Mean Dice Similarity Coefficient Research Articles

SSCK-Net: Spine segmentation in MRI based on cross attention and key-points recognition-assisted learner

Multiple spine segmentation of vertebrae and intervertebral discs for magnetic resonance images (MRI) plays a significant role in various spinal diseases diagnosis and treatments of spine disorder. However, the inherent complexity and characteristics of the spine pose challenges in balancing the inter-class similarity and intra-class variety of spine, improving generalization ability, learning rate, and accuracy. To address the above issues, a spine segmentation in MRI based on cross attention and key-points recognition-assisted learner (SSCK-Net) is proposed. Firstly, a multi-channel cross attention (MCCA) mechanism is put forward to generate a comprehensive description of the spine by fusing complementary inter-class feature and intra-class feature. Secondly, a key-points recognition-assisted learner (KRAL) is designed, which includes mixed-supervision recognition-assisted label fusion (RALF), to reduce dependence on a single dataset and improve the generalization ability of the network. Subsequently, adaptive transformer (AT) is presented to perform parallel computation through selective kernel fusion, nonlinear activation function with simple inverse and rescaling layer normalization, which reduces training time and prevents performance degradation due to long-term dependence. Finally, a feature enhances unit (FEU) is proposed to fuse the key point features and the segmented features while constraining the key-point features to the fused features again by the refine unit. Our experiments on T2-weighted volumetric MRI show that SSCK-Net achieves impressive performance with a mean Dice similarity coefficient (DSC) 96.12% for 5 vertebral bodies and 95.07% for 5 intervertebral discs on public datasets, outperforming state-of-the-art techniques.

Auto-segmentation of thoracic organs in CT scans of breast cancer patients using a 3D U-net cascaded into 2D patchGANs

The quality of organ volume delineation significantly influences the efficacy of radiotherapy treatment for breast cancer patients. This study introduces a novel method for auto-segmentation of the breasts, lungs and heart. The proposed pipeline leverages a multi-class 3D U-Net with a pre-trained ResNet(2+1)D-18 encoder branch, cascaded with a 2D PatchGAN mask correction model for each class. This approach requires a single 3D model, providing a relatively efficient solution. The models were trained and evaluated on 70 thoracic DICOM datasets belonging to breast cancer patients. The evaluation demonstrated state-of-the-art segmentation performance, with mean Dice similarity coefficient values ranging from 0.89 to 0.98, Hausdorff distance values ranging from 2.25 to 8.68 mm, and mean surface distance values ranging from 0.62 to 2.79 mm. These results underscore the pipeline’s potential to enhance breast cancer diagnosis and treatment strategies, with possible applications in other medical sectors utilizing auto-segmentation.

Weakly supervised learning-based 3D bladder reconstruction from 2D ultrasound images for bladder volume measurement.

Accurate measurement of bladder volume is necessary to maintain the consistency of the patient's anatomy in radiation therapy for pelvic tumors. As the diversity of the bladder shape, traditional methods for bladder volume measurement from 2D ultrasound have been found to produce inaccurate results. To improve the accuracy of bladder volume measurement from 2D ultrasound images for patients with pelvic tumors. The bladder ultrasound images from 130 patients with pelvic cancer were collected retrospectively. All data were split into a training set (80 patients), a validation set (20 patients), and a test set (30 patients). A total of 12 transabdominal ultrasound images for one patient were captured by automatically rotating the ultrasonic probe with an angle step of 15°. An incomplete 3D ultrasound volume was synthesized by arranging these 2D ultrasound images in 3D space according to the acquisition angles. With this as input, a weakly supervised learning-based 3D bladder reconstruction neural network model was built to predict the complete 3D bladder. The key point is that we designed a novel loss function, including the supervised loss of bladder segmentation in the ultrasound images at known angles and the compactness loss of the 3D bladder. Bladder volume was calculated by counting the number of voxels belonging to the 3D bladder. The dice similarity coefficient (DSC) was used to evaluate the accuracy of bladder segmentation, and the relative standard deviation (RSD) was used to evaluate the calculation accuracy of bladder volume with that of computed tomography (CT) images as the gold standard. The results showed that the mean DSC was up to 0.94 and the mean absolute RSD can be reduced to 6.3% when using 12 ultrasound images of one patient. Further, the mean DSC also was up to 0.90 and the mean absolute RSD can be reduced to 9.0% even if only two ultrasound images were used (i.e., the angle step is 90°). Compared with the commercial algorithm in bladder scanners, which has a mean absolute RSD of 13.6%, our proposed method showed a considerably huge improvement. The proposed weakly supervised learning-based 3D bladder reconstruction method can greatly improve the accuracy of bladder volume measurement. It has great potential to be used in bladder volume measurement devices in the future.

Applying a novel two-step deep learning network to improve the automatic delineation of esophagus in non-small cell lung cancer radiotherapy.

To introduce a model for automatic segmentation of thoracic organs at risk (OARs), especially the esophagus, in non-small cell lung cancer radiotherapy, using a novel two-step deep learning network. A total of 59 lung cancer patients' CT images were enrolled, of which 39 patients were randomly selected as the training set, 8 patients as the validation set, and 12 patients as the testing set. The automatic segmentations of the six OARs including the esophagus were carried out. In addition, two sets of treatment plans were made on the basis of the manually delineated tumor and OARs (Plan1) as well as the manually delineated tumor and the automatically delineated OARs (Plan2). The Dice similarity coefficient (DSC), 95% Hausdorff distance (HD95), and average surface distance (ASD) of the proposed model were compared with those of U-Net as a benchmark. Next, two groups of plans were also compared according to the dose-volume histogram parameters. The DSC, HD95, and ASD of the proposed model were better than those of U-Net, while the two groups of plans were almost the same. The highest mean DSC of the proposed method was 0.94 for the left lung, and the lowest HD95 and ASD were 3.78 and 1.16mm for the trachea, respectively. Moreover, the DSC reached 0.73 for the esophagus. The two-step segmentation method can accurately segment the OARs of lung cancer. The mean DSC of the esophagus realized preliminary clinical significance (>0.70). Choosing different deep learning networks based on different characteristics of organs offers a new option for automatic segmentation in radiotherapy.

Comparative Analysis of Image Processing Techniques for Enhanced MRI Image Quality: 3D Reconstruction and Segmentation Using 3D U-Net Architecture.

Osteosarcoma is a common type of bone tumor, particularly prevalent in children and adolescents between the ages of 5 and 25 who are experiencing growth spurts during puberty. Manual delineation of tumor regions in MRI images can be laborious and time-consuming, and results may be subjective and difficult to replicate. Therefore, a convolutional neural network (CNN) was developed to automatically segment osteosarcoma cancerous cells in three types of MRI images. The study consisted of five main stages. First, 3692 DICOM format MRI images were acquired from 46 patients, including T1-weighted, T2-weighted, and T1-weighted with injection of Gadolinium (T1W + Gd) images. Contrast stretching and median filter were applied to enhance image intensity and remove noise, and the pre-processed images were reconstructed into NIfTI format files for deep learning. The MRI images were then transformed to fit the CNN's requirements. A 3D U-Net architecture was proposed with optimized parameters to build an automatic segmentation model capable of segmenting osteosarcoma from the MRI images. The 3D U-Net segmentation model achieved excellent results, with mean dice similarity coefficients (DSC) of 83.75%, 85.45%, and 87.62% for T1W, T2W, and T1W + Gd images, respectively. However, the study found that the proposed method had some limitations, including poorly defined borders, missing lesion portions, and other confounding factors. In summary, an automatic segmentation method based on a CNN has been developed to address the challenge of manually segmenting osteosarcoma cancerous cells in MRI images. While the proposed method showed promise, the study revealed limitations that need to be addressed to improve its efficacy.

Reliability of whole mount radical prostatectomy histopathology as the ground truth for artificial intelligence assisted prostate imaging

The development of artificial intelligence–based imaging techniques for prostate cancer (PCa) detection and diagnosis requires a reliable ground truth, which is generally based on histopathology from radical prostatectomy specimens. This study proposes a comprehensive protocol for the annotation of prostatectomy pathology slides. To evaluate the reliability of the protocol, interobserver variability was assessed between five pathologists, who annotated ten radical prostatectomy specimens consisting of 74 whole mount pathology slides. Interobserver variability was assessed for both the localization and grading of PCa. The results indicate excellent overall agreement on the localization of PCa (Gleason pattern ≥ 3) and clinically significant PCa (Gleason pattern ≥ 4), with Dice similarity coefficients (DSC) of 0.91 and 0.88, respectively. On a per-slide level, agreement for primary and secondary Gleason pattern was almost perfect and substantial, with Fleiss Kappa of .819 (95% CI .659–.980) and .726 (95% CI .573–.878), respectively. Agreement on International Society of Urological Pathology Grade Group was evaluated for the index lesions and showed agreement in 70% of cases, with a mean DSC of 0.92 for all index lesions. These findings show that a standardized protocol for prostatectomy pathology annotation provides reliable data on PCa localization and grading, with relatively high levels of interobserver agreement. More complicated tissue characterization, such as the presence of cribriform growth and intraductal carcinoma, remains a source of interobserver variability and should be treated with care when used in ground truth datasets.

Paper 44: Deep Learning Artificial Intelligence Tool for Automated Radiographic Determination of Posterior Tibial Slope in Patients with Anterior Cruciate Ligament Injuries

Objectives: An increased posterior tibial slope (PTS) corresponds with an increased risk of graft failure following anterior cruciate ligament reconstruction (ACLR). Validated methods of human PTS measurements are subject to interobserver variability and can be inefficient on large-scale sets of images. The purpose of this study is to develop a deep learning artificial intelligence technique for the automated measurement of PTS from standard lateral knee radiographs. Methods: A deep learning U-Net model was developed on a cohort of 300 postoperative short leg lateral radiographs from ACLR patients to segment the tibial shaft, tibial joint surface, and tibial tuberosity. The model was trained via a random split following an 80:20 train-validation scheme. Masks for training images were manually segmented and the model was trained for 400 epochs. An image processing pipeline was then deployed to annotate and measure the PTS using the predicted segmentation masks. Finally, the performance of this combined pipeline was compared to human measurements performed by two study personnel using a previously validated manual technique for measuring PTS on short leg lateral radiographs on an independent test-set consisting of both preoperative and postoperative images. Results: The U-Net semantic segmentation model achieved a mean Dice similarity coefficient of 0.885 on the validation cohort. The mean difference between human-made and computer-vision measurements was 1.92⁰ (σ = 2.81⁰, P=0.24). Extreme disagreements between human and machine measurements as defined by differences ≥5⁰ occurred less than 5% of the time. The model was incorporated into a web based digital application for demonstration purposes which can provide measurement of an uploaded image in portable network graphics format in less than 5 seconds. Conclusions: We developed an accurate and reliable deep learning computer vision algorithm to automate the measurement of PTS on lateral knee radiographs. This tool will be deployed for clinical use on an institution-wide basis and, pending external validation, be made available to outside orthopaedic surgeons. This will provide an efficient and accurate tool to measure PTS during the preoperative assessment of ACL-injured patients.

Contouring in the optic plane improves the accuracy of computed tomography-based segmentation of the optic pathway.

Canine optic pathway structures are often contoured on CT images, despite the difficulty of visualizing the optic pathway with CT using standard planes. The purpose of this prospective, analytical, diagnostic accuracy study was to examine the accuracy of optic pathway contouring by veterinary radiation oncologists (ROs) before and after training on optic plane contouring. Optic pathway contours used as the gold standard for comparison were created based on expert consensus from registered CT and MRI for eight dogs. Twenty-one ROs contoured the optic pathway on CT using their preferred method, and again following atlas and video training demonstrating contouring on the optic plane. The Dice similarity coefficient (DSC) was used to assess contour accuracy. A multilevel mixed model with random effects to account for repeated measures was used to examine DSC differences. The median DSC (5th and 95th percentile) before and after training was 0.31 (0.06, 0.48) and 0.41 (0.18, 0.53), respectively. The mean DSC was significantly higher after training compared with before training (mean difference=0.10; 95% CI, 0.08-0.12; P<0.001) across all observers and patients. DSC values were comparable to those reported (0.4-0.5) for segmentation of the optic chiasm and nerves in human patients. Contour accuracy improved after training but remained low, potentially due to the small optic pathway volumes. When registered CT-MRI images are not available, our study supports routine addition of an optic plane with specific window settings to improve segmentation accuracy in mesaticephalic dogs ≥11kg.

Evaluation of a nnU-Net type automated clinical volumetric tumor segmentation tool for diffuse low-grade glioma follow-up

Background and purposeDiffuse low-grade gliomas (DLGG) are characterized by a slow and continuous growth and always evolve towards an aggressive grade. Accurate prediction of the malignant transformation is essential as it requires immediate therapeutic intervention. One of its most precise predictors is the velocity of diameter expansion (VDE). Currently, the VDE is estimated either by linear measurements or by manual delineation of the DLGG on T2 FLAIR acquisitions. However, because of the DLGG's infiltrative nature and its blurred contours, manual measures are challenging and variable, even for experts. Therefore we propose an automated segmentation algorithm using a 2D nnU-Net, to 1) gain time and 2) standardize VDE assessment. Materials and MethodsThe 2D nnU-Net was trained on 318 acquisitions (T2 FLAIR & 3DT1 longitudinal follow-up of 30 patients, including pre- & post-surgery acquisitions, different scanners, vendors, imaging parameters…). Automated vs. manual segmentation performance was evaluated on 167 acquisitions, and its clinical interest was validated by quantifying the amount of manual correction required after automated segmentation of 98 novel acquisitions. ResultsAutomated segmentation showed a good performance with a mean Dice Similarity Coefficient (DSC) of 0.82±0.13 with manual segmentation and a substantial concordance between VDE calculations. Major manual corrections (i.e., DSC<0.7) were necessary only in 3/98 cases and 81% of the cases had a DSC>0.9. ConclusionThe proposed automated segmentation algorithm can successfully segment DLGG on highly variable MRI data. Although manual corrections are sometimes necessary, it provides a reliable, standardized and time-winning support for VDE extraction to asses DLGG growth.

Validation of a New Semiautomated Segmentation Pipeline Based on the Spinal Cord Toolbox DeepSeg Algorithm to Estimate the Cervical Canal Area.

As in the brain reserve concept, a larger cervical canal area may also protect against disability. In this context, a semiautomated pipeline has been developed to obtain quantitative estimations of the cervical canal area. The aim of the study was to validate the pipeline, to evaluate the consistency of the cervical canal area measurements during a 1-year period, and to compare cervical canal area estimations obtained from brain and cervical MRI acquisitions. Eight healthy controls and 18 patients with MS underwent baseline and follow-up 3T brain and cervical spine sagittal 3D MPRAGE. The cervical canal area was measured in all acquisitions, and estimations obtained with the proposed pipeline were compared with manual segmentations performed by 1 evaluator using the Dice similarity coefficient. The cervical canal area estimations obtained on baseline and follow-up T1WI were compared; brain and cervical cord acquisitions were also compared using the individual and average intraclass correlation coefficients. The agreement between the manual cervical canal area masks and the masks provided by the proposed pipeline was excellent, with a mean Dice similarity coefficient mean of 0.90 (range, 0.73-0.97). The cervical canal area estimations obtained from baseline and follow-up scans showed a good level of concordance (intraclass correlation coefficient = 0.76; 95% CI, 0.44-0.88); estimations obtained from brain and cervical MRIs also had good agreement (intraclass correlation coefficient = 0.77; 95% CI, 0.45-0.90). The proposed pipeline is a reliable tool to estimate the cervical canal area. The cervical canal area is a stable measure across time; moreover, when cervical sequences are not available, the cervical canal area could be estimated using brain T1WI.

Incremental retraining, clinical implementation, and acceptance rate of deep learning auto-segmentation for male pelvis in a multiuser environment.

Deep learning auto-segmentation (DLAS) models have been adopted in the clinic; however, they suffer from performance deterioration owing to the clinical practice variability. Some commercial DLAS software provide an incremental retraining function that enables users to train a custom model using their institutional data to account for clinical practice variability. This study was performed to evaluate and implement the commercial DLAS software with the incremental retraining function for definitive treatment of patients with prostate cancer in a multi-user environment. CT-based target organs and organs-at-risk (OAR) delineation of 215 prostate cancer patients were utilized. The performance of three commercial DLAS software built-in models was validated with 20 patients. A retrained custom model was developed using 100 patients and evaluated on the remaining data (n=115). Dice similarity coefficient (DSC), Hausdorff distance (HD), mean surface distance (MSD), and surface DSC (SDSC) were utilized for quantitative evaluation. A multi-rater qualitative evaluation was blindly performed with a five-level scale. Visual inspection was performed in consensus and non-consensus unacceptable cases to identify the failure modes. Three commercial DLAS vendor built-in models achieved sub-optimal performance in 20 patients. The retrained custom model had a mean DSC of 0.82 for prostate, 0.48 for seminal vesicles (SV), and 0.92 for rectum, respectively. This represents a significant improvement over the built-in model with DSC of 0.73, 0.37, and 0.81 for the corresponding structures. Compared to the acceptance rate of 96.5% and consensus unacceptable rate (i.e., both reviewers rated as unacceptable) of 3.5% achieved by manual contours, the custom model achieved a 91.3% acceptance rate and 8.7% consensus unacceptable rate. The failure modes of retrained custom model were attributed to the following: cystogram (n=2), hip prosthesis (n=2), low dose rate brachytherapy seeds (n=2), air in endorectal balloon(n=1), non-iodinated spacer (n=2), and giant bladder(n=1). The commercial DLAS software with the incremental retraining function was validated and clinically adopted for prostate patients in a multi-user environment. AI-based auto-delineation of the prostate and OARs is shown to achieve improved physician acceptance, overall clinical utility, and accuracy.

Comparison of the output of a deep learning segmentation model for locoregional breast cancer radiotherapy trained on 2 different datasets.

The development of deep learning (DL) models for auto-segmentation is increasing and more models become commercially available. Mostly, commercial models are trained on external data. To study the effect of using a model trained on external data, compared to the same model trained on in-house collected data, the performance of these two DL models was evaluated. The evaluation was performed using in-house collected data of 30 breast cancer patients. Quantitative analysis was performed using Dice similarity coefficient (DSC), surface DSC (sDSC) and 95th percentile of Hausdorff Distance (95% HD). These values were compared with previously reported inter-observer variations (IOV). For a number of structures, statistically significant differences were found between the two models. For organs at risk, mean values for DSC ranged from 0.63 to 0.98 and 0.71 to 0.96 for the in-house and external model, respectively. For target volumes, mean DSC values of 0.57 to 0.94 and 0.33 to 0.92 were found. The difference of 95% HD values ranged 0.08 to 3.23mm between the two models, except for CTVn4 with 9.95mm. For the external model, both DSC and 95% HD are outside the range of IOV for CTVn4, whereas this is the case for the DSC found for the thyroid of the in-house model. Statistically significant differences were found between both models, which were mostly within published inter-observer variations, showing clinical usefulness of both models. Our findings could encourage discussion and revision of existing guidelines, to further decrease inter-observer, but also inter-institute variability.

Agreement and accuracy of fully automated morphometric femorotibial cartilage analysis in radiographic knee osteoarthritis

ObjectiveTo examine the performance of automated convolutional neuronal network cartilage segmentation in knees with radiographic osteoarthritis (ROA), and its dependence on the OA status of the training set and MRI sequences. DesignWe studied 122 ROA and 92 healthy reference cohort (HRC) knees from the Osteoarthritis Initiative. All knees had expert manual segmentation of the femorotibial cartilages based on coronal FLASH and sagittal DESS MRI. Two U-net convolutional neural networks were trained on 86/50 ROA/HRC knees, validated on 18/21, and tested on 18/21. ResultsOf 122 ROA knees, 43 (35%) were KLG2, 41 (34%) KLG3, and 38 (31%) KLG4. In the ROA test set, Dice Similarity Coefficients (DSCs) were 0.86/0.86 (FLASH/DESS) for the algorithm trained on ROA, and 0.82/0.82 for that trained on HRC knees. In the HRC test set, mean DSCs of 0.91/0.90 were observed with FLASH/DESS, both with the HRC- and with the ROA-trained algorithm. Cartilage thickness computation from automated segmentation in the FLASH ROA test set showed a correlation of r = 0.94 with manual segmentation for the ROA-trained, and of r = 0.89 for the HRC-trained algorithm. In the FLASH HRC test set, it exhibited a correlation of r = 0.96 for the HRC-trained and r = 0.88 for the ROA-trained algorithm. Results were similar between the DESS and FLASH, but were less accurate for KLG4 than for KLG2/3 knees. ConclusionsAn automated algorithm trained on ROA knees was able to accurately segment and compute cartilage thickness on FLASH and DESS MRI, in both ROA and healthy knees, whereas an algorithm trained on healthy knees did not perform as well in ROA knees.

Clinical evaluation of a deep learning segmentation model including manual adjustments afterwards for locally advanced breast cancer

Deep learning (DL) models are increasingly developed for auto-segmentation in radiotherapy. Qualitative analysis is of great importance for clinical implementation, next to quantitative. This study evaluates a DL segmentation model for left- and right-sided locally advanced breast cancer both quantitatively and qualitatively. For each side a DL model was trained, including primary breast CTV (CTVp), lymph node levels 1-4, heart, lungs, humeral head, thyroid and esophagus. For evaluation, both automatic segmentation, including correction of contours when needed, and manual delineation was performed and both processes were timed. Quantitative scoring with dice-similarity coefficient (DSC), 95% Hausdorff Distance (95%HD) and surface DSC (sDSC) was used to compare both the automatic (not-corrected) and corrected contours with the manual contours. Qualitative scoring was performed by five radiotherapy technologists and five radiation oncologists using a 3-point Likert scale. Time reduction was achieved using auto-segmentation in 95% of the cases, including correction. The time reduction (mean±std) was 42.4% ± 26.5% and 58.5% ± 19.1% for OARs and CTVs, respectively, corresponding to an absolute mean reduction (hh:mm:ss) of 00:08:51 and 00:25:38. Good quantitative results were achieved before correction, e.g. mean DSC for the right-sided CTVp was 0.92±0.06, whereas correction statistically significantly improved this contour by only 0.02±0.05, respectively. In 92% of the cases, auto-contours were scored as clinically acceptable, with or without corrections. A DL segmentation model was trained and was shown to be a time-efficient way to generate clinically acceptable contours for locally advanced breast cancer.

Lithium Metal Battery Quality Control via Transformer-CNN Segmentation.

Lithium metal battery (LMB) has the potential to be the next-generation battery system because of its high theoretical energy density. However, defects known as dendrites are formed by heterogeneous lithium (Li) plating, which hinders the development and utilization of LMBs. Non-destructive techniques to observe the dendrite morphology often use X-ray computed tomography (XCT) to provide cross-sectional views. To retrieve three-dimensional structures inside a battery, image segmentation becomes essential to quantitatively analyze XCT images. This work proposes a new semantic segmentation approach using a transformer-based neural network called TransforCNN that is capable of segmenting out dendrites from XCT data. In addition, we compare the performance of the proposed TransforCNN with three other algorithms, U-Net, Y-Net, and E-Net, consisting of an ensemble network model for XCT analysis. Our results show the advantages of using TransforCNN when evaluating over-segmentation metrics, such as mean intersection over union (mIoU) and mean Dice similarity coefficient (mDSC), as well as through several qualitatively comparative visualizations.

Comprehensive Evaluation of a Deep Learning Model for Automatic Organs-at-Risk Segmentation on Heterogeneous Computed Tomography Images for Abdominal Radiation Therapy

Our purpose was to develop a deep learning model (AbsegNet) that produces accurate contours of 16 organs at risk (OARs) for abdominal malignancies as an essential part of fully automated radiation treatment planning. Three data sets with 544 computed tomography scans were retrospectively collected. Data set 1 was split into 300 training cases and 128 test cases (cohort 1) for AbsegNet. Data set 2, including cohort 2 (n=24) and cohort 3 (n=20), were used to validate AbsegNet externally. Data set 3, including cohort 4 (n=40) and cohort 5 (n=32), were used to clinically assess the accuracy of AbsegNet-generated contours. Each cohort was from a different center. The Dice similarity coefficient and 95th-percentile Hausdorff distance were calculated to evaluate the delineation quality for each OAR. Clinical accuracy evaluation was classified into 4 levels: no revision, minor revisions (0% < volumetric revision degrees [VRD] ≤ 10%), moderate revisions (10% ≤ VRD < 20%), and major revisions (VRD ≥20%). For all OARs, AbsegNet achieved a mean Dice similarity coefficient of 86.73%, 85.65%, and 88.04% in cohorts 1, 2, and 3, respectively, and a mean 95th-percentile Hausdorff distance of 8.92, 10.18, and 12.40 mm, respectively. The performance of AbsegNet outperformed SwinUNETR, DeepLabV3+, Attention-UNet, UNet, and 3D-UNet. When experts evaluated contours from cohorts 4 and 5, 4 OARs (liver, kidney_L, kidney_R, and spleen) of all patients were scored as having no revision, and over 87.5% of patients with contours of the stomach, esophagus, adrenals, or rectum were considered as having no or minor revisions. Only 15.0% of patients with colon and small bowel contours required major revisions. We propose a novel deep-learning model to delineate OARs on diverse data sets. Most contours produced by AbsegNet are accurate and robust and are, therefore, clinically applicable and helpful to facilitate radiation therapy workflow.

AI-Based Isotherm Prediction for Focal Cryoablation of Prostate Cancer

Focal therapies have emerged as minimally invasive alternatives for patients with localized low-risk prostate cancer (PCa) and those with postradiation recurrence. Among the available focal treatment methods for PCa, cryoablation offers several technical advantages, including the visibility of the boundaries of frozen tissue on the intraprocedural images, access to anterior lesions, and the proven ability to treat postradiation recurrence. However, predicting the final volume of the frozen tissue is challenging as it depends on several patient-specific factors, such as proximity to heat sources and thermal properties of the prostatic tissue. This paper presents a convolutional neural network model based on 3D-Unet to predict the frozen isotherm boundaries (iceball) resultant from a given a cryo-needle placement. Intraprocedural magnetic resonance images acquired during 38 cases of focal cryoablation of PCa were retrospectively used to train and validate the model. The model accuracy was assessed and compared against a vendor-provided geometrical model, which is used as a guideline in routine procedures. The mean Dice Similarity Coefficient using the proposed model was 0.79±0.08 (mean+SD) vs 0.72±0.06 using the geometrical model (P<.001). The model provided an accurate iceball boundary prediction in less than 0.4second and has proven its feasibility to be implemented in an intraprocedural planning algorithm.

Deep learning to detect macular atrophy in wet age-related macular degeneration using optical coherence tomography

Here, we have developed a deep learning method to fully automatically detect and quantify six main clinically relevant atrophic features associated with macular atrophy (MA) using optical coherence tomography (OCT) analysis of patients with wet age-related macular degeneration (AMD). The development of MA in patients with AMD results in irreversible blindness, and there is currently no effective method of early diagnosis of this condition, despite the recent development of unique treatments. Using OCT dataset of a total of 2211 B-scans from 45 volumetric scans of 8 patients, a convolutional neural network using one-against-all strategy was trained to present all six atrophic features followed by a validation to evaluate the performance of the models. The model predictive performance has achieved a mean dice similarity coefficient score of 0.706 ± 0.039, a mean Precision score of 0.834 ± 0.048, and a mean Sensitivity score of 0.615 ± 0.051. These results show the unique potential of using artificially intelligence-aided methods for early detection and identification of the progression of MA in wet AMD, which can further support and assist clinical decisions.

Automatic multi-label temporal bone computed tomography segmentation with deep learning.

Manually segmenting temporal bone computed tomography (CT) images is difficult. Despite accurate automatic segmentation in previous studies using deep learning, they did not consider clinical differences, such as variations in CT scanners. Such differences can significantly affect the accuracy of segmentation. Our dataset included 147 scans from three different scanners, and we used Res U-Net, SegResNet, and UNETR neural networks to segment four structures: the ossicular chain (OC), internal auditory canal (IAC), facial nerve (FN), and labyrinth (LA). The experimental results yielded high mean Dice similarity coefficients of 0.8121, 0.8809, 0.6858, 0.9329, and a low mean of 95% Hausdorff distances of 0.1431mm, 0.1518mm, 0.2550mm, and 0.0640mm for OC, IAC, FN, and LA, respectively. This study shows that automated deep learning-based segmentation techniques successfully segment temporal bone structures using CT data from different scanners. Our research can further promote its clinical application.

Automatic segmentation and quantification of the optic nerve on MRI using a 3D U-Net.

Pathological conditions associated with the optic nerve (ON) can cause structural changes in the nerve. Quantifying these changes could provide further understanding of disease mechanisms. We aim to develop a framework that automatically segments the ON separately from its surrounding cerebrospinal fluid (CSF) on magnetic resonance imaging (MRI) and quantifies the diameter and cross-sectional area along the entire length of the nerve. Multicenter data were obtained from retinoblastoma referral centers, providing a heterogeneous dataset of 40 high-resolution 3D T2-weighted MRI scans with manual ground truth delineations of both ONs. A 3D U-Net was used for ON segmentation, and performance was assessed in a tenfold cross-validation () and on a separate test-set () by measuring spatial, volumetric, and distance agreement with manual ground truths. Segmentations were used to quantify diameter and cross-sectional area along the length of the ON, using centerline extraction of tubular 3D surface models. Absolute agreement between automated and manual measurements was assessed by the intraclass correlation coefficient (ICC). The segmentation network achieved high performance, with a mean Dice similarity coefficient score of 0.84, median Hausdorff distance of 0.64mm, and ICC of 0.95 on the test-set. The quantification method obtained acceptable correspondence to manual reference measurements with mean ICC values of 0.76 for the diameter and 0.71 for the cross-sectional area. Compared with other methods, our method precisely identifies the ON from surrounding CSF and accurately estimates its diameter along the nerve's centerline. Our automated framework provides an objective method for ON assessment in vivo.