Computed Tomography In Patients Research Articles

Developing a virtual Endoscopic Surgery Planning system to optimize surgical outcomes.

Planning and predicting functional outcomes of endoscopic sinus surgeries (e.g., nasal airflow) based solely on visualizing Computerized Tomography (CT) or endoscopy poses a challenge to produce optimal clinical outcomes. Technology development, retrospective case report. A virtual surgery planning (VSP) tool is developed that can load any patient's CT data and allow surgeons to remove obstructive tissue using both visual and haptic feedback endoscopically. Pre-calculated airflow resistance, wall shear stress, pressure drop are displayed on the anatomy to identify potential sites of obstruction. After each virtual surgery, changes in nasal airflow can be computed, and the process is reiterated until an optimal result is reached. As proof-of-concept, a series of isolated or combined procedures were performed on CT of one patient, who had olfactory losses that may involve obstructions blocking the air/odor flow to the olfactory fossa (OF). For this patient, an isolated medial partial middle turbinectomy (PMT) demonstrated the best outcome, better than traditionally performed lateral PMT, while septal body reduction worsened air/odor flow to OF. This proof of concept case report demonstrates the potential usefulness of VSP in preoperative planning based on objective benchmarks and could be a valuable tool for optimizing future surgical outcomes.

A Concise and Accurate Method for Evaluating Alveolar Cleft ReconstructionDiet al Concise and Accurate Method for Evaluating.

Currently, there remain unresolved issues in the treatment of alveolar clefts, the resolution of which could greatly benefit many patients with cleft lip and palate. In alveolar cleft treatment research, a reliable tool for pre- and postoperative assessment is crucial. This study presents a concise and accurate method for postoperative evaluations of alveolar treatment, which can rapidly and accurately obtain the shape and volume of the newly formed bone. This study included helical computed tomography (CT) datasets of 20 patients who underwent alveolar bone grafting at our institute. Two observers independently measured the volume of newly formed bone in the patient's CT images one year postoperatively, with each observer performing the measurement twice. To acquire the volume of the newly formed bone at 1 year postoperatively, the model of the newly formed bone must be constructed first. The acquisition of this model involves Boolean operations on registered preoperative and postoperative cranial 3-dimensional (3D) images. The registration of the preoperative and postoperative models is performed in MIMICS software, and the registration results can be directly confirmed layer by layer on the CT images to ensure accuracy. The mean newly formed bone ratio in this study was 39.81%±17.96%, and the mean processing time was 05:11±01:41 minutes. The intraclass correlation coefficient for bone volume measurements between the two observers was 0.999, indicating high consistency and reproducibility. This method enhances accuracy, is time-efficient, and demonstrates high reliability in evaluating postoperative bone formation.

INCIDENTAL BONE METASTASIS FINDING FROM CT SIMULATION FOR BREAST CANCER RADIOTHERAPY

Breast cancer has the highest tendency for bone metastasis. This study aimed to evaluate bone metastasis using computed tomography (CT) simulation for breast cancer radiotherapy. We retrospectively reviewed the patient's CT images from September 2022 to February 2023. The median age was 48 (34–80), and almost half of the patients had T3-4 and N2-3 disease. Nine out of 53 patients (17%) had bone metastasis, and 8 had multiple lesions. The vertebral spine was the predominant site of bone metastasis. Due to the high rate of bone spreading, bone evaluation is mandatory during CT simulation for more appropriate treatment. Keywords: Bone metastasis, computed tomography, breast cancer

Effect of the application of digital technology-assisted optimization in the process of adjusting jaw position.

The aim of this study was to demonstrate a novel jaw position adjustment technique derived from digital twins and evaluate the application effect of digital technology-assisted optimization in the process of adjusting jaw position on patients with temporomandibular disorders (TMD). A total of 74 patients with TMD who attended the Department of Temporomandibular Joint, West China Hospital of Stomatology, Si-chuan University, between June 2022 and May 2023 were selected. The patient's initial computed tomography (CT) and bilateral temporomandibular joint data obtained by magnetic resonance imaging (MRI) were collected. The 148 joints were divided into the normal disc-condyle relationship (N) group, disc displacement with reduction (DDWR) group, and disc displacement without reduction (DDWoR) group. Assisted by digital technology, the patient's CT data were reconstructed, and a personalized reference plane was established to adjust the jaw position. A three-point bite guiding splint was designed by the adjusted occlusal space and then fabricated by 3D printing technology. It was worn by the patients and then reviewed by MRI. Before and after the adjustment of jaw position, the amount and direction of condyle and disc displacement and the angle between condyle and disc were measured as the evaluation indexes of the effect of the adjustment. The correlation with condylar displacement was evaluated. In the N group, the disc moved backward and downward along the X and Z axes by (-0.60±0.62) and (0.51±0.71) mm, respectively. In the DDWR group, the disc moved backward and upward along the X and Z axes by (-1.33±1.38) and (-0.09±1.31) mm, respectively. In the DDWoR group, the disc moved forward and downward along the X and Z axes by (0.49±1.76) and (1.35±1.76) mm, respectively. The angle between the condyle and the disc decreased after adjustment of the jaw position in all three groups. All patients showed improvement in symptoms after adjustment. Digital technology-assisted jaw position adjustment can simplify the process, reduce the sensitivity of the technique, and improve patients' disc-condyle structure and symptoms. Therefore, its application in the treatment of patients with TMD is of great clinical significance.

Machine Learning Model Based on Radiomics for Preoperative Differentiation of Jaw Cystic Lesions.

This study aims to use machine learning techniques together with radiomics methods to build a preoperative predictive diagnostic model from spiral computed tomography (CT) images. The model is intended for the differential diagnosis of common jaw cystic lesions. Retrospective, case-control study. This retrospective study was conducted at Sun Yat-sen Memorial Hospital of Sun Yat-sen University (Guangzhou, Guangdong, China). All the data used to build the predictive diagnostic model were collected from 160 patients, who were treated at the Department of Oral and Maxillofacial Surgery at Sun Yat-sen Memorial Hospital of Sun Yat-sen University between 2019 and 2023. We included a total of 160 patients in this study. We extracted 107 radiomic features from each patient's CT scan images. After a feature selection process, we chose 15 of these radiomic features to construct the predictive diagnostic model. Among the preoperative predictive diagnostic models built using 3 different machine learning methods (support vector machine, random forest [RF], and multivariate logistic regression), the RF model showed the best predictive performance. It demonstrated a sensitivity of 0.923, a specificity of 0.643, an accuracy of 0.825, and an area under the receiver operating characteristic curve of 0.810. The preoperative predictive model, based on spiral CT radiomics and machine learning algorithms, shows promising differential diagnostic capabilities. For common jaw cystic lesions, this predictive model has potential clinical application value, providing a scientific reference for treatment decisions.

SeRL: Style-embedding representation learning for unsupervised CT images synthesis from unpaired MR images

Hepatocellular carcinoma (HCC) is the third leading cause of cancer-related deaths worldwide, and the fastest-growing cause of cancer deaths in the United States. Computed tomography (CT) and magnetic resonance (MR) imaging are the key image-based examination methods for HCC patients. Patients may alternatively undergo CT and MR exams given a radiation dose of the former and the cost of the latter. Due to organ representation differences between CT and MR, it’s a great challenge to detect the tumors from a time series of data consisting of both CT and MR by a single model. The annotations for these time series data will consume much time and labor for the physicians. Thus, we propose our style-embedding representation learning (SeRL) for unsupervised and unpaired abdomen CT and MR translation. Different from current medical image translation models, the style-representation information from real CT and real MR images has been embedded in the translation process to bypass some local minima during the convergence process and improve the synthesis results. Our patch-corrosion augmentation method enhances the style-embedding representation learning by bringing more diversity to the training data. Combined with the self-attention module, SeRL eliminates the noise caused by low grayscale pixel values during translation. Results on the unpaired HCC patient's CT and MR images show that our proposed SeRL is able to generate high quality CT images from MR ones. Evaluations such as Frechet Inception Distance (FID), Sliced Wasserstein Distance (SWD), and liver segmentation dice score are utilized to demonstrate our advantages over other state-of-the-art unsupervised methods.

GIFNet: an effective global infection feature network for automatic COVID-19 lung lesions segmentation.

The ongoing COronaVIrus Disease 2019 (COVID-19) pandemic carried by the SARS-CoV-2 virus spread worldwide in early 2019, bringing about an existential health catastrophe. Automatic segmentation of infected lungs from COVID-19 X-ray and computer tomography (CT) images helps to generate a quantitative approach for treatment and diagnosis. The multi-class information about the infected lung is often obtained from the patient's CT dataset. However, the main challenge is the extensive range of infected features and lack of contrast between infected and normal areas. To resolve these issues, a novel Global Infection Feature Network (GIFNet)-based Unet with ResNet50 model is proposed for segmenting the locations of COVID-19 lung infections. The Unet layers have been used to extract the features from input images and select the region of interest (ROI) by using the ResNet50 technique for training it faster. Moreover, integrating the pooling layer into the atrous spatial pyramid pooling (ASPP) mechanism in the bottleneck helps for better feature selection and handles scale variation during training. Furthermore, the partial differential equation (PDE) approach is used to enhance the image quality and intensity value for particular ROI boundary edges in the COVID-19 images. The proposed scheme has been validated on two datasets, namely the SARS-CoV-2 CT scan and COVIDx-19, for detecting infected lung segmentation (ILS). The experimental findings have been subjected to a comprehensive analysis using various evaluation metrics, including accuracy (ACC), area under curve (AUC), recall (REC), specificity (SPE), dice similarity coefficient (DSC), mean absolute error (MAE), precision (PRE), and mean squared error (MSE) to ensure rigorous validation. The results demonstrate the superior performance of the proposed system compared to the state-of-the-art (SOTA) segmentation models on both X-ray and CT datasets.

Robotic-assisted navigation system for preoperative lung nodule localization: a pilot study.

Preoperative percutaneous computed tomography (CT)-guided localization of pulmonary nodules plays a pivotal role in the diagnosis and treatment of early-stage lung cancer. However, conventional manual localization techniques have inherent limitations in achieving a high degree of accuracy. Consequently, a novel robotic-assisted navigation system was developed to attain precise localization of small lung nodules. This study aims to investigate the accuracy and safety of this system in clinical applications. Patients with peripheral solitary pulmonary nodules measuring less than 20 mm were enrolled. The robotic-assisted navigation system generated a three-dimensional (3D) model based on the patient's CT images, determining the optimal puncture path. The robotic arm then accurately located the nodule and, following percutaneous puncture, indocyanine green (ICG) was injected. The primary outcome measure was the accuracy of pulmonary nodule localization, while secondary outcomes included the complication rate, procedural duration, and total radiation exposure. A total of 33 nodules were successfully localized using the robotic-assisted navigation system and resected through video-assisted thoracoscopic surgery (VATS). The first-pass success rate was 100%, with a median deviation of 6.1 mm [interquartile range (IQR), 2.5-7.2 mm] between the localizer and the nodule. The median localization time was 25.0 minutes, and the single and cumulative exam dose-length products (DLP) were 534.0 and 1491.0 mGy·cm, respectively. Notably, no observable complications were reported during the procedures. The innovative robotic-assisted navigation system demonstrated satisfactory accuracy and holds promise for improving the percutaneous localization of lung nodules. This method represents a safe and viable alternative to traditional CT-guided manual localization techniques.

Development and evaluation of an integrated model based on a deep segmentation network and demography-added radiomics algorithm for segmentation and diagnosis of early lung adenocarcinoma.

Non-invasive early detection and differentiation grading of lung adenocarcinoma using computed tomography (CT) images are clinically important for both clinicians and patients, including determining the extent of lung resection. However, these are difficult to accomplish using preoperative images, with CT-based diagnoses often being different from postoperative pathologic diagnoses. In this study, we proposed an integrated detection and classification algorithm (IDCal) for diagnosing ground-glass opacity nodules (GGN) using CT images and other patient informatics, and compared its performance with that of other diagnostic modalities. All labeling was confirmed by a thoracic surgeon by referring to the patient's CT image and biopsy report. The detection phase was implemented via a modified FC-DenseNet to contour the lesions as elaborately as possible and secure the reliability of the classification phase for subsequent applications. Then, by integrating radiomics features and other patients' general information, the lesions were dichotomously reclassified into "non-invasive" (atypical adenomatous hyperplasia, adenocarcinoma in situ, and minimally invasive adenocarcinoma) and "invasive" (invasive adenocarcinoma). Data from 168 GGN cases were used to develop the IDCal, which was then validated in 31 independent CT scans. IDCal showed a high accuracy of GGN detection (sensitivity, 0.970; false discovery rate, 0.697) and classification (accuracy, 0.97; f1-score, 0.98; ROAUC, 0.96). In conclusion, the proposed IDCal detects and classifies GGN with excellent performance. Thus, it can be suggested that our multimodal prediction model has high potential as an auxiliary diagnostic tool of GGN to help clinicians.

Development and Validation of a Deep Learning-Based Auto-Delineation of Target Volume and Organs at Risk in Pancreatic Cancer Radiotherapy.

The delineation of the clinical target volume (CTV), gross target volume (GTV) and organs at risk (OARs) is a crucial and laborious in pancreatic cancer radiotherapy. In this work, we propose and evaluate a three-dimensional (3D) novel convolutional neural network (CNN) for automatic and accurate CTV, GTV and OARs in pancreatic cancer. A total of 120 computed tomography (CT) scans patients with pancreatic cancer were collected. A novel 3D CNN network, called ResUNet3D, was developed to achieve auto-delineation. 96 patients chosen randomly were used for training, 12 patients for validation, and 12 patients for testing. Meanwhile, the Dice similarity coefficient (DSC) and 95th percentile Hausdorff distance (HD95%) were used to assess the performance. The DSC values for the test data were 80.9±8.6%, 77.5±5.6%, 94.5±1.3%, 66.2±13.4%, 73.6±7.6%, 79.0±8.7%, 94.1±1.9%, 94.6±1.4%, 87.3±5.8% for CTV, GTV, liver, duodenum, spinal cord, bowel, kidney left, kidney right, stomach. The corresponding HD95% values were 10.7±6.9mm, 7.8±5.7mm, 11.6±5.6mm, 18.6±5.6mm, 2.7±0.7mm, 17.7±8.6mm, 3.9±1.4mm, 3.7±1.9mm, 13.4±5.7mm, respectively. The average delineation time for one patient's CT images was within 5 seconds. The experimental results demonstrate that the CTV, GTV and OARs delineated for pancreatic cancer by ResUNet3D achieved a close agreement with the ground truth. ResUNet3D could significantly reduce the radiation oncologists' contouring time.

Utilizing patient-specific 3D printed kidney surgical guide with realistic phantom for partial nephrectomy

Partial nephrectomy has been demonstrated to preserve renal function compared with radical nephrectomy. Computed tomography (CT) is used to reveal localized renal cell carcinoma (RCC). However, marking RCC directly and quantitatively on a patient's kidney during an operation is difficult. We fabricated and evaluated a 3D-printed kidney surgical guide (3DP-KSG) with a realistic kidney phantom. The kidney phantoms including parenchyma and three different RCC locations and 3DP-KSG were designed and fabricated based on a patient's CT image. 3DP-KSG was used to insert 16-gauge intravenous catheters into the kidney phantoms, which was scanned by CT. The catheter insertion points and angle were evaluated. The measurement errors of insertion points were 1.597 ± 0.741 mm, and cosine similarity of trajectories was 0.990 ± 0.010. The measurement errors for X-axis, Y-axis, and Z-axis in the insertion point were 0.611 ± 0.855 mm, 0.028 ± 1.001 mm, and − 0.510 ± 0.923 mm. The 3DP-KSG targeted the RCC accurately, quantitatively, and immediately on the surface of the kidney, and no significant difference was shown between the operators. Partial nephrectomy will accurately remove the RCC using 3DP-KSG in the operating room.

Estimation of Bone Mineral Density in the Femoral Neck and Lumbar Spine using Texture Analysis of Chest and Pelvis Computed Tomography Hounsfield Unit.

This study aimed to establish an academic basis for using a computed tomography (CT) model for predicting osteoporosis in the clinical setting by illustrating the effectiveness of morphometric texture analysis. We introduce texture analysis and quantitative approaches using CT Hounsfield units (HU) to screen osteoporosis. From March 6th, 2013, to August 11th, 2020, a total of 4,333 cases (1,766 patients) were included in the study. After applying exclusion criteria concerning the patient status and scan interval between CT and DXA, we selected only 1,647 samples (736 patients) and analyzed both their CT and DXA bone mineral density (BMD) results. BMD was measured in the femoral neck and L1 spine body. A region of interest (ROI) was extracted from each patient's CT as the maximum trabecular area of the L1 spine body and femoral neck. A total of 45 texture features were extracted from every ROI using gray-level co-occurrence matrices. Machine-learning techniques, including linear regression (LR) and artificial neural network (ANN), were applied to predict BMD. We assigned samples to (1) Set 1 (857 lumbar spine samples in chest model, L1 spine DXA BMD), (2) Set 2 (392 lumbar spine samples in lumbar spine CT model, L1 spine DXA BMD), (3) Set 3 (1,249 lumbar spine samples in both chest and lumbar spine CT model, L1 spine DXA BMD), (4) Set 4 (398 femoral neck samples in hip and pelvis CT model, femoral neck DXA BMD), and (5) Set 5 (a total of 1,647 samples). When we applied LR, the correlation coefficients between estimated and reference values for Sets 1, 2, 3, and 4 were 0.783, 0.784, 0.757, and 0.652, respectively. For total samples (Set 5), LR and ANN provided correlation coefficients of 0.707 and 0.782, respectively. The modality using morphometric texture analysis with CT HU can be an additional diagnostic tool for osteoporosis and an alternative for DXA.

CT-based radiomic phenotypes of lung adenocarcinoma: a preliminary comparative analysis with targeted next-generation sequencing.

This study aimed to explore the relationship between computed tomography (CT)-based radiomic phenotypes and genomic profiles, including expression of programmed cell death-ligand 1 (PD-L1) and the 10 major genes, such as epidermal growth factor receptor (EGFR), tumor protein 53 (TP53), and Kirsten rat sarcoma viral oncogene (KRAS), in patients with lung adenocarcinoma (LUAD). In total, 288 consecutive patients with pathologically confirmed LUAD were enrolled in this retrospective study. Radiomic features were extracted from preoperative CT images, and targeted genomic data were profiled through next-generation sequencing. PD-L1 expression was assessed by immunohistochemistry staining (chi-square test or Fisher's exact test for categorical data and the Kruskal-Wallis test for continuous data). A total of 1,013 radiomic features were obtained from each patient's CT images. Consensus clustering was used to cluster patients on the basis of radiomic features. The 288 patients were classified according to consensus clustering into four radiomic phenotypes: Cluster 1 (n = 11) involving mainly large solid masses with a maximum diameter of 5.1 ± 2.0 cm; Clusters 2 and 3 involving mainly part-solid and solid masses with maximum diameters of 2.1 ± 1.4 cm and 2.1 ± 0.9 cm, respectively; and Cluster 4 involving mostly small ground-glass opacity lesions with a maximum diameter of 1.0 ± 0.9 cm. Differences in maximum diameter, PD-L1 expression, and TP53, EGFR, BRAF, ROS1, and ERBB2 mutations among the four clusters were statistically significant. Regarding targeted therapy and immunotherapy, EGFR mutations were highest in Cluster 2 (73.1%); PD-L1 expression was highest in Cluster 1 (45.5%). Our findings provide evidence that CT-based radiomic phenotypes could non-invasively identify LUADs with different molecular characteristics, showing the potential to provide personalized treatment decision-making support for LUAD patients.

A deep-learning-based dose verification tool utilizing fluence maps for a cobalt-60 compensator-based intensity-modulated radiation therapy system.

A novel cobalt-60 compensator-based intensity-modulated radiation therapy (IMRT) system was developed for a resource-limited environment but lacked an efficient dose verification algorithm. The aim of this study was to develop a deep-learning-based dose verification algorithm for accurate and rapid dose predictions. A deep-learning network was employed to predict the doses from static fields related to beam commissioning. Inputs were a cube-shaped phantom, a beam binary mask, and an intersecting volume of the phantom and beam binary mask, while output was a 3-dimensional (3D) dose. The same network was extended to predict patient-specific doses for head and neck cancers using two different approaches. A field-based method predicted doses for each field and combined all calculated doses into a plan, while the plan-based method combined all nine fluences into a plan to predict doses. Inputs included patient computed tomography (CT) scans, binary beam masks, and fluence maps truncated to the patient's CT in 3D. For static fields, predictions agreed well with ground truths with average deviations of less than 0.5% for percent depth doses and profiles. Even though the field-based method showed excellent prediction performance for each field, the plan-based method showed better agreement between clinical and predicted dose distributions. The distributed dose deviations for all planned target volumes and organs at risk were within 1.3Gy. The calculation speed for each case was within two seconds. A deep-learning-based dose verification tool can accurately and rapidly predict doses for a novel cobalt-60 compensator-based IMRT system.

Learning from pseudo-lesion: a self-supervised framework for COVID-19 diagnosis.

The Coronavirus disease 2019 (COVID-19) has rapidly spread all over the world since its first report in December 2019, and thoracic computed tomography (CT) has become one of the main tools for its diagnosis. In recent years, deep learning-based approaches have shown impressive performance in myriad image recognition tasks. However, they usually require a large number of annotated data for training. Inspired by ground glass opacity, a common finding in COIVD-19 patient's CT scans, we proposed in this paper a novel self-supervised pretraining method based on pseudo-lesion generation and restoration for COVID-19 diagnosis. We used Perlin noise, a gradient noise based mathematical model, to generate lesion-like patterns, which were then randomly pasted to the lung regions of normal CT images to generate pseudo-COVID-19 images. The pairs of normal and pseudo-COVID-19 images were then used to train an encoder-decoder architecture-based U-Net for image restoration, which does not require any labeled data. The pretrained encoder was then fine-tuned using labeled data for COVID-19 diagnosis task. Two public COVID-19 diagnosis datasets made up of CT images were employed for evaluation. Comprehensive experimental results demonstrated that the proposed self-supervised learning approach could extract better feature representation for COVID-19 diagnosis, and the accuracy of the proposed method outperformed the supervised model pretrained on large-scale images by 6.57% and 3.03% on SARS-CoV-2 dataset and Jinan COVID-19 dataset, respectively.

Prevalence of positive findings of brain computed tomography scans in pediatric population.

Computed tomography (CT) scans are used more frequently in medical centers, increasing unnecessary requests for it as a first-line evaluation. This study aimed to investigate the rate and prevalence of abnormal findings in a brain CT scan in children at the Children's Hospital of Tabriz, Iran. This study was a cross-sectional descriptive-analytical study that included all children under 15 years old undergoing a brain CT scan at the Children's Hospital of Tabriz, Iran. All patients who fulfilled the inclusion criteria and were referred to the Children's Hospital of Tabriz during the spring of 2018 entered the study. Age, gender, patient history, and clinical findings were examined. In the next step, the results of each patient's CT scan were evaluated. In this study, 108 patients were studied with a median age of 18.0 months. CT scan results were normal in seventy-four cases (68.5%), hydrocephalus was seen in 15 (13.9%), and benign infantile hydrocephalus was seen in eight (7.4%). The study revealed a statistically significant relationship between patient history and CT scan results (p=0.017). A statistically significant relationship was observed between the clinical findings and CT scan results (p=0.042). Brain CT scans have more abnormal findings in patients with positive clinical findings. Although a CT scan is a highly sensitive and specificmodality in diagnosing central nervous system (CNS) abnormalities, its value depends on the underlying medical history and physical exam.

Development of a X-ray Exposure Dose Control Tool for CT Examination by Using Spreadsheet Software

We developed X-ray dose monitoring and optimization software for computed tomography (CT) by using a spreadsheet software. The analysis was carried out on 1,212 patients using 2,128 CT examinations at our institution, in the period of April 2020 to April 2022. These cases were extracted in the limitation of patient's weight in the range of 50 to 70 kg. The individual radiation report, including CTDIvol and DLP, on the console displayed by a CT system was used and manually transferred to our software. The X-ray dose distribution was evaluated by using a boxplot chart and a scatter plot, and the data were referred to Japan DRLs 2020 (National Diagnostic Reference Levels in Japan 2020). Almost all patient's doses in our institution were distributed in the acceptable ranges in comparison with the Japan DRLs 2020. Neither extremely high nor low dose data were recorded. However, seven cases recorded outlier dose values. Three cases were caused by errors during manually inputting the data. Four from seven cases were explored the reason to record outlier dose values. Our software on a spreadsheet software worked well to explore the CT dose data, such as CTDIvol and DLP. Our software was able to find cases that were recorded as outlier dose values, but their values were not extremely high/low values. The cautionary notice of either high or low dose, as a function of our software, will be able to keep our patient's CT dose in the universal-standard "justification and optimization" level.

Prediction of obstructive sleep apnea using deep learning in 3D craniofacial reconstruction.

Obstructive sleep apnea (OSA) is a common sleep disorder. However, current diagnostic methods are labor-intensive and require professionally trained personnel. We aimed to develop a deep learning model using upper airway computed tomography (CT) to predict OSA and to warn the medical technician if a patient has OSA while the patient is undergoing any head and neck CT scan, even for other diseases. A total of 219 patients with OSA [apnea-hypopnea index (AHI) ≥10/h] and 81 controls (AHI <10/h) were enrolled. We reconstructed each patient's CT into 3 types (skeletal structures, external skin structures, and airway structures) and captured reconstructed models in 6 directions (front, back, top, bottom, left profile, and right profile). The 6 images from each patient were imported into the ResNet-18 network to extract features and output the probability of OSA using two fusion methods: Add and Concat. Five-fold cross-validation was used to reduce bias. Finally, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) were calculated. All 18 views with Add as the feature fusion performed better than did the other reconstruction and fusion methods. This gave the best performance for this prediction method with an AUC of 0.882. We present a model for predicting OSA using upper airway CT and deep learning. The model has satisfactory performance and enables CT to accurately identify patients with moderate to severe OSA.

Clinical Validation and Treatment Plan Evaluation Based on Autodelineation of the Clinical Target Volume for Prostate Cancer Radiotherapy.

Clinical target volumes (CTVs) and organs at risk (OARs) could be autocontoured to save workload. This study aimed to assess a convolutional neural network for automatic and accurate CTV and OARs in prostate cancer, while comparing possible treatment plans based on autocontouring CTV to clinical treatment plans. Computer tomography (CT) scans of 217 patients with locally advanced prostate cancer treated at our hospital were retrospectively collected and analyzed from January 2013 to January 2019. A deep learning-based method, CUNet, was used to delineate CTV and OARs. A training set of 195 CT scans and a test set of 28 CT scans were randomly chosen from the dataset. The mean Dice similarity coefficient (DSC), 95th percentile Hausdorff distance (95HD), and subjective evaluation were used to evaluate the performance of this strategy. Predetermined evaluation criteria were used to grade treatment plans, and percentage errors for clinical doses to the planned target volume (PTV) and OARs were calculated. The mean DSC and 95HD values of the defined CTVs were (0.84 ± 0.05) and (5.04 ± 2.15) mm, respectively. The average delineation time was < 15 s for each patient's CT scan. The overall positive rates for clinicians A and B were 53.15% versus 46.85%, and 54.05% versus 45.95%, respectively (P > .05) when CTV outlines from CUNet were blindly chosen and compared with the ground truth (GT). Furthermore, 8 test patients were randomly chosen to design the predicted plan based on the autocontouring CTVs and OARs, demonstrating acceptable agreement with the clinical plan: average absolute dose differences in mean value of D2, D50, D98, Dmax, and Dmean for PTV were within 0.74%, and average absolute volume differences in mean value of V45 and V50 for OARs were within 3.4%. Our results revealed that the CTVs and OARs for prostate cancer defined by CUNet were close to the GT. CUNet could halve the time spent by radiation oncologists in contouring, demonstrating the potential of the novel autocontouring method.

A Radiomics-Based Nomogram for the Prediction of Occult Lymph Node Metastasis in cN0 Supraglottic Carcinoma

<h3>Purpose/Objective(s)</h3> To develop a radiomics-based nomogram for predicting pathologically occult lymph node metastasis in patients with cN0 supraglottic carcinoma (SGC) to inform the optimal neck treatment. <h3>Materials/Methods</h3> The training set included 131 patients with SGC from 1/2010 to 12/2019 in our hospital, and the validation set included 108 patients from the same time in other hospital. All patents with cN0, defined by computed tomography (CT) and ultrasound or magnetic resonance imaging of the neck, had positive pathologically lymph node. The primary tumors of ROI regions were outlined in each patient's CT image by two radiotherapists. Using the least absolute shrinkage and selection operator (LASSO) method to select radiomics features. The multivariable logistic regression analysis was adopted to develop a radiomics nomogram combining radiomics score and clinical variables. The predictive performance of the model was evaluated by the calibration and decision curve analysis. <h3>Results</h3> Four radiomics features were selected to constitute radiomics signature, which was associated with LNM. Clinical T stage and pathology differentiation were also significant associated with occult metastasis, founded by multivariable logistic analysis. The multiparametric radiomics nomogram incorporate these factors and radiomics signature, which showing better prediction efficacy (C-index 0.805, 95%CI: 0.720-0.890) than radiomics score alone (C-index 0.776, 95%CI: 0.686-0.866) in training set. The area under curve yielded was 0.767 (95%CI: 0.666-0.869) in the validation set. Decision curve analysis showed our model achieved favorable discrimination and calibration. <h3>Conclusion</h3> The radiomics-based nomogram including radiomics signature, clinical T stage, and pathology differentiation has favorable individualized prediction of occult LNM in cN0 SGC patients. This nomogram could offer optimal neck dissection and prophylactic neck irradiation for cN0 SGC patients.