Mean Dice Similarity Coefficient Research Articles

Comparison of commercial atlas-based automatic segmentation software for prostate radiotherapy treatment planning.

This study aims to assess the accuracy of automatic atlas-based contours for various key anatomical structures in prostate radiotherapy treatment planning. The evaluated structures include the bladder, rectum, prostate, seminal vesicles, femoral heads and penile bulb. CT images from 20 patients who underwent intensity-modulated radiotherapy were randomly chosen to create an atlas library. Atlas contours of the seven anatomical structures were generated using four software packages: ABAS, Eclipse, MIM, and RayStation. These contours were then compared to manual delineations performed by oncologists, which served as the ground truth. Evaluation metrics such as dice similarity coefficient (DSC), mean distance to agreement (MDA), and volume ratio (VR) were calculated to assess the accuracy of the contours. Additionally, the time taken by each software to generate the atlas contour was recorded. The mean DSC values for the bladder exhibited strong agreement (>0.8) with manual delineations for all software except for Eclipse and RayStation. Similarly, the femoral heads showed significant similarity between the atlas contours and ground truth across all software, with mean DSC values exceeding 0.9 and MDA values close to zero. On the other hand, the penile bulb displayed only moderate agreement with the ground truth, with mean DSC values ranging from 0.5 to 0.7 for all software. A similar trend was observed in the prostate atlas contours, except for MIM, which achieved a mean DSC of over 0.8. For the rectum, both ABAS and MIM atlases demonstrated strong agreement with the ground truth, resulting in mean DSC values of more than 0.8. Overall, MIM and ABAS outperformed Eclipse and RayStation in both DSC and MDA. These results indicate that the atlas-based segmentation employed in this study produces acceptable contours for the anatomical structures of interest in prostate radiotherapy treatment planning.

Improving hybrid image and structure-based deformable image registration for large internal deformations

Objective. Deformable image registration (DIR) is a widely used technique in radiotherapy. Complex deformations, resulting from large anatomical changes, are a regular challenge. DIR algorithms generally seek a balance between capturing large deformations and preserving a smooth deformation vector field (DVF). We propose a novel structure-based term that can enhance the registration efficacy while ensuring a smooth DVF. Approach. The proposed novel similarity metric for controlling structures was introduced as a new term into a commercially available algorithm. Its performance was compared to the original algorithm using a dataset of 46 patients who received pelvic re-irradiation, many of which exhibited complex deformations. Main results. The mean Dice Similarity Coefficient (DSC) under the improved algorithm was 0.96, 0.94, 0.76, and 0.91 for bladder, rectum, colon, and bone respectively, compared to 0.69, 0.89, 0.62, and 0.88 for the original algorithm. The improvement was more pronounced for complex deformations. Significance. With this work, we have demonstrated that the proposed term is able to improve registration accuracy for complex cases while maintaining realistic deformations.

Is 3T MR nerve-bone fusion imaging a viable alternative to MRI-CBCT to identify the relationship between the inferior alveolar nerve and mandibular third molar.

To investigate the feasibility of MRI nerve-bone fusion imaging in assessing the relationship between inferior alveolar nerve (IAN) / mandibular canal (MC) and mandibular third molar (MTM) compared with MRI-CBCT fusion. The MRI nerve-bone fusion and MRI-CBCT fusion imaging were performed in 20 subjects with 37 MTMs. The Hausdorff distance (HD) value and dice similarity coefficient (DSC) was calculated. The relationship between IAN/MC and MTM roots, inflammatory, and fusion patterns were compared between these two fused images. The reliability was assessed using a weighted κ statistic. The mean HD and DSC ranged from 0.62~1.35 and 0.83~0.88 for MRI nerve-bone fusion, 0.98 ~ 1.50 and 0.76 ~ 0.83 for MRI-CBCT fusion. MR nerve-bone fusion had considerable reproducibility compared to MRI-CBCT fusion in relation classification (MR nerve-bone fusion κ = 0.694, MRI-CBCT fusion κ = 0.644), direct contact (MR nerve-bone fusion κ = 0.729, MRI-CBCT fusion κ = 0.720), and moderate to good agreement for inflammation detection (MR nerve-bone fusion κ = 0.603, MRI-CBCT fusion κ = 0.532, average). The MR nerve-bone fusion imaging showed a lower ratio of larger pattern compared to MR-CBCT fusion (16.2% VS 27.3% in the molar region, and 2.7% VS 5.4% in the retromolar region). And the average time spent on MR nerve-bone fusion and MRI-CBCT fusion was 1min and 3min, respectively. Both MR nerve-bone fusion and MRI-CBCT fusion exhibited good consistency in evaluating the spatial relationship between IAN/MC and MTM, fusion effect, and inflammation detection. MR nerve-bone fusion imaging can be a preoperative one-stop radiation-free examination for patients at high risk for MTM surgery.

Automatic segmentation of hemorrhagic transformation on follow-up non-contrast CT after acute ischemic stroke.

Hemorrhagic transformation (HT) following reperfusion therapies is a serious complication for patients with acute ischemic stroke. Segmentation and quantification of hemorrhage provides critical insights into patients' condition and aids in prognosis. This study aims to automatically segment hemorrhagic regions on follow-up non-contrast head CT (NCCT) for stroke patients treated with endovascular thrombectomy (EVT). Patient data were collected from 10 stroke centers across two countries. We propose a semi-automated approach with adaptive thresholding methods, eliminating the need for extensive training data and reducing computational demands. We used Dice Similarity Coefficient (DSC) and Lin's Concordance Correlation Coefficient (Lin's CCC) to evaluate the performance of the algorithm. A total of 51 patients were included, with 28 Type 2 hemorrhagic infarction (HI2) cases and 23 parenchymal hematoma (PH) cases. The algorithm achieved a mean DSC of 0.66 ± 0.17. Notably, performance was superior for PH cases (mean DSC of 0.73 ± 0.14) compared to HI2 cases (mean DSC of 0.61 ± 0.18). Lin's CCC was 0.88 (95% CI 0.79-0.93), indicating a strong agreement between the algorithm's results and the ground truth. In addition, the algorithm demonstrated excellent processing time, with an average of 2.7 s for each patient case. To our knowledge, this is the first study to perform automated segmentation of post-treatment hemorrhage for acute stroke patients and evaluate the performance based on the radiological severity of HT. This rapid and effective tool has the potential to assist with predicting prognosis in stroke patients with HT after EVT.

3D auto-segmentation of biliary structure of living liver donors using magnetic resonance cholangiopancreatography for enhanced preoperative planning.

This study aimed to develop an automated segmentation system for biliary structures using a deep learning model, based on data from magnetic resonance cholangiopancreatography (MRCP). Living liver donors who underwent MRCP using the gradient and spin echo technique followed by three-dimensional modeling were eligible for this study. A three-dimensional residual U-Net model was implemented for the deep learning process. Data were divided into training and test sets at a 9:1 ratio. Performance was assessed using the dice similarity coefficient to compare the model's segmentation with the manually labeled ground truth. The study incorporated 250 cases. There was no difference in the baseline characteristics between the train set (n=225) and test set (n=25). The overall mean Dice Similarity Coefficient was 0.80±0.20 between the ground truth and inference result. The qualitative assessment of the model showed relatively high accuracy especially for the common bile duct (88%), common hepatic duct (92%), hilum (96%), right hepatic duct (100%), and left hepatic duct (96%), while the third-order branch of the right hepatic duct (18.2%) showed low accuracy. The developed automated segmentation model for biliary structures, utilizing MRCP data and deep learning techniques, demonstrated robust performance and holds potential for further advancements in automation.

A deep learning model for translating CT to ventilation imaging: analysis of accuracy and impact on functional avoidance radiotherapy planning.

Radiotherapy planning incorporating functional lung images has the potential to reduce pulmonary toxicity. Free-breathing 4DCT-derived ventilation image (CTVI) may help quantify lung function. This study introduces a novel deep-learning model directly translating planning CT images into CTVI. We investigated the accuracy of generated images and the impact on functional avoidance planning. Paired planning CT and 4DCT scans from 48 patients with NSCLC were randomized to training (n = 41) and testing (n = 7) data sets. The ventilation maps were generated from 4DCT using a Jacobian-based algorithm to provide ground truth labels (CTVI4DCT). A 3D U-Net-based model was trained to map CT to synthetic CTVI (CTVISyn) and validated using fivefold cross-validation. The highest-performing model was applied to the testing set. Spearman's correlation (rs) and Dice similarity coefficient (DSC) determined voxel-wise and functional-wise concordance between CTVI4DCT and CTVISyn. Three plans were designed per patient in the testing set: one clinical plan without CTVI and two functional avoidance plans combined with CTVI4DCT or CTVISyn, aimed at sparing high-functional lungs defined as the top 50% of the percentile ventilation ranges. Dose-volume (DVH) parameters regarding the planning target volume (PTV) and organs at risk (OARs) were recorded. Radiation pneumonitis (RP) risk was estimated using a dose-function (DFH)-based normal tissue complication probability (NTCP) model. CTVISyn showed a mean rs value of 0.65 ± 0.04 compared to CTVI4DCT. Mean DSC values over the top 50% and 60% of ventilation ranges were 0.41 ± 0.07 and 0.52 ± 0.10, respectively. In the test set (n = 7), all patients' RP-risk benefited from CTVI4DCT-guided plans (Riskmean_4DCT_vs_Clinical: 29.24% vs. 49.12%, P = 0.016), and six patients benefited from CTVISyn-guided plans (Riskmean_Syn_vs_Clinical: 31.13% vs. 49.12%, P = 0.022). There were no significant differences in DVH and DFH metrics between CTVISyn and CTVI4DCT-guided plan (P > 0.05). Using deep-learning techniques, CTVISyn generated from planning CT exhibited a moderate-to-high correlation with CTVI4DCT. The CTVISyn-guided plans were comparable to the CTVI4DCT-guided plans, effectively reducing pulmonary toxicity in patients while maintaining acceptable plan quality. Further prospective trials are needed to validate these findings.

Quantitative measurement of the ureter on three-dimensional magnetic resonance urography images using deep learning.

Accurate measurement of ureteral diameters plays a pivotal role in diagnosing and monitoring urinary tract obstruction (UTO). While three-dimensional magnetic resonance urography (3D MRU) represents a significant advancement in imaging, the traditional manual methods for assessing ureteral diameters are characterized by labor-intensive procedures and inherent variability. In the realm of medical image analysis, deep learning has led to a paradigm shift, yet the development of a comprehensive automated tool for the precise segmentation and measurement of ureters in MR images is an unaddressed challenge. The ureter was quantitatively measured on 3D MRU images using a deep learning model. A retrospective cohort of 445 3D MRU scans (443 patients, 52±18 years; 217 female patients) was collected and split into training, validation, and internal testing cohorts. A 3D V-Net model was trained for urinary tract segmentation, and a post-processing algorithm was developed for ureteral measurements. The accuracy of the segmentation was evaluated using the Dice similarity coefficient (DSC) and volume intraclass correlation coefficient (ICC), with ground truth segmentations provided by experienced radiologists. The external cohort comprised 50 scans (50 patients, 55±21 years; 30 female patients), and the model-predicted ureteral diameter measurements were compared with manual measurements to assess system performance. The various diameter parameters of ureter among the different measurement methods (ground truth, auto-segmentation with automatic diameter extraction, and manual segmentation with automatic diameter extraction) were assessed with Friedman tests and post hoc Dunn test. The effectiveness of the UTO diagnosis was assessed by receiver operating characteristic (ROC) curves and their respective areas under the curve (AUC) between different methods. In both the internal test and external cohorts, the mean DSC values for bilateral ureters exceeded 0.70. The ICCs for the bilateral ureter volume obtained by comparing the model and manual segmentation were all greater than 0.96 (p < 0.05), except for the right ureter in the internal test cohort, for which the ICC was 0.773 (p < 0.05). The mean DSCs for interobserver and intraobserver reliability were all above 0.97. The maximum diameter of the ureter exhibited no statistically significant differences either in the dilated (p=0.08) or in the non-dilated (p=0.32) ureters across the three measurement methods. The AUCs of ground truth, auto-segmentation with automatic diameter extraction, and manual segmentation with automatic diameter extraction in diagnosing UTO were 0.988 (95% CI: 0.934, 1.000), 0.961 (95% CI: 0.893, 0.991), and 0.979 (95% CI: 0.919, 0.998), respectively. There was no statistical difference between AUCs of the different methods (p>0.05). The proposed deep learning model and post-processing algorithm provide an effective means for the quantitative evaluation of urinary diseases using 3D MRU images.

Dual-branch feature fusion S3D V-Net network for lung nodules segmentation.

Accurate segmentation of lung nodules can help doctors get more accurate results and protocols in early lung cancer diagnosis and treatment planning, so that patients can be better detected and treated at an early stage, and the mortality rate of lung cancer can be reduced. Currently, the improvement of lung nodule segmentation accuracy has been limited by his heterogeneous performance in the lungs, the imbalance between segmentation targets and background pixels, and other factors. We propose a new 2.5D lung nodule segmentation network model for lung nodule segmentation. This network model can well improve the extraction of edge information of lung nodules, and fuses intra-slice and inter-slice features, which makes good use of the three-dimensional structural information of lung nodules and can more effectively improve the accuracy of lung nodule segmentation. Our approach is based on a typical encoding-decoding network structure for improvement. The improved model captures the features of multiple nodules in both 3-D and 2-D CT images, complements the information of the segmentation target's features and enhances the texture features at the edges of the pulmonary nodules through the dual-branch feature fusion module (DFFM) and the reverse attention context module (RACM), and employs central pooling instead of the maximal pooling operation, which is used to preserve the features around the target and to eliminate the edge-irrelevant features, to further improve the performance of the segmentation of the pulmonary nodules. We evaluated this method on a wide range of 1186 nodules from the LUNA16 dataset, and averaging the results of ten cross-validated, the proposed method achieved the mean dice similarity coefficient (mDSC) of 84.57%, the mean overlapping error (mOE) of 18.73% and average processing of a case is about 2.07 s. Moreover, our results were compared with inter-radiologist agreement on the LUNA16 dataset, and the average difference was 0.74%. The experimental results show that our method improves the accuracy of pulmonary nodules segmentation and also takes less time than more 3-D segmentation methods in terms of time.

Automatic segmentation of hepatocellular carcinoma on dynamic contrast-enhanced MRI based on deep learning

Objective. Precise hepatocellular carcinoma (HCC) detection is crucial for clinical management. While studies focus on computed tomography-based automatic algorithms, there is a rareness of research on automatic detection based on dynamic contrast enhanced (DCE) magnetic resonance imaging. This study is to develop an automatic detection and segmentation deep learning model for HCC using DCE. Approach: DCE images acquired from 2016 to 2021 were retrospectively collected. Then, 382 patients (301 male; 81 female) with 466 lesions pathologically confirmed were included and divided into an 80% training-validation set and a 20% independent test set. For external validation, 51 patients (42 male; 9 female) in another hospital from 2018 to 2021 were included. The U-net architecture was modified to accommodate multi-phasic DCE input. The model was trained with the training-validation set using five-fold cross-validation, and furtherly evaluated with the independent test set using comprehensive metrics for segmentation and detection performance. The proposed automatic segmentation model consisted of five main steps: phase registration, automatic liver region extraction using a pre-trained model, automatic HCC lesion segmentation using the multi-phasic deep learning model, ensemble of five-fold predictions, and post-processing using connected component analysis to enhance the performance to refine predictions and eliminate false positives. Main results. The proposed model achieved a mean dice similarity coefficient (DSC) of 0.81 ± 0.11, a sensitivity of 94.41 ± 15.50%, a precision of 94.19 ± 17.32%, and 0.14 ± 0.48 false positive lesions per patient in the independent test set. The model detected 88% (80/91) HCC lesions in the condition of DSC > 0.5, and the DSC per tumor was 0.80 ± 0.13. In the external set, the model detected 92% (58/62) lesions with 0.12 ± 0.33 false positives per patient, and the DSC per tumor was 0.75 ± 0.10. Significance. This study developed an automatic detection and segmentation deep learning model for HCC using DCE, which yielded promising post-processed results in accurately identifying and delineating HCC lesions.

Fully Automated Identification of Lymph Node Metastases and Lymphovascular Invasion in Endometrial Cancer From Multi-Parametric MRI by Deep Learning.

Early and accurate identification of lymphatic node metastasis (LNM) and lymphatic vascular space invasion (LVSI) for endometrial cancer (EC) patients is important for treatment design, but difficult on multi-parametric MRI (mpMRI) images. To develop a deep learning (DL) model to simultaneously identify of LNM and LVSI of EC from mpMRI images. Retrospective. Six hundred twenty-one patients with histologically proven EC from two institutions, including 111 LNM-positive and 168 LVSI-positive, divided into training, internal, and external test cohorts of 398, 169, and 54 patients, respectively. T2-weighted imaging (T2WI), contrast-enhanced T1WI (CE-T1WI), and diffusion-weighted imaging (DWI) were scanned with turbo spin-echo, gradient-echo, and two-dimensional echo-planar sequences, using either a 1.5 T or 3 T system. EC lesions were manually delineated on T2WI by two radiologists and used to train an nnU-Net model for automatic segmentation. A multi-task DL model was developed to simultaneously identify LNM and LVSI positive status using the segmented EC lesion regions and T2WI, CE-T1WI, and DWI images as inputs. The performance of the model for LNM-positive diagnosis was compared with those of three radiologists in the external test cohort. Dice similarity coefficient (DSC) was used to evaluate segmentation results. Receiver Operating Characteristic (ROC) analysis was used to assess the performance of LNM and LVSI status identification. P value <0.05 was considered significant. EC lesion segmentation model achieved mean DSC values of 0.700 ± 0.25 and 0.693 ± 0.21 in the internal and external test cohorts, respectively. For LNM positive/LVSI positive identification, the proposed model achieved AUC values of 0.895/0.848, 0.806/0.795, and 0.804/0.728 in the training, internal, and external test cohorts, respectively, and better than those of three radiologists (AUC = 0.770/0.648/0.674). The proposed model has potential to help clinicians to identify LNM and LVSI status of EC patients and improve treatment planning. 3 TECHNICAL EFFICACY: Stage 2.

Auto-segmentation of neck nodal metastases using self-distilled masked image transformer on longitudinal MR images.

Auto-segmentation promises greater speed and lower inter-reader variability than manual segmentations in radiation oncology clinical practice. This study aims to implement and evaluate the accuracy of the auto-segmentation algorithm, "Masked Image modeling using the vision Transformers (SMIT)," for neck nodal metastases on longitudinal T2-weighted (T2w) MR images in oropharyngeal squamous cell carcinoma (OPSCC) patients. This prospective clinical trial study included 123 human papillomaviruses (HPV-positive [+]) related OSPCC patients who received concurrent chemoradiotherapy. T2w MR images were acquired on 3 T at pre-treatment (Tx), week 0, and intra-Tx weeks (1-3). Manual delineations of metastatic neck nodes from 123 OPSCC patients were used for the SMIT auto-segmentation, and total tumor volumes were calculated. Standard statistical analyses compared contour volumes from SMIT vs manual segmentation (Wilcoxon signed-rank test [WSRT]), and Spearman's rank correlation coefficients (ρ) were computed. Segmentation accuracy was evaluated on the test data set using the dice similarity coefficient (DSC) metric value. P-values <0.05 were considered significant. No significant difference in manual and SMIT delineated tumor volume at pre-Tx (8.68 ± 7.15 vs 8.38 ± 7.01 cm3, P = 0.26 [WSRT]), and the Bland-Altman method established the limits of agreement as -1.71 to 2.31 cm3, with a mean difference of 0.30 cm3. SMIT model and manually delineated tumor volume estimates were highly correlated (ρ = 0.84-0.96, P < 0.001). The mean DSC metric values were 0.86, 0.85, 0.77, and 0.79 at the pre-Tx and intra-Tx weeks (1-3), respectively. The SMIT algorithm provides sufficient segmentation accuracy for oncological applications in HPV+ OPSCC. First evaluation of auto-segmentation with SMIT using longitudinal T2w MRI in HPV+ OPSCC.

Localization and Risk Stratification of Thyroid Nodules in Ultrasound Images Through Deep Learning

ObjectiveDeep learning algorithms have commonly been used for the differential diagnosis between benign and malignant thyroid nodules. In this study, we aimed to develop an integrated system that combines deep learning model and clinical standard thyroid imaging reporting and data system (TI-RADS) for the simultaneous segmentation and risk stratification of thyroid nodules. Methods304 ultrasound images from two independent sites with TI-RADS 4 thyroid nodules were collected. Edge-connection and Criminisi algorithm was used to remove manually-induced markers in ultrasound images. An integrated system based on TI-RADS and mask region-based convolution neural network (Mask R-CNN) was proposed to stratify subclasses of TI-RADS 4 thyroid nodules and to segment thyroid nodules in the ultrasound images. Accuracy and precision-recall curve were used to evaluate stratification performance, dice similarity coefficient (DSC) between the segmentation of Mask R-CNN and the radiologist's contour was used to evaluate segmentation performance of the model. ResultsThe combined approach could significantly enhance the performance of the proposed integrated system. Overall stratification accuracy of TI-RADS 4 thyroid nodules, mean average precision, and mean DSC of the proposed model in the independent test set were 90.79%, 0.8579, and 0.83, respectively. Specifically, stratification accuracy for TI-RADS 4a, 4b and 4c thyroid nodules were 95.83%, 84.21%, and 77.78%. ConclusionAn integrated system combining TI-RADS and deep learning model was developed. The system can provide clinicians with not only diagnostic assistance from TI-RADS, but also accurate segmentation of thyroid nodules, which improves the applicability of the system in clinical practice.

Sliding transformer with uncertainty estimation for vestibular schwannoma automatic segmentation.

Objective.Automated segmentation of vestibular schwannoma (VS) using magnetic resonance imaging (MRI) can enhance clinical efficiency. Though many advanced methods exist for automated VS segmentation, the accuracy is hindered by ambivalent tumor borders and cystic regions in some patients. In addition, these methods provide results that do not indicate segmentation uncertainty, making their translation into clinical workflows difficult due to potential errors. Providing a definitive segmentation result along with segmentation uncertainty or self-confidence is crucial for the conversion of automated segmentation programs to clinical aid diagnostic tools.Approach.To address these issues, we propose a U-shaped cascade transformer structure with a sliding window that utilizes multiple sliding samples, a segmentation head, and an uncertainty head to obtain both the segmentation mask and uncertainty map. We collected multimodal MRI data from 60 clinical patients with VS from Xuanwu Hospital. Each patient case includes T1-weighted images, contrast-enhanced T1-weighted images, T2-weighted images, and a tumor mask. The images exhibit an in-plane resolution ranging from 0.70 × 0.70 to 0.76 × 0.76 mm, an in-plane matrix spanning from 216 × 256 to 284 × 256, a slice thickness varying between 0.50 and 0.80 mm, and a range of slice numbers from 72 to 120.Main results.Extensive experimental results show that our method achieves comparable or higher results than previous state-of-the-art brain tumor segmentation methods. On our collected multimodal MRI dataset of clinical VS, our method achieved the dice similarity coefficient (DSC) of 96.08% ± 1.30. On a publicly available VS dataset, our method achieved the mean DSC of 94.23% ± 2.53.Significance.The method efficiently solves the VS segmentation task while providing an uncertainty map of the segmentation results, which helps clinical experts review the segmentation results more efficiently and helps to transform the automated segmentation program into a clinical aid diagnostic tool.

Fully-automated multi-organ segmentation tool applicable to both non-contrast and post-contrast abdominal CT: deep learning algorithm developed using dual-energy CT images

A novel 3D nnU-Net-based of algorithm was developed for fully-automated multi-organ segmentation in abdominal CT, applicable to both non-contrast and post-contrast images. The algorithm was trained using dual-energy CT (DECT)-obtained portal venous phase (PVP) and spatiotemporally-matched virtual non-contrast images, and tested using a single-energy (SE) CT dataset comprising PVP and true non-contrast (TNC) images. The algorithm showed robust accuracy in segmenting the liver, spleen, right kidney (RK), and left kidney (LK), with mean dice similarity coefficients (DSCs) exceeding 0.94 for each organ, regardless of contrast enhancement. However, pancreas segmentation demonstrated slightly lower performance with mean DSCs of around 0.8. In organ volume estimation, the algorithm demonstrated excellent agreement with ground-truth measurements for the liver, spleen, RK, and LK (intraclass correlation coefficients [ICCs] > 0.95); while the pancreas showed good agreements (ICC = 0.792 in SE-PVP, 0.840 in TNC). Accurate volume estimation within a 10% deviation from ground-truth was achieved in over 90% of cases involving the liver, spleen, RK, and LK. These findings indicate the efficacy of our 3D nnU-Net-based algorithm, developed using DECT images, which provides precise segmentation of the liver, spleen, and RK and LK in both non-contrast and post-contrast CT images, enabling reliable organ volumetry, albeit with relatively reduced performance for the pancreas.

Development of statistical auto-segmentation method for diffusion restriction gray matter lesions in patients with newly diagnosed sporadic Creutzfeldt–Jakob disease

Quantification of diffusion restriction lesions in sporadic Creutzfeldt-Jakob disease (sCJD) may provide information of the disease burden. We aim to develop an automatic segmentation model for sCJD and to evaluate the volume of disease extent as a prognostic marker for overall survival. Fifty-six patients (mean age ± SD, 61.2 ± 9.9 years) were included from February 2000 to July 2020. A threshold-based segmentation was used to obtain abnormal signal intensity masks. Segmented volumes were compared with the visual grade. The Dice similarity coefficient was calculated to measure the similarity between the automatic vs. manual segmentation. Cox proportional hazards regression analysis was performed to evaluate the volume of disease extent as a prognostic marker. The automatic segmentation showed good correlation with the visual grading. The cortical lesion volumes significantly increased as the visual grade aggravated (extensive: 112.9 ± 73.2; moderate: 45.4 ± 30.4; minimal involvement: 29.6 ± 18.1 mm3) (P < 0.001). The deep gray matter lesion volumes were significantly higher for positive than for negative involvement of the deep gray matter (5.6 ± 4.6 mm3 vs. 1.0 ± 1.3 mm3, P < 0.001). The mean Dice similarity coefficients were 0.90 and 0.94 for cortical and deep gray matter lesions, respectively. However, the volume of disease extent was not associated with worse overall survival (cortical extent: P = 0.07; deep gray matter extent: P = 0.12).

Automatic segmentation of hemispheric CSF on MRI using deep learning: Quantifying cerebral edema following large hemispheric infarction

Background and objectiveCerebral edema (CED) is a serious complication of acute ischemic stroke (AIS), especially in patients with large hemispheric infarction (LHI). Herein, a deep learning-based approach is implemented to extract CSF from T2-Weighted Imaging (T2WI) and evaluate the relationship between quantified cerebrospinal fluid and outcomes. MethodsPatients with acute LHI who underwent magnetic resonance imaging (MRI) were included. We used a deep learning algorithm to segment the CSF from T2WI. The hemispheric CSF ratio was calculated to evaluate its relationship with the degree of brain edema and prognosis in patients with LHI. ResultsFor the 93 included patients, the left and right cerebrospinal fluid regions were automatically extracted with a mean Dice similarity coefficient of 0.830. Receiver operating characteristic analysis indicated that hemispheric CSF ratio was an accurate marker for qualitative severe cerebral edema (area under receiver-operating-characteristic curve 0.867 [95% CI, 0.781–0.929]). Multivariate logistic regression analysis of functional prognosis showed that previous stroke (OR = 5.229, 95% CI 1.013–26.984), ASPECT≤6 (OR = 13.208, 95% CI 1.136–153.540) and low hemispheric CSF ratio (OR = 0.966, 95% CI 0.937–0.997) were significantly associated with higher chances for unfavorable functional outcome in patients with LHI. ConclusionsAutomated assessment of CSF volume provides an objective biomarker of cerebral edema that can be leveraged to quantify the degree of cerebral edema and confirm its predictive effect on outcomes after LHI.

Technical note: Minimizing CIED artifacts on a 0.35 T MRI-Linac using deep learning.

Artifacts from implantable cardioverter defibrillators (ICDs) are a challenge to magnetic resonance imaging (MRI)-guided radiotherapy (MRgRT). This study tested an unsupervised generative adversarial network to mitigate ICD artifacts in balanced steady-state free precession (bSSFP) cine MRIs and improve image quality and tracking performance for MRgRT. Fourteen healthy volunteers (Group A) were scanned on a 0.35 T MRI-Linac with and without an MR conditional ICD taped to their left pectoral to simulate an implanted ICD. bSSFP MRI data from 12 of the volunteers were used to train a CycleGAN model to reduce ICD artifacts. The data from the remaining two volunteers were used for testing. In addition, the dataset was reorganized three times using a Leave-One-Out scheme. Tracking metrics [Dice similarity coefficient (DSC), target registration error (TRE), and 95 percentile Hausdorff distance (95% HD)] were evaluated for whole-heart contours. Image quality metrics [normalized root mean square error (nRMSE), peak signal-to-noise ratio (PSNR), and multiscale structural similarity (MS-SSIM) scores] were evaluated. The technique was also tested qualitatively on three additional ICD datasets (Group B) including a patient with an implanted ICD. For the whole-heart contour with CycleGAN reconstruction: 1) Mean DSC rose from 0.910 to 0.935; 2) Mean TRE dropped from 4.488 to 2.877mm; and 3) Mean 95% HD dropped from 10.236 to 7.700mm. For the whole-body slice with CycleGAN reconstruction: 1) Mean nRMSE dropped from 0.644 to 0.420; 2) Mean MS-SSIM rose from 0.779 to 0.819; and 3) Mean PSNR rose from 18.744 to 22.368. The three Group B datasets evaluated qualitatively displayed a reduction in ICD artifacts in the heart. CycleGAN-generated reconstructions significantly improved both tracking and image quality metrics when used to mitigate artifacts from ICDs.

Development and Validation of Deep Learning-Based Automated Detection of Cervical Lymphadenopathy in Patients with Lymphoma for Treatment Response Assessment: A Bi-institutional Feasibility Study.

The purpose is to train and evaluate a deep learning (DL) model for the accurate detection and segmentation of abnormal cervical lymph nodes (LN) on head and neck contrast-enhanced CT scans in patients diagnosed with lymphoma and evaluate the clinical utility of the DL model in response assessment. This retrospective study included patients who underwent CT for abnormal cervical LN and lymphoma assessment between January 2021 and July 2022. Patients were grouped into the development (n = 76), internal test 1 (n = 27), internal test 2 (n = 87), and external test (n = 26) cohorts. A 3D SegResNet model was used to train the CT images. The volume change rates of cervical LN across longitudinal CT scans were compared among patients with different treatment outcomes (stable, response, and progression). Dice similarity coefficient (DSC) and the Bland-Altman plot were used to assess the model's segmentation performance and reliability, respectively. No significant differences in baseline clinical characteristics were found across cohorts (age, P = 0.55; sex, P = 0.13; diagnoses, P = 0.06). The mean DSC was 0.39 ± 0.2 with a precision and recall of 60.9% and 57.0%, respectively. Most LN volumes were within the limits of agreement on the Bland-Altman plot. The volume change rates among the three groups differed significantly (progression (n = 74), 342.2%; response (n = 8), - 79.2%; stable (n = 5), - 8.1%; all P < 0.01). Our proposed DL segmentation model showed modest performance in quantifying the cervical LN burden on CT in patients with lymphoma. Longitudinal changes in cervical LN volume, as predicted by the DL model, were useful for treatment response assessment.

Custom-Trained Deep Learning-Based Auto-Segmentation for Male Pelvic Iterative CBCT on C-Arm Linear Accelerators

Purpose: To evaluate the clinical applicability of a commercial artificial intelligence (AI)-driven deep learning auto-segmentation (DLAS) tool on enhanced iterative cone-beam CT (iCBCT) acquisitions for intact prostate and prostate bed treatments. Methods and MaterialsDLAS models were trained using 116 iCBCT datasets with manually delineated organs-at-risk (OARs – bladder, femoral heads, and rectum) and target volumes (intact prostate and prostate bed) adhering to institution-specific contouring guidelines. An additional 25 intact prostate and prostate bed iCBCT datasets were utilized for model testing. Segmentation accuracy relative to a reference structure set was quantified using various geometric comparison metrics and qualitatively evaluated by trained physicists and physicians. These results were compared to those obtained for an additional DLAS-based model trained on planning CT (pCT) datasets and for a deformable image registration (DIR)-based automatic contour propagation method. Results: In most instances, statistically significant differences in the Dice similarity coefficient (DSC), 95% directed Hausdorff distance, and mean surface distance metrics were observed between the models, as the iCBCT-trained DLAS model outperformed the pCT-trained DLAS model and DIR-based method for all OARs and the intact prostate target volume. Mean DSC values for the proposed method were ≥0.90 for these volumes-of-interest. The iCBCT-trained DLAS model demonstrated a relatively suboptimal performance for the prostate bed segmentation, as the mean DSC value was <0.75 for this target contour. Overall, 90% of bladder, 93% of femoral head, 67% of rectum, and 92% of intact prostate contours generated by the proposed method were deemed clinically acceptable based on qualitative scoring, while approximately 63% of prostate bed contours required moderate or major manual editing to adhere to institutional contouring guidelines. Conclusion: The proposed method presents the potential for improved segmentation accuracy and efficiency when compared to the DIR-based automatic contour propagation method as commonly applied in CBCT-based dose evaluation and calculation studies.

Abstract TMP84: Multi-Institutional Validation of an Automated Segmentation Tool for Intracerebral Hemorrhage (ICH) and Peri-Hematomal Edema (PHE) on Head CT

Objective: In patients with acute intracerebral hemorrhage (ICH), the volume of hemorrhage, and perihematomal edema (PHE) are representative of primary, and secondary brain injury, respectively. Automated quantification of ICH and PHE volumes from admission non-contrast head CT can facilitate evaluation of large stroke datasets, and expedite treatment triage when volume cutoffs are applied. We aimed to train and externally validate an automated model for segmentation and quantification of ICH and PHE volumes on non-contrast head CT scans. Methods: For training of the model, we used the data from multicentric ATACH-2 clinical trial with head CTs from eleven medical institutes in six countries. The model was then exported and tested on two external datasets from Yale and University of Berlin. We designed an automated pipeline for extracting brain window from 3D non-contrast head CT, skull stripping, resampling to homogenous voxel size, and segmentation. For segmentation we applied a deep learning model based on 3D full resolution nnUNet. We used the Dice Similarity Coefficient (DSC), Hausdorff Distance (HD), and Volume Similarity (VS) between automated segmentation and ground truth manual segmentation volumes to evaluate the models’ performance. Results: A total of 854 patients from the ATACH-2 trial (854 х2, baseline and follow-up CT scans) were used for training of the model in 5-fold cross-validation. In validation folds, the model achieved a mean DSC=0.90±0.10, HD =1.42±16.0mm, and VS=0.95±0.10 for ICH; and DSC=0.74±0.12, HD =3.0±16.5 mm, and VS=0.91±0.12 for PHE. In the externaltesting cohort from Yale (200 patients х2, baseline and follow-up CT scans), the model archieved median DSC=0.93, HD=0.98 mm, and VS=0.97 for ICH; and median DSC=0.69, HD=4.52 mm, and VS=0.84 for PHE. In the externaltesting cohort from Charité Hospital, University of Berlin (915 baseline CT scans), the model archived median DSC=0.87, HD=4 mm, and VS=0.90 for ICH; and median DSC=0.65, HD=5 mm, VS=0.85 for PHE. Conclusion: We trained, validated, and externally tested an end-to-end automated tool for segmentation of both ICH and PHE on non-contrast head CT. Such tool can facilitate treatment triage for trials when volume cutoffs are applied for enrollment.