A cascaded FAS-UNet+ framework with iterative optimization strategy for segmentation of organs at risk.

Segmentation of organs at risks (OARs) in the thorax plays a critical role in radiation therapy for lung and esophageal cancer. Although automatic segmentation of OARs has been extensively studied, it remains challenging due to the varying sizes and shapes of organs, as well as the low contrast between the target and background. This paper proposes a cascaded FAS-UNet+ framework, which integrates convolutional neural networks and nonlinear multi-grid theory to solve a modified Mumford-shah model for segmenting OARs. This framework is equipped with an enhanced iteration block, a coarse-to-fine multiscale architecture, an iterative optimization strategy, and a model ensemble technique. The enhanced iteration block aims to extract multiscale features, while the cascade module is used to refine coarse segmentation predictions. The iterative optimization strategy improves the network parameters to avoid unfavorable local minima. An efficient data augmentation method is also developed to train the network, which significantly improves its performance. During the prediction stage, a weighted ensemble technique combines predictions from multiple models to refine the final segmentation. The proposed cascaded FAS-UNet+ framework was evaluated on the SegTHOR dataset, and the results demonstrate significant improvements in Dice score and Hausdorff Distance (HD). The Dice scores were 95.22%, 95.68%, and HD values were 0.1024, and 0.1194 for the segmentations of the aorta and heart in the official unlabeled dataset, respectively. Our code and trained models are available at https://github.com/zhuhui100/C-FASUNet-plus .

Clinical evaluation of the convolutional neural network­based automatic delineation tool in determining the clinical target volume and organs at risk in rectal cancer radiotherapy.

Delineating the clinical target volume (CTV) and organs at risk (OARs) is crucial in rectal cancer radiotherapy. However, the accuracy of manual delineation (MD) is variable and the process is time consuming. Automatic delineation (AD) may be a solution to produce quicker and more accurate contours. In the present study, a convolutional neural network (CNN)-based AD tool was clinically evaluated to analyze its accuracy and efficiency in rectal cancer. CT images were collected from 148 supine patients in whom tumor stage and type of surgery were not differentiated. The rectal cancer contours consisted of CTV and OARs, where the OARs included the bladder, left and right femoral head, left and right kidney, spinal cord and bowel bag. The MD contours reviewed and modified together by a senior radiation oncologist committee were set as the reference values. The Dice similarity coefficient (DSC), Jaccard coefficient (JAC) and Hausdorff distance (HD) were used to evaluate the AD accuracy. The correlation between CT slice number and AD accuracy was analyzed, and the AD accuracy for different contour numbers was compared. The time recorded in the present study included the MD time, AD time for different CT slice and contour numbers and the editing time for AD contours. The Pearson correlation coefficient, paired-sample t-test and unpaired-sample t-test were used for statistical analyses. The results of the present study indicated that the DSC, JAC and HD for CTV using AD were 0.80±0.06, 0.67±0.08 and 6.96±2.45 mm, respectively. Among the OARs, the highest DSC and JAC using AD were found for the right and left kidney, with 0.91±0.06 and 0.93±0.04, and 0.84±0.09 and 0.88±0.07, respectively, and HD was lowest for the spinal cord with 2.26±0.82 mm. The lowest accuracy was found for the bowel bag. The more CT slice numbers, the higher the accuracy of the spinal cord analysis. However, the contour number had no effect on AD accuracy. To obtain qualified contours, the AD time plus editing time was 662.97±195.57 sec, while the MD time was 3294.29±824.70 sec. In conclusion, the results of the present study indicate that AD can significantly improve efficiency and a higher number of CT slices and contours can reduce AD efficiency. The AD tool provides acceptable CTV and OARs for rectal cancer and improves efficiency for delineation.

Automated confidence estimation in deep learning auto-segmentation for brain organs at risk on MRI for radiotherapy.

PURPOSE: We have built a novel AI-driven QA method called AutoConfidence (ACo), to estimate segmentation confidence on a per-voxel basis without gold standard segmentations, enabling robust, efficient review of automated segmentation (AS). We have demonstrated this method in brain OAR AS on MRI, using internal and external (third-party) AS models. METHODS: Thirty-two retrospectives, MRI planned, glioma cases were randomly selected from a local clinical cohort for ACo training. A generator was trained adversarialy to produce internal autosegmentations (IAS) with a discriminator to estimate voxel-wise IAS uncertainty, given the input MRI. Confidence maps for each proposed segmentation were produced for operator use in AS editing and were compared with "difference to gold-standard" error maps. Nine cases were used for testing ACo performance on IAS and validation with two external deep learning segmentation model predictions [external model with low-quality AS (EM-LQ) and external model with high-quality AS (EM-HQ)]. Matthew's correlation coefficient (MCC), false-positive rate (FPR), false-negative rate (FNR), and visual assessment were used for evaluation. Edge removal and geometric distance corrections were applied to achieve more useful and clinically relevant confidence maps and performance metrics. RESULTS: ACo showed generally excellent performance on both internal and external segmentations, across all OARs (except lenses). MCC was higher on IAS and low-quality external segmentations (EM-LQ) than high-quality ones (EM-HQ). On IAS and EM-LQ, average MCC (excluding lenses) varied from 0.6 to 0.9, while average FPR and FNR were ≤0.13 and ≤0.21, respectively. For EM-HQ, average MCC varied from 0.4 to 0.8, while average FPR and FNR were ≤0.37 and ≤0.22, respectively. CONCLUSION: ACo was a reliable predictor of uncertainty and errors on AS generated both internally and externally, demonstrating its potential as an independent, reference-free QA tool, which could help operators deliver robust, efficient autosegmentation in the radiotherapy clinic.

Proton therapy for breast cancer: Reducing toxicity.

Radiotherapy is a crucial component in the holistic management of breast cancer, with approximately 60% of individuals diagnosed with breast cancer requiring this treatment. As the survival rate of individuals with breast cancer has significantly increased, there is a growing focus on the long-term well-being of patients. Proton therapy (PT) is a new and rapidly developing radiotherapy method. In comparison with conventional photon therapy, PT offers the benefits of decreased radiation toxicity and increased dosage in the designated region. This can extend patients' lifespan and enhance their overall well-being. The present analysis examines the function of PT in diminishing the harmful effects of radiation in cases of breast cancer, while also providing a brief overview of the future potential and obstacles associated with PT for breast cancer.

Inter-fraction motion robustness in a prospective phase II trial on dose-escalated proton reirradiation for locally recurrent rectal cancer.

Background and purpose: Intensity modulated proton therapy (IMPT) enables generation of conformal dose plans with organ at risk (OAR) sparing potential. However, pelvic IMPT robustness is challenged by inter-fraction motion caused by constant anatomical variations. In this study, the dosimetric impact of inter-fraction motion on target coverage and dose to OAR was quantified in the prospective phase II study ReRad-II on dose-escalated proton reirradiation for locally recurrent rectal cancer (LRRC). Materials and methods: The inter-fraction motion robustness was assessed for the initial twelve patients enrolled in the ReRad-II study. Patients with resectable LRRC were assessed for neoadjuvant IMPT (55 Gy(RBE)/44Fx) and unresectable recurrences for definitive IMPT (57.5-65 Gy(RBE)/ 46-52Fx). Target coverage and dose to OAR were assessed for robustly optimised three-field IMPT, on 12 plan computerized tomography (CT) scans (pCT) - and 47 repetitive control CT scans (cCTs) during the treatment. The target coverage and doses to OAR were re-calculated on each cCT and the mean dose ratio (pCT/cCT-ratio) and target coverage (V95%) was evaluated. Results: The target coverage was robust with a mean dose pCT/cCT-ratio of 1.00 (+/-1%). The V95% target coverage for every cCT were above the accepted worst-case scenario in the robust evaluation. Considerable variation in bladder-, bowel bag-, and bowel loop volume was observed. The OAR with the largest variation in ratio was the bladder (pCT/cCT-ratio: 1.3 (range: 0.5-4.7). Conclusions: IMPT for dose-escalated reirradiation of LRRC provided anatomically robust target coverage despite OAR changes. Inter-fraction motion resulted in OAR doses varying within clinically acceptable range.

Cardiac conduction system as a new organ at risk in radiotherapy.

BACKGROUND: Cardiovascular diseases represent the most common non-oncologic cause of death in patients following radiotherapy (RT) in the thoracic region. Radiation-induced heart disease (RIHD) can manifest as various heterogeneous clinical entities. However, the influence of RT on the cardiac conduction system has only recently gained more attention. Arrhythmogenic toxicity, i.e., conduction disorders and arrhythmias, constitutes a significant part of these adverse effects. The cardiac conduction system is not routinely monitored as an organ at risk (OaR). Its specific histological nature and function suggest different sensitivity and response to radiation. The heart is a highly heterogeneous organ, and the routinely monitored dose to the whole heart may not adequately characterize the risk of increased arrhythmogenic toxicity from RT. Cardiac structures, including the conduction system, appear to be additional OaRs for which dose distribution should be monitored. MATERIAL AND METHODS: For the systematic selection of studies, we utilized the PubMed database with keywords derived from the analysis of existing literature. The search was limited to English-language publications, and the selection criteria included relevance to the topic and the quality of methodology. PURPOSE: This article summarizes the impact of RT on the cardiac conduction system. CONCLUSION: Radiotherapy-induced cardiotoxicity significantly affects morbidity and mortality. The heart exhibits heterogeneity in terms of radiosensitivity. Certain cardiac subregions in the dose distribution show a higher correlation with poorer overall survival than routinely monitored doses to the whole heart and derived parameters (the volumes irradiated with the doses of 5 or 30 Gy - V5 or V30, respectively). The most radiosensitive subregions appear to be the base of the heart, including the beginning of the conduction system. Higher doses to the conduction system, especially the sinoatrial (SA) node, are associated with a higher incidence of a wide range of arrhythmias and poorer overall survival. However, dose limits (Dmean and Dmax) for the conduction system have not yet been established. Dosimetric studies have identified cutoff doses to the SA node, exceeding which there is a significant increase in mortality and the occurrence of arrhythmias.

Analyzing the Relationship between Dose and Geometric Agreement Metrics for Auto-Contouring in Head and Neck Normal Tissues.

This study aimed to determine the relationship between geometric and dosimetric agreement metrics in head and neck (H&N) cancer radiotherapy plans. A total 287 plans were retrospectively analyzed, comparing auto-contoured and clinically used contours using a Dice similarity coefficient (DSC), surface DSC (sDSC), and Hausdorff distance (HD). Organs-at-risk (OARs) with ≥200 cGy dose differences from the clinical contour in terms of Dmax (D0.01cc) and Dmean were further examined against proximity to the planning target volume (PTV). A secondary set of 91 plans from multiple institutions validated these findings. For 4995 contour pairs across 19 OARs, 90% had a DSC, sDSC, and HD of at least 0.75, 0.86, and less than 7.65 mm, respectively. Dosimetrically, the absolute difference between the two contour sets was <200 cGy for 95% of OARs in terms of Dmax and 96% in terms of Dmean. In total, 97% of OARs exhibiting significant dose differences between the clinically edited contour and auto-contour were within 2.5 cm PTV regardless of geometric agreement. There was an approximately linear trend between geometric agreement and identifying at least 200 cGy dose differences, with higher geometric agreement corresponding to a lower fraction of cases being identified. Analysis of the secondary dataset validated these findings. Geometric indices are approximate indicators of contour quality and identify contours exhibiting significant dosimetric discordance. For a small subset of OARs within 2.5 cm of the PTV, geometric agreement metrics can be misleading in terms of contour quality.

Radiation induced brachial plexopathy in head and neck cancer patients treated with definitive radiotherapy and correlation with disease characteristics and dosimetric parameters.

Background: Definitive concurrent chemoradiotherapy (CRT) is the standard of care in advanced stages of head and neck cancer (HNC). With evident increase in survival rate there is also simultaneous increase in toxicity affecting the quality of life. One of the less researched late toxicity is radiation induced brachial plexopathy (RIBP). In this dosimetric study we intent to contour the brachial plexus (BP) as an organ at risk (OAR) and determine the factors that contribute to dose variations to BP, and clinically evaluate the patients for RIBP during follow-up using a questionnaire. Materials and methods: 30 patients with HNC planned for CRT from September 2020 to June 2022 were accrued. Patients were treated to a dose of 6600 cGy with intensity modulated radiotherapy using the simultaneous integrated boost technique. From the dose-volume histogram (DVH) statistics the BP volume, Dmax and other parameters like V66, V60 were assessed and was correlated with respect to primary tumour and nodal stage. Results: On corelation, more than the T stage, the N stage and the primary location had a significant impact on the Dmax. With a median follow-up of 17.9 months, the incidence of RIBP was 6.67%. The 2-year disease free survival and the 2-year overall survival were 53.7% and 59.4%, respectively. Conclusions: In oropharyngeal/hypopharyngeal primaries and in advanced nodal disease, BP receives higher doses contributing to RIBP. Primary tumor and nodal stage also impacted V60 and V66 of BP. Hence, contouring of BP as an OAR becomes imperative, and respecting the DVH parameters is essential.

Effect of model-based dose-calculation algorithms in high dose rate brachytherapy of cervical carcinoma.

Background: Task Group 43 (TG-43) formalism does not consider the tissue and applicator heterogeneities. This study is to compare the effect of model-based dose calculation algorithms, like Advanced Collapsed Cone Engine (ACE), on dose calculation with the TG-43 dose calculation formalism in patients with cervical carcinoma. Materials and methods: 20 patients of cervical carcinoma treated with a high dose rate of intracavitary brachytherapy were prospectively studied. The target volume and organs at risk (OARs) were contoured in the Oncentra treatment planning system (Elekta, Veenendaal, The Netherlands). All patients were planned with cobalt-60 (Co-60) and iridium-192 (Ir-192) sources with doses of 21 Gy in 3 fractions. These plans were calculated with TG-43 formalism and a model-based dose calculation algorithm ACE. The dosimetric parameters of TG-43 and ACE-based plans were compared in terms of target coverage and OAR doses. Results: For Co-60-based plans, the percentage differences in the D90 and V100 values for high-risk clinical target volume (HR-CTV) were 0.36 ± 0.43% and 0.17 ± 0.31%, respectively. For the bladder, rectum and sigmoid, the percentage differences for D2cc volumes were -0.50 ± 0.51%, -0.16 ± 0.53% and -0.37 ± 1.21%, respectively. For Ir-192-based plans, the percentage difference in the D90 for HR-CTV was 0.54 ± 0.79%, while V100 was 0.24 ± 0.29%. For the bladder, rectum and sigmoid, the doses to 2cc volume were 0.35 ± 1.06%, 0.99 ± 0.74% and 0.74 ± 1.92%, respectively. No significant differences were found in the dosimetric parameters calculated with ACE and TG-43. Conclusion: The ACE algorithm reduced doses to OARs and targets. However, ACE and TG-43 did not show significant differences in the dosimetric parameters of the target and OARs with both sources.

Dosimetric evaluation of intensity modulated photon versus proton reirradiation of head and neck cancer.

BACKGROUND: Reirradiation of head and neck cancer (HNC) became more accessible in the last decade, owing to modern irradiation techniques which offer a reduction in treatment related toxicities. The aim of this paper was to comparatively evaluate the dosimetric aspects derived from intensity modulated photon vs. proton treatment planning in reirradiated HNC patients. METHODS: Six recurrent HNC patients were enrolled in this retrospective study. For each patient two treatment plans were created: one IMRT/VMAT and one IMPT plan. The prescribed dose for the second irradiation was between 50 and 70 Gy RBE. The study comparatively analyzed the CTV coverage, doses to organs at risk (OARs) and low doses received by the healthy tissue (other than OAR). RESULTS: Similar CTV coverage was achieved for photon vs proton plans, with the latter presenting better homogeneity in four cases. Maximum dose to CTV was generally higher for photon plans, with differences ranging from 0.3 to 1.9%. For parotid glands and body, the mean dose was lower for proton plans. A notable reduction of low dose to healthy tissue (other than OARs) could be achieved with protons, with an average of 60% and 64% for D10% and Dmean, respectively. CONCLUSION: The dosimetric comparison between photon and proton reirradiation of HNC showed a great need for treatment individualization, concluding that protons should be considered for reirradiation on an individual basis.