET-positive LNs (n = 412) were identified in 103 of 190 patients (54%). Compared to anal canal tumors with extension into the rectum, anal canal tumors with perianal extension more often had inguinal (P < 0.001) and less often perirectal (P < 0.001) and internal iliac (P < 0.001) PET-positive LNs. Forty-two patients had PET-positive LNs confined to a solitary region, corresponding to first echelon nodes. The most common solitary LN region was inguinal (25 of 42; 60%) followed by perirectal (26%), internal iliac (10%), and external iliac (2%). No PET-positive LNs were identified in the ischiorectal fossa or in the inguinal area located posterolateral to deep vessels. Skip metastases above the bottom of the sacroiliac joint were quite rare. Most external iliac PET-positive LNs were located posterior to the external iliac vein; only one was located in the lateral external iliac sub-region.ConclusionsThe results support some specific modifications to the elective clinical target volume (CTV) in anal cancer. These changes would lead to reduced volumes of normal tissue being irradiated, which could contribute to a reduction in radiation side-effects.
| S-EPMC8063376 | biostudies-literature
Project description:ObjectivesTo propose an easily applicable segmentation method (perPET-RT) for delineation of tumour volume during radiotherapy on interim fluorine 18 fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) in patients with non-small cell lung cancer (NSCLC).Material and methodsSixty-seven patients (51 primary tumours, 60 lymph nodes), from 4 prospective studies, underwent an FDG PET/CT scan during the fifth week of radiation therapy, using different generations of PET/CT. Per-therapeutic PET/CT scans were delineated in consensus by two experienced physicians leading to the gold standard threshold to be applied. The mathematical expression of Thopt, the optimal threshold to be applied as a function of the maximum standard uptake value (SUVmax), was determined. The performance of this method (perPET-RT) was assessed by computing the DICE similarity coefficient (DSC) and was compared with 8 fixed threshold values and 3 adaptive thresholding methods.ResultsThopt verified the following expression: Thopt = A.ln(1/SUVmax) + B where A and B were 2 constants. A and B were independent from the generation of PET/CT, but depended on the type of lesions (primary lung tumours vs. lymph nodes). PerPET-RT showed good to very good agreement in comparison to the gold standard. The mean and standard deviation of DSC value was 0.81 ± 0.13 for lung lesions and 0.78 ± 0.15 for lymph nodes. PerPET-RT showed a significant better agreement than the other segmentation methods (p < 0.001), except for one of the adaptive thresholding method ADT (p = 0.11).ConclusionOn the database used, perPET-RT has proven its reliability and accuracy for tumour delineation on per-therapeutic FDG PET/CT using only SUVmax measurement. This method may be used to delineate tumour volume for dose-escalation planning.Trial registrationNCT01261598 , NCT01261585 , NCT01576796 .
| S-EPMC6233287 | biostudies-literature
Project description:BackgroundThe clinical validation of the EARL harmonization program for standardised uptake value (SUV) metrics is well documented; however, its potential for defining metabolic active tumour volume (MATV) has not yet been investigated. We aimed to compare delineation of MATV on images reconstructed using conventional ordered subset expectation maximisation (OSEM) with those reconstructed using point spread function modelling (PSF-reconstructed images), and either optimised for diagnostic potential (PSF) or filtered to meet the EANM/EARL harmonising standards (PSF7).MethodsImages from 18 stage IIIA-IIIB lung cancer patients were reconstructed using all the three methods. MATVs were then delineated using both a 40% isocontour and a gradient-based method. MATVs were compared by means of Bland-Altman analyses, and Dice coefficients and concordance indices based on the unions and intersections between each pair of reconstructions (PSF vs OSEM, PSF7 vs PSF and PSF7 vs OSEM).ResultsUsing the 40% isocontour method and taking the MATVs delineated on OSEM images as a reference standard, the use of PSF7 images led to significantly higher Dice coefficients (median value = 0.96 vs 0.77; P < 0.0001) and concordance indices (median value = 0.92 vs 0.64; P < 0.0001) than those obtained using PSF images. The gradient-based methodology was less sensitive to reconstruction variability than the 40% isocontour method; Dice coefficients and concordance indices were superior to 0.8 for both PSF reconstruction methods. However, the use of PSF7 images led to narrower interquartile ranges and significantly higher Dice coefficients (median value = 0.96 vs 0.94; P = 0.01) and concordance indices (median value = 0.89 vs 0.85; P = 0.003) than those obtained with PSF images.ConclusionThis study demonstrates that automatic contouring of lung tumours on EARL-compliant PSF images using the widely adopted automatic isocontour methodology is an accurate means of overcoming reconstruction variability in MATV delineation. Although gradient-based methodology appears to be less sensitive to reconstruction variability, the use of EARL-compliant PSF images significantly improved the Dice coefficients and concordance indices, demonstrating the importance of harmonised-images, even when more advanced contouring algorithms are used.
| S-EPMC5374086 | biostudies-literature
Project description:PurposeAutomatic segmentation offers many benefits for radiotherapy treatment planning; however, the lack of publicly available benchmark datasets limits the clinical use of automatic segmentation. In this work, we present a well-curated computed tomography (CT) dataset of high-quality manually drawn contours from patients with thoracic cancer that can be used to evaluate the accuracy of thoracic normal tissue auto-segmentation systems.Acquisition and validation methodsComputed tomography scans of 60 patients undergoing treatment simulation for thoracic radiotherapy were acquired from three institutions: MD Anderson Cancer Center, Memorial Sloan Kettering Cancer Center, and the MAASTRO clinic. Each institution provided CT scans from 20 patients, including mean intensity projection four-dimensional CT (4D CT), exhale phase (4D CT), or free-breathing CT scans depending on their clinical practice. All CT scans covered the entire thoracic region with a 50-cm field of view and slice spacing of 1, 2.5, or 3 mm. Manual contours of left/right lungs, esophagus, heart, and spinal cord were retrieved from the clinical treatment plans. These contours were checked for quality and edited if necessary to ensure adherence to RTOG 1106 contouring guidelines.Data format and usage notesThe CT images and RTSTRUCT files are available in DICOM format. The regions of interest were named according to the nomenclature recommended by American Association of Physicists in Medicine Task Group 263 as Lung_L, Lung_R, Esophagus, Heart, and SpinalCord. This dataset is available on The Cancer Imaging Archive (funded by the National Cancer Institute) under Lung CT Segmentation Challenge 2017 (http://doi.org/10.7937/K9/TCIA.2017.3r3fvz08).Potential applicationsThis dataset provides CT scans with well-delineated manually drawn contours from patients with thoracic cancer that can be used to evaluate auto-segmentation systems. Additional anatomies could be supplied in the future to enhance the existing library of contours.
| S-EPMC8344378 | biostudies-literature