Adrenalectomy for the treatment of hypotension in a cat with phaeochromocytoma associated with caudal vena cava syndrome

An 11-year-old spayed female, Persian cat was referred to the Jeonbuk Animal Medical Center for evaluation of a 2-month history of lethargy and anorexia. Physical examination revealed tachycardia and hypotension. Abdominal imaging via sonography and CT identified a right adrenal gland mass causing severe deviation and compression of the caudal vena cava. After stabilising the blood pressure and heart rate through positive inotropes and fluid therapy, right adrenalectomy was performed. Surgery confirmed the adrenal gland mass was severely compressing the caudal vena cava. Histopathological examination revealed that the mass was a pheochromocytoma. After adrenalectomy, blood pressure and heart rate stabilised and remained unaffected 8 months postsurgery. This report describes a rare case of an adrenal pheochromocytoma leading to caudal vena cava compression in a cat presenting with hypotension.

Computed tomography-based automated measurement of abdominal aortic aneurysm using semantic segmentation with active learning

Accurate measurement of abdominal aortic aneurysm is essential for selecting suitable stent-grafts to avoid complications of endovascular aneurysm repair. However, the conventional image-based measurements are inaccurate and time-consuming. We introduce the automated workflow including semantic segmentation with active learning (AL) and measurement using an application programming interface of computer-aided design. 300 patients underwent CT scans, and semantic segmentation for aorta, thrombus, calcification, and vessels was performed in 60-300 cases with AL across five stages using UNETR, SwinUNETR, and nnU-Net consisted of 2D, 3D U-Net, 2D-3D U-Net ensemble, and cascaded 3D U-Net. 7 clinical landmarks were automatically measured for 96 patients. In AL stage 5, 3D U-Net achieved the highest dice similarity coefficient (DSC) with statistically significant differences (p < 0.01) except from the 2D-3D U-Net ensemble and cascade 3D U-Net. SwinUNETR excelled in 95% Hausdorff distance (HD95) with significant differences (p < 0.01) except from UNETR and 3D U-Net. DSC of aorta and calcification were saturated at stage 1 and 4, whereas thrombus and vessels were continuously improved at stage 5. The segmentation time between the manual and AL-corrected segmentation using the best model (3D U-Net) was reduced to 9.51 ± 1.02, 2.09 ± 1.06, 1.07 ± 1.10, and 1.07 ± 0.97 min for the aorta, thrombus, calcification, and vessels, respectively (p < 0.001). All measurement and tortuosity ratio measured - 1.71 ± 6.53 mm and - 0.15 ± 0.25. We developed an automated workflow with semantic segmentation and measurement, demonstrating its efficiency compared to conventional methods.

Analyzing to discover origins of CNNs and ViT architectures in medical images

In this paper, we introduce in-depth the analysis of CNNs and ViT architectures in medical images, with the goal of providing insights into subsequent research direction. In particular, the origins of deep neural networks should be explainable for medical images, but there has been a paucity of studies on such explainability in the aspect of deep neural network architectures. Therefore, we investigate the origin of model performance, which is the clue to explaining deep neural networks, focusing on the two most relevant architectures, such as CNNs and ViT. We give four analyses, including (1) robustness in a noisy environment, (2) consistency in translation invariance property, (3) visual recognition with obstructed images, and (4) acquired features from shape or texture so that we compare origins of CNNs and ViT that cause the differences of visual recognition performance. Furthermore, the discrepancies between medical and generic images are explored regarding such analyses. We discover that medical images, unlike generic ones, exhibit class-sensitive. Finally, we propose a straightforward ensemble method based on our analyses, demonstrating that our findings can help build follow-up studies. Our analysis code will be publicly available.

Machine learning models with time-series clinical features to predict radiographic progression in patients with ankylosing spondylitis

Objective

Ankylosing spondylitis (AS) is chronic inflammatory arthritis causing structural damage and radiographic progression to the spine due to repeated and continuous inflammation over a long period. This study establishes the application of machine learning models to predict radiographic progression in AS patients using time-series data from electronic medical records (EMRs).

Methods

EMR data, including baseline characteristics, laboratory findings, drug administration, and modified Stoke AS Spine Score (mSASSS), were collected from 1,123 AS patients between January 2001 and December 2018 at a single center at the time of first (T1), second (T2), and third (T3) visits. The radiographic progression of the (n+1)th visit (Pn+1=(mSASSSn+1-mSASSSn)/(Tn+1-Tn)≥1 unit per year) was predicted using follow-up visit datasets from T1 to Tn. We used three machine learning methods (logistic regression with the least absolute shrinkage and selection operation, random forest, and extreme gradient boosting algorithms) with three-fold cross-validation.

Results

The random forest model using the T1 EMR dataset best predicted the radiographic progression P2 among the machine learning models tested with a mean accuracy and area under the curves of 73.73% and 0.79, respectively. Among the T1 variables, the most important variables for predicting radiographic progression were in the order of total mSASSS, age, and alkaline phosphatase.

Conclusion

Prognosis predictive models using time-series data showed reasonable performance with clinical features of the first visit dataset when predicting radiographic progression.

Educational simulator for mastoidectomy considering mechanical properties using 3D printing and its usability evaluation

Complex temporal bone anatomy complicates operations; thus, surgeons must engage in practice to mitigate risks, improving patient safety and outcomes. However, existing training methods often involve prohibitive costs and ethical problems. Therefore, we developed an educational mastoidectomy simulator, considering mechanical properties using 3D printing. The mastoidectomy simulator was modeled on computed tomography images of a patient undergoing a mastoidectomy. Infill was modeled for each anatomical part to provide a realistic drilling sensation. Bone and other anatomies appear in assorted colors to enhance the simulator's educational utility. The mechanical properties of the simulator were evaluated by measuring the screw insertion torque for infill specimens and cadaveric temporal bones and investigating its usability with a five-point Likert-scale questionnaire completed by five otolaryngologists. The maximum insertion torque values of the sigmoid sinus, tegmen, and semicircular canal were 1.08 ± 0.62, 0.44 ± 0.42, and 1.54 ± 0.43 N mm, displaying similar-strength infill specimens of 40%, 30%, and 50%. Otolaryngologists evaluated the quality and usability at 4.25 ± 0.81 and 4.53 ± 0.62. The mastoidectomy simulator could provide realistic bone drilling feedback for educational mastoidectomy training while reinforcing skills and comprehension of anatomical structures.

Supervised representation learning based on various levels of pediatric radiographic views for transfer learning

Transfer learning plays a pivotal role in addressing the paucity of data, expediting training processes, and enhancing model performance. Nonetheless, the prevailing practice of transfer learning predominantly relies on pre-trained models designed for the natural image domain, which may not be well-suited for the medical image domain in grayscale. Recognizing the significance of leveraging transfer learning in medical research, we undertook the construction of class-balanced pediatric radiograph datasets collectively referred to as PedXnets, grounded in radiographic views using the pediatric radiographs collected over 24 years at Asan Medical Center. For PedXnets pre-training, approximately 70,000 X-ray images were utilized. Three different pre-training weights of PedXnet were constructed using Inception V3 for various radiation perspective classifications: Model-PedXnet-7C, Model-PedXnet-30C, and Model-PedXnet-68C. We validated the transferability and positive effects of transfer learning of PedXnets through pediatric downstream tasks including fracture classification and bone age assessment (BAA). The evaluation of transfer learning effects through classification and regression metrics showed superior performance of Model-PedXnets in quantitative assessments. Additionally, visual analyses confirmed that the Model-PedXnets were more focused on meaningful regions of interest.

Patient-specific, deliverable, and self-expandable surgical guide development and evaluation using 4D printing for laparoscopic partial nephrectomy

Accurate lesion diagnosis through computed tomography (CT) and advances in laparoscopic or robotic surgeries have increased partial nephrectomy survival rates. However, accurately marking the kidney resection area through the laparoscope is a prevalent challenge. Therefore, we fabricated and evaluated a 4D-printed kidney surgical guide (4DP-KSG) for laparoscopic partial nephrectomies based on CT images. The kidney phantom and 4DP-KSG were designed based on CT images from a renal cell carcinoma patient. 4DP-KSG were fabricated using shape-memory polymers. 4DP-KSG was compressed to a 10 mm thickness and restored to simulate laparoscopic port passage. The Bland-Altman evaluation assessed 4DP-KSG shape and marking accuracies before compression and after restoration with three operators. The kidney phantom's shape accuracy was 0.436 ± 0.333 mm, and the 4DP-KSG's shape accuracy was 0.818 ± 0.564 mm before compression and 0.389 ± 0.243 mm after restoration, with no significant differences. The 4DP-KSG marking accuracy was 0.952 ± 0.682 mm before compression and 0.793 ± 0.677 mm after restoration, with no statistical differences between operators (p = 0.899 and 0.992). In conclusion, our 4DP-KSG can be used for laparoscopic partial nephrectomies, providing precise and quantitative kidney tumor marking between operators before compression and after restoration.

Updated Primer on Generative Artificial Intelligence and Large Language Models in Medical Imaging for Medical Professionals

The emergence of Chat Generative Pre-trained Transformer (ChatGPT), a chatbot developed by OpenAI, has garnered interest in the application of generative artificial intelligence (AI) models in the medical field. This review summarizes different generative AI models and their potential applications in the field of medicine and explores the evolving landscape of Generative Adversarial Networks and diffusion models since the introduction of generative AI models. These models have made valuable contributions to the field of radiology. Furthermore, this review also explores the significance of synthetic data in addressing privacy concerns and augmenting data diversity and quality within the medical domain, in addition to emphasizing the role of inversion in the investigation of generative models and outlining an approach to replicate this process. We provide an overview of Large Language Models, such as GPTs and bidirectional encoder representations (BERTs), that focus on prominent representatives and discuss recent initiatives involving language-vision models in radiology, including innovative large language and vision assistant for biomedicine (LLaVa-Med), to illustrate their practical application. This comprehensive review offers insights into the wide-ranging applications of generative AI models in clinical research and emphasizes their transformative potential.

Evaluation of the accuracy of an augmented reality-based tumor-targeting guide for breast-conserving surgery

BACKGROUND AND OBJECTIVES

Although magnetic resonance imaging (MRI) is commonly used for breast tumor detection, significant challenges remain in determining and presenting the three-dimensional (3D) morphology of tumors to guide breast-conserving surgery. To address this challenge, we have developed the augmented reality-breast surgery guide (AR-BSG) and compared its performance with that of a traditional 3D-printed breast surgical guide (3DP-BSG).

METHODS

Based on the MRI results of a breast cancer patient, a breast phantom made of skin, body, and tumor was fabricated through 3D printing and silicone-casting. AR-BSG and 3DP-BSG were executed using surgical plans based on the breast phantom's computed tomography scan images. Three operators independently inserted a catheter into the phantom using each guide. Their targeting accuracy was then evaluated using Bland-Altman analysis with limits of agreement (LoA). Differences between the users of each guide were evaluated using the intraclass correlation coefficient (ICC).

RESULTS

The entry and end point errors associated with AR-BSG were -0.34±0.68 mm (LoA: -1.71-1.01 mm) and 0.81±1.88 mm (LoA: -4.60-3.00 mm), respectively, whereas 3DP-BSG was associated with entry and end point errors of -0.28±0.70 mm (LoA: -1.69-1.11 mm) and -0.62±1.24 mm (LoA: -3.00-1.80 mm), respectively. The AR-BSG's entry and end point ICC values were 0.99 and 0.97, respectively, whereas 3DP-BSG was associated with entry and end point ICC values of 0.99 and 0.99, respectively.

CONCLUSIONS

AR-BSG can consistently and accurately localize tumor margins for surgeons without inferior guiding accuracy AR-BSG can consistently and accurately localize tumor margins for surgeons without inferior guiding accuracy compared to 3DP-BSG. Additionally, when compared with 3DP-BSG, AR-BSG can offer better spatial perception and visualization, lower costs, and a shorter setup time.

Evaluation of retrieval accuracy and visual similarity in content-based image retrieval of chest CT for obstructive lung disease

The aim of our study was to assess the performance of content-based image retrieval (CBIR) for similar chest computed tomography (CT) in obstructive lung disease. This retrospective study included patients with obstructive lung disease who underwent volumetric chest CT scans. The CBIR database included 600 chest CT scans from 541 patients. To assess the system performance, follow-up chest CT scans of 50 patients were evaluated as query cases, which showed the stability of the CT findings between baseline and follow-up chest CT, as confirmed by thoracic radiologists. The CBIR system retrieved the top five similar CT scans for each query case from the database by quantifying and comparing emphysema extent and size, airway wall thickness, and peripheral pulmonary vasculatures in descending order from the database. The rates of retrieval of the same pairs of query CT scans in the top 1-5 retrievals were assessed. Two expert chest radiologists evaluated the visual similarities between the query and retrieved CT scans using a five-point scale grading system. The rates of retrieving the same pairs of query CTs were 60.0% (30/50) and 68.0% (34/50) for top-three and top-five retrievals. Radiologists rated 64.8% (95% confidence interval 58.8-70.4) of the retrieved CT scans with a visual similarity score of four or five and at least one case scored five points in 74% (74/100) of all query cases. The proposed CBIR system for obstructive lung disease integrating quantitative CT measures demonstrated potential for retrieving chest CT scans with similar imaging phenotypes. Further refinement and validation in this field would be valuable.

Approximating Intermediate Feature Maps of Self-Supervised Convolution Neural Network to Learn Hard Positive Representations in Chest Radiography

Recent advances in contrastive learning have significantly improved the performance of deep learning models. In contrastive learning of medical images, dealing with positive representation is sometimes difficult because some strong augmentation techniques can disrupt contrastive learning owing to the subtle differences between other standardized CXRs compared to augmented positive pairs; therefore, additional efforts are required. In this study, we propose intermediate feature approximation (IFA) loss, which improves the performance of contrastive convolutional neural networks by focusing more on positive representations of CXRs without additional augmentations. The IFA loss encourages the feature maps of a query image and its positive pair to resemble each other by maximizing the cosine similarity between the intermediate feature outputs of the original data and the positive pairs. Therefore, we used the InfoNCE loss, which is commonly used loss to address negative representations, and the IFA loss, which addresses positive representations, together to improve the contrastive network. We evaluated the performance of the network using various downstream tasks, including classification, object detection, and a generative adversarial network (GAN) inversion task. The downstream task results demonstrated that IFA loss can improve the performance of effectively overcoming data imbalance and data scarcity; furthermore, it can serve as a perceptual loss encoder for GAN inversion. In addition, we have made our model publicly available to facilitate access and encourage further research and collaboration in the field.

Accuracy of artificial intelligence-assisted growth prediction in skeletal Class I preadolescent patients using serial lateral cephalograms for a 2-year growth interval

OBJECTIVE

To investigate the accuracy of artificial intelligence-assisted growth prediction using a convolutional neural network (CNN) algorithm and longitudinal lateral cephalograms (Lat-cephs).

MATERIALS AND METHODS

A total of 198 Japanese preadolescent children, who had skeletal Class I malocclusion and whose Lat-cephs were available at age 8 years (T0) and 10 years (T1), were allocated into the training, validation, and test phases (n = 161, n = 17, n = 20). Orthodontists and the CNN model identified 28 hard-tissue landmarks (HTL) and 19 soft-tissue landmarks (STL). The mean prediction error values were defined as 'excellent,' 'very good,' 'good,' 'acceptable,' and 'unsatisfactory' (criteria: 0.5 mm, 1.0 mm, 1.5 mm, and 2.0 mm, respectively). The degree of accurate prediction percentage (APP) was defined as 'very high,' 'high,' 'medium,' and 'low' (criteria: 90%, 70%, and 50%, respectively) according to the percentage of subjects that showed the error range within 1.5 mm.

RESULTS

All HTLs showed acceptable-to-excellent mean PE values, while the STLs Pog', Gn', and Me' showed unsatisfactory values, and the rest showed good-to-acceptable values. Regarding the degree of APP, HTLs Ba, ramus posterior, Pm, Pog, B-point, Me, and mandibular first molar root apex exhibited low APPs. The STLs labrale superius, lower embrasure, lower lip, point of lower profile, B', Pog,' Gn' and Me' also exhibited low APPs. The remainder of HTLs and STLs showed medium-to-very high APPs.

CONCLUSION

Despite the possibility of using the CNN model to predict growth, further studies are needed to improve the prediction accuracy in HTLs and STLs of the chin area.

Accuracy and clinical validity of automated cephalometric analysis using convolutional neural networks

BACKGROUND

This study aimed to assess the error range of cephalometric measurements based on the landmarks detected using cascaded CNNs and determine how horizontal and vertical positional errors of individual landmarks affect lateral cephalometric measurements.

METHODS

In total, 120 lateral cephalograms were obtained consecutively from patients (mean age, 32.5 ± 11.6) who visited the Asan Medical Center, Seoul, Korea, for orthodontic treatment between 2019 and 2021. An automated lateral cephalometric analysis model previously developed from a nationwide multi-centre database was used to digitize the lateral cephalograms. The horizontal and vertical landmark position error attributable to the AI model was defined as the distance between the landmark identified by the human and that identified by the AI model on the x- and y-axes. The differences between the cephalometric measurements based on the landmarks identified by the AI model vs those identified by the human examiner were assessed. The association between the lateral cephalometric measurements and the positioning errors in the landmarks comprising the cephalometric measurement was assessed.

RESULTS

The mean difference in the angular and linear measurements based on AI vs human landmark localization was .99 ± 1.05°, and .80 ± .82 mm, respectively. Significant differences between the measurements derived from AI-based and human localization were observed for all cephalometric variables except SNA, pog-Nperp, facial angle, SN-GoGn, FMA, Bjork sum, U1-SN, U1-FH, IMPA, L1-NB (angular) and interincisal angle.

CONCLUSIONS

The errors in landmark positions, especially those that define reference planes, may significantly affect cephalometric measurements. The possibility of errors generated by automated lateral cephalometric analysis systems should be considered when using such systems for orthodontic diagnoses.

Accuracy of posteroanterior cephalogram landmarks and measurements identification using a cascaded convolutional neural network algorithm: A multicenter study

Objective

: To quantify the effects of midline-related landmark identification on midline deviation measurements in posteroanterior (PA) cephalograms using a cascaded convolutional neural network (CNN).

Methods

: A total of 2,903 PA cephalogram images obtained from 9 university hospitals were divided into training, internal validation, and test sets (n = 2,150, 376, and 377). As the gold standard, 2 orthodontic professors marked the bilateral landmarks, including the frontozygomatic suture point and latero-orbitale (LO), and the midline landmarks, including the crista galli, anterior nasal spine (ANS), upper dental midpoint (UDM), lower dental midpoint (LDM), and menton (Me). For the test, Examiner-1 and Examiner-2 (3-year and 1-year orthodontic residents) and the Cascaded-CNN models marked the landmarks. After point-to-point errors of landmark identification, the successful detection rate (SDR) and distance and direction of the midline landmark deviation from the midsagittal line (ANS-mid, UDM-mid, LDM-mid, and Me-mid) were measured, and statistical analysis was performed.

Results

: The cascaded-CNN algorithm showed a clinically acceptable level of point-to-point error (1.26 mm vs. 1.57 mm in Examiner-1 and 1.75 mm in Examiner-2). The average SDR within the 2 mm range was 83.2%, with high accuracy at the LO (right, 96.9%; left, 97.1%), and UDM (96.9%). The absolute measurement errors were less than 1 mm for ANS-mid, UDM-mid, and LDM-mid compared with the gold standard.

Conclusions

: The cascaded-CNN model may be considered an effective tool for the auto-identification of midline landmarks and quantification of midline deviation in PA cephalograms of adult patients, regardless of variations in the image acquisition method.

Generative AI in glioma: ensuring diversity in training image phenotypes to improve diagnostic performance for IDH mutation prediction

BACKGROUND

This study evaluated whether generative artificial intelligence-based augmentation (GAA) can provide diverse and realistic imaging phenotypes and improve deep learning-based classification of isocitrate dehydrogenase (IDH) type in glioma compared with neuroradiologists.

METHODS

For model development, 565 patients (346 IDH-wildtype, 219 IDH-mutant) with paired contrast-enhanced T1 and FLAIR MRI scans were collected from tertiary hospital and The Cancer Imaging Archive. Performance was tested on internal (119, 78 IDH-wildtype, 41 IDH-mutant [IDH1 and 2]) and external test sets (108, 72 IDH-wildtype, 36 IDH-mutant). GAA was developed using score-based diffusion model and ResNet50 classifier. The optimal GAA was selected in comparison with null model. Two neuroradiologists (R1, R2) assessed realism, diversity of imaging phenotypes, and predicted IDH mutation. The performance of a classifier trained with optimal GAA was compared with that of neuroradiologists using area under the receiver operating characteristics curve (AUC). The effect of tumor size and contrast enhancement on GAA performance was tested.

RESULTS

Generated images demonstrated realism (Turing's test: 47.5%-50.5%) and diversity indicating IDH type. Optimal GAA was achieved with augmentation with 110 000 generated slices (AUC: 0.938). The classifier trained with optimal GAA demonstrated significantly higher AUC values than neuroradiologists in both the internal (R1, P=.003; R2, P<.001) and external test sets (R1, P<.01; R2, P<.001). GAA with large-sized tumors or predominant enhancement showed comparable performance to optimal GAA (internal test: AUC 0.956 and 0.922; external test: 0.810 and 0.749).

CONCLUSIONS

Application of generative AI with realistic and diverse images provided better diagnostic performance than neuroradiologists for predicting IDH type in glioma.

Ensemble Deep Learning Model to Predict Lymphovascular Invasion in Gastric Cancer

Lymphovascular invasion (LVI) is one of the most important prognostic factors in gastric cancer as it indicates a higher likelihood of lymph node metastasis and poorer overall outcome for the patient. Despite its importance, the detection of LVI(+) in histopathology specimens of gastric cancer can be a challenging task for pathologists as invasion can be subtle and difficult to discern. Herein, we propose a deep learning-based LVI(+) detection method using H&E-stained whole-slide images. The ConViT model showed the best performance in terms of both AUROC and AURPC among the classification models (AUROC: 0.9796; AUPRC: 0.9648). The AUROC and AUPRC of YOLOX computed based on the augmented patch-level confidence score were slightly lower (AUROC: -0.0094; AUPRC: -0.0225) than those of the ConViT classification model. With weighted averaging of the patch-level confidence scores, the ensemble model exhibited the best AUROC, AUPRC, and F1 scores of 0.9880, 0.9769, and 0.9280, respectively. The proposed model is expected to contribute to precision medicine by potentially saving examination-related time and labor and reducing disagreements among pathologists.

Machine learning-based clinical decision support system for treatment recommendation and overall survival prediction of hepatocellular carcinoma: a multi-center study

The treatment decisions for patients with hepatocellular carcinoma are determined by a wide range of factors, and there is a significant difference between the recommendations of widely used staging systems and the actual initial treatment choices. Herein, we propose a machine learning-based clinical decision support system suitable for use in multi-center settings. We collected data from nine institutions in South Korea for training and validation datasets. The internal and external datasets included 935 and 1750 patients, respectively. We developed a model with 20 clinical variables consisting of two stages: the first stage which recommends initial treatment using an ensemble voting machine, and the second stage, which predicts post-treatment survival using a random survival forest algorithm. We derived the first and second treatment options from the results with the highest and the second-highest probabilities given by the ensemble model and predicted their post-treatment survival. When only the first treatment option was accepted, the mean accuracy of treatment recommendation in the internal and external datasets was 67.27% and 55.34%, respectively. The accuracy increased to 87.27% and 86.06%, respectively, when the second option was included as the correct answer. Harrell's C index, integrated time-dependent AUC curve, and integrated Brier score of survival prediction in the internal and external datasets were 0.8381 and 0.7767, 91.89 and 86.48, 0.12, and 0.14, respectively. The proposed system can assist physicians by providing data-driven predictions for reference from other larger institutions or other physicians within the same institution when making treatment decisions.

Orthognathic surgical planning using graph CNN with dual embedding module: External validations with multi-hospital datasets

BACKGROUND AND OBJECTIVE

Despite recent development of AI, prediction of the surgical movement in the maxilla and mandible by OGS might be more difficult than that of tooth movement by orthodontic treatment. To evaluate the prediction accuracy of the surgical movement using pairs of pre-(T0) and post-surgical (T1) lateral cephalograms (lat-ceph) of orthognathic surgery (OGS) patients and dual embedding module-graph convolution neural network (DEM-GCNN) model.

METHODS

599 pairs from 3 institutions were used as training, internal validation, and internal test sets and 201 pairs from other 6 institutions were used as external test set. DEM-GCNN model (IEM, learning the lat-ceph images; LTEM, learning the landmarks) was developed to predict the amount and direction of surgical movement of ANS and PNS in the maxilla and B-point and Md1crown in the mandible. The distance between T1 landmark coordinates actually moved by OGS (ground truth) and predicted by DEM-GCNN model and pre-existed CNN-based Model-C (learning the lat-ceph images) was compared.

RESULTS

In both internal and external tests, DEM-GCNN did not exhibit significant difference from ground truth in all landmarks (ANS, PNS, B-point, Md1crown, all P > 0.05). When the accumulated successful detection rate for each landmark was compared, DEM-GCNN showed higher values than Model-C in both the internal and external tests. In violin plots exhibiting the error distribution of the prediction results, both internal and external tests showed that DEM-GCNN had significant performance improvement in PNS, ANS, B-point, Md1crown than Model-C. DEM-GCNN showed significantly lower prediction error values than Model-C (one-jaw surgery, B-point, Md1crown, all P < 0.005; two-jaw surgery, PNS, ANS, all P < 0.05; B point, Md1crown, all P < 0.005).

CONCLUSION

We developed a robust OGS planning model with maximized generalizability despite diverse qualities of lat-cephs from 9 institutions.

Comparing the biomechanical properties of conventional suture and all-suture anchors using patient-specific and realistic osteoporotic and non-osteoporotic phantom using 3D printing

Conventional suture anchors (CAs) and all-suture anchors (ASAs) are used for rotator cuff repair. Pull-out strength (POS) is an important factor that affects surgical outcomes. While the fixation mechanism differs between the anchor types and relies on the quality, few studies have compared biomechanical properties of anchors based on bone quality. This study aimed to compare the biomechanical properties of anchors using osteoporotic bone (OB) and non-osteoporotic bone (NOB) simulators. Humerus simulators were fabricated using fused deposition modeling of 3D printing and acrylonitrile butadiene styrene adjusting the thickness of cortical bone and density of cancellous bone based on CT images. Cyclic loading from 10 to 50 N, 10 to 100 N, and 10 to 150 N for 10 cycles was clinically determined at each anchor because the supraspinatus generates a force of 67-125 N in daily activities of normal control. After cyclic loading, the anchor was extruded at a load of 5 mm/min. Displacement, POS, and stiffness were measured. In OB simulators, CAs revealed bigger gap displacement than ASAs with cyclic loading of 10-150 N. ASA showed higher values for POS and stiffness. In NOB simulators, ASAs revealed bigger gap displacement than CAs with cyclic loading of 10-150 N. ASA showed higher values for POS and CA showed higher values for stiffness. POS of anchors depends on anchors 'displacement and bone stiffness. In conclusion, ASA demonstrated better biomechanical performance than CA in terms of stability under cyclic loading and stiffness with similar POS in OB.

Spatiotemporal association between air pollution and stroke mortality in South Korea

BACKGROUND

Air pollutant concentrations in South Korea vary greatly by region and time. To assess temporal and spatial associations of stroke subtypes with long-term air pollution effects on stroke mortality, we studied ischemic stroke (IS), intracerebral hemorrhage (ICH), and subarachnoid hemorrhage (SAH).

METHODS

This was an observational study conducted in South Korea from 2001-2018. Concentrations of carbon monoxide (CO), nitrogen dioxide (NO2), sulfur dioxide (SO2), and particulate matter ≤10 µm in diameter (PM10) were determined from 332 stations. Average air pollutant concentrations in each district were determined by distance-weighted linear interpolation. The nationwide stroke mortality rates in 249 districts were obtained from the Korean Statistical Information Service. Time intervals were divided into three consecutive 6-year periods: 2001-2006, 2007-2012, and 2013-2018.

RESULTS

The concentrations of air pollutants gradually decreased from 2001-2018, along with decreases in IS and ICH mortality rates. However, mortality rates associated with SAH remained constant. From 2001-2006, NO2 (adjusted odds ratio [aOR]:1.13, 95% confidence interval: 1.08-1.19), SO2 (aOR: 1.10, 1.07-1.13), and PM10 (aOR: 1.12, 1.06-1.18) concentrations were associated with IS mortality, and SO2 (aOR: 1.07, 1.02-1.13) and PM10 (aOR:1.11,1.06-1.22) concentrations were associated with SAH-associated mortality. Air pollution was no longer associated with stroke mortality from 2007 onward, as the air pollution concentration continued to decline. Throughout the entire 18-year period, ICH-associated mortality was not associated with air pollution.

CONCLUSIONS

Considering temporal and spatial trends, high concentrations of air pollutants were most likely to be associated with IS mortality. Our results strengthen the existing evidence of the deleterious effects of air pollution on IS mortality.

Overcoming the Challenges in the Development and Implementation of Artificial Intelligence in Radiology: A Comprehensive Review of Solutions Beyond Supervised Learning

Artificial intelligence (AI) in radiology is a rapidly developing field with several prospective clinical studies demonstrating its benefits in clinical practice. In 2022, the Korean Society of Radiology held a forum to discuss the challenges and drawbacks in AI development and implementation. Various barriers hinder the successful application and widespread adoption of AI in radiology, such as limited annotated data, data privacy and security, data heterogeneity, imbalanced data, model interpretability, overfitting, and integration with clinical workflows. In this review, some of the various possible solutions to these challenges are presented and discussed; these include training with longitudinal and multimodal datasets, dense training with multitask learning and multimodal learning, self-supervised contrastive learning, various image modifications and syntheses using generative models, explainable AI, causal learning, federated learning with large data models, and digital twins.

Screening obstructive sleep apnea patients via deep learning of knowledge distillation in the lateral cephalogram

The lateral cephalogram in orthodontics is a valuable screening tool on undetected obstructive sleep apnea (OSA), which can lead to consequences of severe systematic disease. We hypothesized that a deep learning-based classifier might be able to differentiate OSA as anatomical features in lateral cephalogram. Moreover, since the imaging devices used by each hospital could be different, there is a need to overcome modality difference of radiography. Therefore, we proposed a deep learning model with knowledge distillation to classify patients into OSA and non-OSA groups using the lateral cephalogram and to overcome modality differences simultaneously. Lateral cephalograms of 500 OSA patients and 498 non-OSA patients from two different devices were included. ResNet-50 and ResNet-50 with a feature-based knowledge distillation models were trained and their performances of classification were compared. Through the knowledge distillation, area under receiver operating characteristic curve analysis and gradient-weighted class activation mapping of knowledge distillation model exhibits high performance without being deceived by features caused by modality differences. By checking the probability values predicting OSA, an improvement in overcoming the modality differences was observed, which could be applied in the actual clinical situation.

The prediction of sagittal chin point relapse following two-jaw surgery using machine learning

The study aimed to identify critical factors associated with the surgical stability of pogonion (Pog) by applying machine learning (ML) to predict relapse following two-jaw orthognathic surgery (2 J-OGJ). The sample set comprised 227 patients (110 males and 117 females, 207 training and 20 test sets). Using lateral cephalograms taken at the initial evaluation (T0), pretreatment (T1), after (T2) 2 J-OGS, and post treatment (T3), 55 linear and angular skeletal and dental surgical movements (T2-T1) were measured. Six ML modes were utilized, including classification and regression trees (CART), conditional inference tree (CTREE), and random forest (RF). The training samples were classified into three groups; highly significant (HS) (≥ 4), significant (S) (≥ 2 and < 4), and insignificant (N), depending on Pog relapse. RF indicated that the most important variable that affected relapse rank prediction was ramus inclination (RI), CTREE and CART revealed that a clockwise rotation of more than 3.7 and 1.8 degrees of RI was a risk factor for HS and S groups, respectively. RF, CTREE, and CART were practical tools for predicting surgical stability. More than 1.8 degrees of CW rotation of the ramus during surgery would lead to significant Pog relapse.

Enhancing the performance of premature ventricular contraction detection in unseen datasets through deep learning with denoise and contrast attention module

Premature ventricular contraction (PVC) is a common and harmless cardiac arrhythmia that can be asymptomatic or cause palpitations and chest pain in rare instances. However, frequent PVCs can lead to more serious arrhythmias, such as atrial fibrillation. Several PVC detection models have been proposed to enable early diagnosis of arrhythmias; however, they lack reliability and generalizability due to the variability of electrocardiograms across different settings and noise levels. Such weaknesses are known to aggravate with new data. Therefore, we present a deep learning model with a novel attention mechanism that can detect PVC accurately, even on unseen electrocardiograms with various noise levels. Our method, called the Denoise and Contrast Attention Module (DCAM), is a two-step process that denoises signals with a convolutional neural network (CNN) in the frequency domain and attends to differences. It focuses on differences in the morphologies and intervals of the remaining beats, mimicking how trained clinicians identify PVCs. Using three different encoder types, we evaluated 1D U-Net with DCAM on six external test datasets. The results showed that DCAM significantly improved the F1-score of PVC detection performance on all six external datasets and enhanced the performance of balancing both the sensitivity and precision of the models, demonstrating its robustness and generalization ability regardless of the encoder type. This demonstrates the need for a trainable denoising process before applying the attention mechanism. Our DCAM could contribute to the development of a reliable algorithm for cardiac arrhythmia detection under real clinical electrocardiograms.

Enhancement of Non-Linear Deep Learning Model by Adjusting Confounding Variables for Bone Age Estimation in Pediatric Hand X-rays

In medicine, confounding variables in a generalized linear model are often adjusted; however, these variables have not yet been exploited in a non-linear deep learning model. Sex plays important role in bone age estimation, and non-linear deep learning model reported their performances comparable to human experts. Therefore, we investigate the properties of using confounding variables in a non-linear deep learning model for bone age estimation in pediatric hand X-rays. The RSNA Pediatric Bone Age Challenge (2017) dataset is used to train deep learning models. The RSNA test dataset is used for internal validation, and 227 pediatric hand X-ray images with bone age, chronological age, and sex information from Asan Medical Center (AMC) for external validation. U-Net based autoencoder, U-Net multi-task learning (MTL), and auxiliary-accelerated MTL (AA-MTL) models are chosen. Bone age estimations adjusted by input, output prediction, and without adjusting the confounding variables are compared. Additionally, ablation studies for model size, auxiliary task hierarchy, and multiple tasks are conducted. Correlation and Bland-Altman plots between ground truth and model-predicted bone ages are evaluated. Averaged saliency maps based on image registration are superimposed on representative images according to puberty stage. In the RSNA test dataset, adjusting by input shows the best performances regardless of model size, with mean average errors (MAEs) of 5.740, 5.478, and 5.434 months for the U-Net backbone, U-Net MTL, and AA-MTL models, respectively. However, in the AMC dataset, the AA-MTL model that adjusts the confounding variable by prediction shows the best performance with an MAE of 8.190 months, whereas the other models show the best performances by adjusting the confounding variables by input. Ablation studies of task hierarchy reveal no significant differences in the results of the RSNA dataset. However, predicting the confounding variable in the second encoder layer and estimating bone age in the bottleneck layer shows the best performance in the AMC dataset. Ablations studies of multiple tasks reveal that leveraging confounding variables plays an important role regardless of multiple tasks. To estimate bone age in pediatric X-rays, the clinical setting and balance between model size, task hierarchy, and confounding adjustment method play important roles in performance and generalizability; therefore, proper adjusting methods of confounding variables to train deep learning-based models are required for improved models.

Deep Learning for Automated Triaging of Stable Chest Radiographs in a Follow-up Setting

Background Most artificial intelligence algorithms that interpret chest radiographs are restricted to an image from a single time point. However, in clinical practice, multiple radiographs are used for longitudinal follow-up, especially in intensive care units (ICUs). Purpose To develop and validate a deep learning algorithm using thoracic cage registration and subtraction to triage pairs of chest radiographs showing no change by using longitudinal follow-up data. Materials and Methods A deep learning algorithm was retrospectively developed using baseline and follow-up chest radiographs in adults from January 2011 to December 2018 at a tertiary referral hospital. Two thoracic radiologists reviewed randomly selected pairs of "change" and "no change" images to establish the ground truth, including normal or abnormal status. Algorithm performance was evaluated using area under the receiver operating characteristic curve (AUC) analysis in a validation set and temporally separated internal test sets (January 2019 to August 2021) from the emergency department (ED) and ICU. Threshold calibration for the test sets was conducted, and performance with 40% and 60% triage thresholds was assessed. Results This study included 3 304 996 chest radiographs in 329 036 patients (mean age, 59 years ± 14 [SD]; 170 433 male patients). The training set included 550 779 pairs of radiographs. The validation set included 1620 pairs (810 no change, 810 change). The test sets included 533 pairs (ED; 265 no change, 268 change) and 600 pairs (ICU; 310 no change, 290 change). The algorithm had AUCs of 0.77 (validation), 0.80 (ED), and 0.80 (ICU). With a 40% triage threshold, specificity was 88.4% (237 of 268 pairs) and 90.0% (261 of 290 pairs) in the ED and ICU, respectively. With a 60% triage threshold, specificity was 79.9% (214 of 268 pairs) and 79.3% (230 of 290 pairs) in the ED and ICU, respectively. For urgent findings (consolidation, pleural effusion, pneumothorax), specificity was 78.6%-100% (ED) and 85.5%-93.9% (ICU) with a 40% triage threshold. Conclusion The deep learning algorithm could triage pairs of chest radiographs showing no change while detecting urgent interval changes during longitudinal follow-up. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Czum in this issue.

Racial Difference in Outcomes in Breast Cancer Patients with Residual Nodal Disease after Neoadjuvant Chemotherapy

PURPOSE/OBJECTIVE(S)

African Americans (AA) requiring neoadjuvant chemotherapy (NAC) have been associated with worse outcomes. Residual nodal disease (ypN+) after NAC represents a highly unfavorable risk factor. We hypothesized that even within this unfavorable subgroup, that racial differences in outcome would persist.

MATERIALS/METHODS

An IRB-approved retrospective review of breast cancer (BC) patients in a multi-institutional study was performed between 2005-2018 to identify ypN+ patients (excluding metastatic or inflammatory BC). Clinico-pathologic parameters stratified by race were collected and analyzed. For molecular subtype analyses, patients were stratified into triple negative (TN), hormone receptor (HR)+/HER2-, and HR+/HER2+, and HR-/HER2+ subtypes. Overall survival (OS), disease free survival (DFS) and recurrence outcomes were obtained, and univariate and multivariate (MVA) logistic regression models were constructed and analyzed.

RESULTS

Among 404 ypN+ patients, 107 (26%) were AA, and 297 (74%) were non-AA. Median follow-up for the non-AA group was 3.8 years (y) (IQR 2.4-6.3) and 3.5y (IQR 2.0-6.2) for the AA group. Clinical and pathologic patient characteristics (age, molecular subtypes, BRCA status, histology, grade, smoking status, primary surgery type, axillary/reconstruction surgery rates, margin status, stage) were without significant statistical differences between the non-AA and AA group, except the non-AA group had proportionally more cN3 disease (10.5% vs. 5.1%; p = .01). Despite this, AA demonstrated worse OS and DFS outcomes (Table). AA also had significantly worse local (15% vs. 6.7%, p = .02), regional (11.2% vs. 5.1%, p = .05) and distant recurrences (32.7% vs. 22.6%, p = .05) compared to non-AA. On MVA for OS and DFS, HR+ status, clinical stage, and AA race (HR 2.1 (CI 1.3-3.4), p = .004 and HR 1.7 (CI 1.1-2.6), p = .01 respectively) remained significant. Molecular subtype analysis demonstrated that AA with HR+/HER2- but not the TN subtypes demonstrated significantly worse outcomes (Table). Utilization of endocrine therapy was not different between AA and non-AA patients (94% vs. 97%, p = 0.3) to explain this discrepancy. Worse outcomes in HER2 subtype for AA group was suggested but could not be statistically verified due to insufficient sample size. There was no discernible difference in chemotherapy and radiation therapy regimen or compliance between the AA and non-AA groups.

CONCLUSION

AA patients who fail to achieve nodal clearance with NAC had higher local, regional and distant recurrence, and worse survival compared to non-AA, particularly those with non-TN status. These differences could not be readily explained by therapeutic disparity, or compliance. These hypothesis generating findings suggest need to explore biological implications, and alternative therapeutic strategies for this unfavorable subgroup.

Multi-Institutional Phase II Trial Using Dose Escalated Five Fraction Stereotactic Partial Breast Irradiation (S-PBI) with GammaPodTM for Early-Stage Breast Cancer

PURPOSE/OBJECTIVE(S)

We report on our early experience of a multi-institutional phase II study of dose escalated five fraction stereotactic partial breast irradiation (S-PBI) for early-stage breast cancer after partial mastectomy using the GammaPodTM stereotactic radiation system.

MATERIALS/METHODS

Patient eligibility included DCIS or invasive epithelial histologies, AJCC clinical stage 0, I, or II with tumor size < 3 cm, and negative margins. Prior safety of Phase I dose escalation has been reported. Dose was 40 Gy delivered in 5 fractions to the CTV, and minimum dose 30 Gy in 5 fractions to the PTV. CTV margin was 1 cm and PTV margin 3 mm. For PTV cavities larger than 100cc, dose was reduced to 35Gy in 5 fractions to the CTV and 30 Gy in 5 fractions to the PTV. Primary endpoint of the study is to determine the 3-year patient global cosmesis score (4-point scale excellent, good, fair, or poor) and adverse cosmesis using a dose escalated approach with smaller PTV margins than conventional methods. Both patients and physicians completed baseline and subsequent cosmesis outcome questionnaires. Treatment related toxicity was graded using the NCI version 4.0 and RTOG/EORTC late radiation scale.

RESULTS

From 3/2019-10/2021, 74 patients were treated respectively. Of these, 38 were treated to 40Gy and 36 were treated to 35 Gy. Median follow up (f/u) was 24 months (mo), range (r) 3-39mo. Median age was 63 years (r 43-77). Histology included 28 DCIS, and 46 invasive carcinomas. 45/46 invasive tumors were ER+. 60/74 (81%) patients received endocrine therapy, and 7/74 patient received chemotherapy. There were 221 acute grade 1 toxicities, and 28 Grade 2 toxicities. No grade 3 or higher acute toxicities were reported (< 90 days). The most common Grade 2 toxicities were radiation dermatitis (10), breast pain (8), blister (4), skin infection (2), nipple discharge (2), and fatigue (2). In the late period, there were 54 Grade 1 late toxicities, 4 Grade 2 late toxicities, and no Grade 3 or higher late toxicities. Grade 2 toxicities included fibrosis (2), and pain (2). Two patients developed grade 1 asymptomatic nonpalpable fat necrosis both diagnosed at 12 months after radiation treatments. The most common grade 1 late toxicities were breast pain (14), hyperpigmentation (8), fibrosis (10), and fatigue (5). Physicians scored cosmesis excellent or good 70/73 (95.8%), 58/60 (96.7%), 36/36 (100%),17/17(100%) respectively at baseline, 12 months, 24 months, and 36months post SBRT, while patients scored the same periods 62/71 (83.7%), 53/59 (89.8%), 33/36 (91.6%), 17/18 (94.4%). There have been no reports of disease recurrences.

CONCLUSION

Results at 24-month median follow-up, of our dose escalated stereotactic partial breast 5 fraction regimen, has low acute and late toxicity, while maintaining high proportion of excellent/good cosmetic outcomes. Continued analysis of all cohorts is in progress.

CLINICAL TRIALS

gov identifier is NCT03581136.

Assessment of Cardiac Radiation Dose in the Co-60 Prone Based Stereotactic Partial Breast Irradiation (CP-sPBI) Using the Distance from the Heart to the Planning Treatment Volume as a Surrogate Marker

PURPOSE/OBJECTIVE(S)

Irradiation of the breast has shown to provide sharp dose gradients using Co-60 prone based stereotactic partial breast irradiation (CP-sPBI), a contemporary device for stereotactic radiotherapy for breast cancer (BC) for accelerated partial breast irradiation (APBI). In addition, the precise setup of CP-sPBI permits a small planning treatment volume (PTV) margin of 3 mm creating a greater distance from PTV to organs at risk. However, to date the factors that influence dose gradients and subsequent cardiac doses of ionizing radiation using CP-sPBI have not been well-studied. Here we evaluate distance of the heart to the lumpectomy PTV cavity and how this effects cardiac dose.

MATERIALS/METHODS

A retrospective database of 113 consecutive patients treated by CP-sPBI for APBI from March 2019 to February 2023 who were treated with 30 Gy in 5 fractions were queried for analysis. The minimum distance from the heart to the PTV (hP) was measured in either the axial or sagittal view. A group of 28 patient cases were randomly selected to achieve an even distribution of 28 cases with hP < 2.75 cm and hP ≥ 2.75 cm to compare cardiac toxicities based on hP. Descriptive analyses were performed to evaluate various cardiac dosimetric parameters based on laterality of BC and hP, using the student's t test.

RESULTS

The mean (range) hP was 4.58 cm (0.80-12.23) for all cases. The subgroup analyses of 28 patient cases with cardiac parameters showed the heart mean (range) dose of 1.20 Gy (0.01-2.11). The mean and max heart dose to the left-sided BC were similar to those to the right-sided BC (mean dose: 1.20 vs. 1.19 Gy; P = 0.97 and max dose: 10.47 vs. 5.66 Gy; P = 0.06). An inverse correlation between hP and mean heart dose was shown with the correlation coefficient of -0.81. Using a cutoff of 2.75 cm hP, the differences between hP < 2.75 and hP ≥ 2.75 cm for all cardiac dosimetric evaluations were all statistically significant, including mean (1.67 vs. 0.79 Gy; p<0.01) and maximal heart dose (14.48 vs. 4.11 Gy; p<0.01) CONCLUSION: CP-sPBI treatment delivery system was able to achieve acceptable clinically relevant heart dosimetric parameters when delivering 5 fraction APBI with a mean heart dose of 1.20 Gy for all locations of PTV cavity volume in the breast. Due to CP-sPBIs excellent dose fall-off characteristics, APBI using CP-SPBI showed clinically acceptable cardiac dosimetric parameters, particularly for PTVs located > 2.75 cm from the heart.

Risk of Sexual Dysfunction in Men Treated with Pelvic Radiation Therapy for Locally Advanced Rectal Cancer: 20 Years of Experience with 451 Patients

PURPOSE/OBJECTIVE(S)

Radiation therapy (RT) is commonly used in the treatment of locally advanced rectal cancer (LARC), but data on its impact on men's sexual health is limited. Given the rising incidence of rectal cancer in younger men, sexual function is an important quality of life factor. We hypothesized that men with LARC treated with RT would be at increased risk of sexual dysfunction compared to men who did not receive RT.

MATERIALS/METHODS

This is a single institution retrospective analysis of outcomes of men ≤50 years diagnosed with LARC between 1999 and 2019. Primary outcomes of erectile dysfunction (ED), ejaculatory dysfunction (EjD), and testosterone deficiency (TD) were assessed via ICD-9/10 codes, and TD was captured with free testosterone <300 ng/dL. Cumulative incidences were calculated with death as a competing risk and p values were calculated using Gray's test. Subdistribution hazard ratios from competing risk regression models were used.

RESULTS

The combined study sample included 451 men: 347 received RT as part of their multimodality treatment, and 104 did not. Median time to last follow up was 5.6 years (IQR 3.3-8.7). Age at diagnosis, stage, and medical comorbidities for sexual dysfunction were similar between the two groups (p>0.05). Cumulative incidence estimates are shown in Table 1, showing a higher cumulative incidence of ED in the RT group, but no difference in EjD or TD between the 2 groups. On univariable analysis, RT, smoking, dyslipidemia, peripheral artery disease, depression, prostate cancer/hyperplasia, closed or current ileostomy, and undergoing rectal cancer surgery were all independent risk factors for ED (p<0.05). On multivariable analysis, RT maintained statistical significance as an independent risk factor for ED (HR 3.87, 95% CI 1.93-7.75, p<0.001). Within the RT group, IMRT compared to 3D (HR 1.54, 95% CI 1.02-2.32, p = 0.040) and groin RT (HR 2.60, 95% CI 1.21-5.59, p = 0.014) were independent risk factors for ED. Within the RT group, groin RT also approached significance as a risk factor for TD (HR 3.61, 95% CI 0.98-13.3, p = 0.054). No RT dose thresholds to external genitals or penile bulb were identified that increased risks of ED, EjD, or TD.

CONCLUSION

RT for LARC independently increases risk of ED but not EjD or TD. IMRT might increase the risk of ED due to increased scatter dose to the genitals and including the inguinal nodes in the target volumes increases the dose to the genitals/testicles, which could translate into a higher risk for ED and TD. Future research on proton RT and prophylactic sildenafil is needed in men ≤50 to decrease the risk of ED.

Financial Toxicity and Patient Experience Outcomes on a Multi-Institutional Phase I Single Fraction Stereotactic Partial Breast Irradiation Protocol for Early-Stage Breast Cancer

PURPOSE/OBJECTIVE(S)

Given the demonstrated financial toxicity (FT) of radiation treatment on breast cancer patients shown in both conventional and our recent 5 fraction stereotactic APBI (S-PBI) study, we assessed the FT, as well as patient-reported utility, quality-of-life and patient experience measures, on patients treated in our phase I single fraction S-PBI trial.

MATERIALS/METHODS

A phase I single fraction dose escalation trial of S-PBI for early-stage breast cancer was conducted. Women with in-situ or stage I-II (AJCC 6) invasive breast cancer following breast conserving surgery were treated with S-PBI in 1 fraction to a total dose of 22.5, 26.5 or 30 Gy (Clinical trials.gov ID NCT02685332). At one month follow-up, patients were asked to complete our novel "Patient Perspective Cost and Convenience of Care Questionnaire". Patients also completed the EQ-5D-5L, including the visual analogue scale of overall health (VAS), at enrollment, 6, 12-, 24-, 36-, and 48-month follow-up.

RESULTS

Of 29 patients enrolled and treated, questionnaire data was available for all patients. Our trial encompassed a wide range of annual household incomes, education, and employment status. Overall, 44.8% (n = 13/29) of patients reported that radiation treatment presented a financial burden. Interestingly, no demographic information, such as patient race, marital status, education, household income, or employment during treatment predicted perceived FT. Patients reporting FT trended towards younger age (median 64 vs 70.5) and having a cancer related co-pay similar to our 5 fraction S-PBI FT trial; however, due to the small size of this study, this did not reach significance (p = 0.24 and 0.10, respectively). VAS and utility scores were calculated per the EQ-5D-5L and remained unchanged from baseline through 4-year follow-up. Likewise, there was no difference in the utility or VAS between patients who reported FT and those who did not. Interestingly, while patient reported cosmesis was similar for all patients at enrollment, patients who reported FT noted significantly worse cosmesis scores (fair/poor vs good/excellent) at 6 month and 2-year follow-ups (p = 0.01 and 0.04, respectively). Finally, patients were surveyed on treatment related disruption to their daily activities and enjoyment of life. The median values were 0 (scale 0-10, with 0 being no disruption) regardless of perceived FT. Patients were also uniformly satisfied with treatment time with a median score of 10 (scale 0-10, 10 being most satisfied).

CONCLUSION

Here, we show that despite using SPBI in a single fraction, nearly half of the patients treated still reported FT of treatment. Importantly, single fraction S-PBI has no negative impact on patient VAS or utility scores, and all patients were uniformly satisfied with treatment time without significant disruption to their life.

MuSiC-ViT: A multi-task Siamese convolutional vision transformer for differentiating change from no-change in follow-up chest radiographs

A major responsibility of radiologists in routine clinical practice is to read follow-up chest radiographs (CXRs) to identify changes in a patient's condition. Diagnosing meaningful changes in follow-up CXRs is challenging because radiologists must differentiate disease changes from natural or benign variations. Here, we suggest using a multi-task Siamese convolutional vision transformer (MuSiC-ViT) with an anatomy-matching module (AMM) to mimic the radiologist's cognitive process for differentiating baseline change from no-change. MuSiC-ViT uses the convolutional neural networks (CNNs) meet vision transformers model that combines CNN and transformer architecture. It has three major components: a Siamese network architecture, an AMM, and multi-task learning. Because the input is a pair of CXRs, a Siamese network was adopted for the encoder. The AMM is an attention module that focuses on related regions in the CXR pairs. To mimic a radiologist's cognitive process, MuSiC-ViT was trained using multi-task learning, normal/abnormal and change/no-change classification, and anatomy-matching. Among 406 K CXRs studied, 88 K change and 115 K no-change pairs were acquired for the training dataset. The internal validation dataset consisted of 1,620 pairs. To demonstrate the robustness of MuSiC-ViT, we verified the results with two other validation datasets. MuSiC-ViT respectively achieved accuracies and area under the receiver operating characteristic curves of 0.728 and 0.797 on the internal validation dataset, 0.614 and 0.784 on the first external validation dataset, and 0.745 and 0.858 on a second temporally separated validation dataset. All code is available at https://github.com/chokyungjin/MuSiC-ViT.

Dosimetric Comparison of Adaptive Radiotherapy Modalities for Stereotactic Partial Breast Irradiation

PURPOSE/OBJECTIVE(S)

An increase in the availability of adaptive radiotherapy (ART) platforms have proven to be effective in the treatment of a variety of sites. In this study, we aim to evaluate the effectiveness of non-adaptive RT and 3 different ART platforms: (1) CBCT-based, (2) CT-based, and (3) MRI-based for stereotactic partial breast irradiation (SPBI).

MATERIALS/METHODS

Data were collected from 32 patients (16 left and 16 right breast) treated at a single institution. 16 patients (8 left and 8 right) treated using the non-ART platform were re-planned onto two different ART platforms, CBCT- and MRI-based. The remaining 16 patients treated using CT-based adaptive platform were not re-planned due to the prone patient treatment position (others systems supine). All cases were planned to 30 Gy in 5 fractions. Plan quality was evaluated based on pre-defined planning goals to the OARS: ipsilateral and contralateral lungs (Dmean, Dmax, V20 Gy, V9 Gy), ipsilateral (V15 Gy, V30 Gy) and contralateral breasts (Dmax), heart (Dmean, Dmax, V3 Gy, V1.5 Gy), skin (Dmax, V36.5 Gy), and rib (Dmax, V30 Gy). Target goals were defined by Dmax, Dmin, gradient index, and paddock conformality index. Re-planned cases were compared within the cohort using a paired t-test and a 2-sided t-test was used comparing to the CT-based platform.

RESULTS

Comparing the left and right breast cohort across all platforms, the CT-based ART system showed a signification dose reduction in Dmean (p<0.001 for all platforms), Dmax (p<0.001 for left breast, p<0.03 for right breast) and V9 Gy (p<0.004 for left breast, p<0.001 for right breast) to the ipsilateral lung, V15 Gy (p<0.004 for left breast cohort) to the ipsilateral breast, and Dmax to the contralateral breast (p<0.001) and ribs (p = 0.01, p<0.001, p = 0.01 for CBCT-ART, MRI-ART, and non-ART for left breast cohort only). On average, the MR-Linac platform showed the least degree of OAR sparing across nearly all dosimetric parameters evaluated when compared to all modalities, especially for contralateral lung Dmean and Dmax (p<0.05 for all dosimetric parameters for all platforms) and contralateral breast Dmax (p<0.003 for all platforms). The CBCT-based platform showed superior dose reduction in contralateral lung mean (p<0.03 for all platforms) and heart Dmean (p = 0.065, p<0.001, p = 0.045 for non-adaptive, MRI-ART, and CT-ART for left breast and p<0.008 for right breast). PTV coverage was comparable across all platforms, averaging at approximately 95%. The CT-based ART platform showed a significantly reduced gradient index relative to the CBCT- and MRI-based platforms (p<0.001).

CONCLUSION

For SPBI treatments, the CT-based ART platforms displayed a higher degree of OAR sparing for many of the dosimetric parameters recorded relative to the other ART and non-ART platforms presented. The MRI-based system typically showed less reduced OAR sparing; however, the advantage of the system is shown if soft tissue contrast is needed. PTV coverage remained comparable across all platforms.

OsCM regulates rice defence system in response to UV light supplemented with drought stress

Studies on plant responses to combined abiotic stresses are very limited, especially in major crop plants. The current study evaluated the response of chorismate mutase overexpressor (OxCM) rice line to combined UV light and drought stress. The experiments were conducted in pots in a growth chamber, and data were assessed for gene expression, antioxidant and hormone regulation, flavonoid accumulation, phenotypic variation, and amino acid accumulation. Wild-type (WT) rice had reduced the growth and vigour, while transgenic rice maintained growth and vigour under combined UV light and drought stress. ROS and lipid peroxidation analysis revealed that chorismate mutase (OsCM) reduced oxidative stress mediated by ROS scavenging and reduced lipid peroxidation. The combined stresses reduced biosynthesis of total flavonoids, kaempferol and quercetin in WT plants, but increased significantly in plants with OxCM. Phytohormone analysis showed that SA was reduced by 50% in WT and 73% in transgenic plants, while ABA was reduced by 22% in WT plants but increased to 129% in transgenic plants. Expression of chorismate mutase regulates phenylalanine biosynthesis, UV light and drought stress-responsive genes, e.g., phenylalanine ammonia lyase (OsPAL), dehydrin (OsDHN), dehydration-responsive element-binding (OsDREB), ras-related protein 7 (OsRab7), ultraviolet-B resistance 8 (OsUVR8), WRKY transcription factor 89 (OsWRKY89) and tryptophan synthase alpha chain (OsTSA). Moreover, OsCM also increases accumulation of free amino acids (aspartic acid, glutamic acid, leucine, tyrosine, phenylalanine and proline) and sodium (Na), potassium (K), and calcium (Ca) ions in response to the combined stresses. Together, these results suggest that chorismate mutase expression induces physiological, biochemical and molecular changes that enhance rice tolerance to combined UV light and drought stresses.

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.

An Image Turing Test on Realistic Gastroscopy Images Generated by Using the Progressive Growing of Generative Adversarial Networks

Generative adversarial networks (GAN) in medicine are valuable techniques for augmenting unbalanced rare data, anomaly detection, and avoiding patient privacy issues. However, there were limits to generating high-quality endoscopic images with various characteristics, such as peristalsis, viewpoints, light sources, and mucous patterns. This study used the progressive growing of GAN (PGGAN) within the normal distribution dataset to confirm the ability to generate high-quality gastrointestinal images and investigated what barriers PGGAN has to generate endoscopic images. We trained the PGGAN with 107,060 gastroscopy images from 4165 normal patients to generate highly realistic 5122 pixel-sized images. For the evaluation, visual Turing tests were conducted on 100 real and 100 synthetic images to distinguish the authenticity of images by 19 endoscopists. The endoscopists were divided into three groups based on their years of clinical experience for subgroup analysis. The overall accuracy, sensitivity, and specificity of the 19 endoscopist groups were 61.3%, 70.3%, and 52.4%, respectively. The mean accuracy of the three endoscopist groups was 62.4 [Group I], 59.8 [Group II], and 59.1% [Group III], which was not considered a significant difference. There were no statistically significant differences in the location of the stomach. However, the real images with the anatomical landmark pylorus had higher detection sensitivity. The images generated by PGGAN showed highly realistic depictions that were difficult to distinguish, regardless of their expertise as endoscopists. However, it was necessary to establish GANs that could better represent the rugal folds and mucous membrane texture.

Generative Adversarial Network-Based Image Conversion Among Different Computed Tomography Protocols and Vendors: Effects on Accuracy and Variability in Quantifying Regional Disease Patterns of Interstitial Lung Disease

OBJECTIVE

To assess whether computed tomography (CT) conversion across different scan parameters and manufacturers using a routable generative adversarial network (RouteGAN) can improve the accuracy and variability in quantifying interstitial lung disease (ILD) using a deep learning-based automated software.

MATERIALS AND METHODS

This study included patients with ILD who underwent thin-section CT. Unmatched CT images obtained using scanners from four manufacturers (vendors A-D), standard- or low-radiation doses, and sharp or medium kernels were classified into groups 1-7 according to acquisition conditions. CT images in groups 2-7 were converted into the target CT style (Group 1: vendor A, standard dose, and sharp kernel) using a RouteGAN. ILD was quantified on original and converted CT images using a deep learning-based software (Aview, Coreline Soft). The accuracy of quantification was analyzed using the dice similarity coefficient (DSC) and pixel-wise overlap accuracy metrics against manual quantification by a radiologist. Five radiologists evaluated quantification accuracy using a 10-point visual scoring system.

RESULTS

Three hundred and fifty CT slices from 150 patients (mean age: 67.6 ± 10.7 years; 56 females) were included. The overlap accuracies for quantifying total abnormalities in groups 2-7 improved after CT conversion (original vs. converted: 0.63 vs. 0.68 for DSC, 0.66 vs. 0.70 for pixel-wise recall, and 0.68 vs. 0.73 for pixel-wise precision; P < 0.002 for all). The DSCs of fibrosis score, honeycombing, and reticulation significantly increased after CT conversion (0.32 vs. 0.64, 0.19 vs. 0.47, and 0.23 vs. 0.54, P < 0.002 for all), whereas those of ground-glass opacity, consolidation, and emphysema did not change significantly or decreased slightly. The radiologists' scores were significantly higher (P < 0.001) and less variable on converted CT.

CONCLUSION

CT conversion using a RouteGAN can improve the accuracy and variability of CT images obtained using different scan parameters and manufacturers in deep learning-based quantification of ILD.

Automated detection of intracranial aneurysms using skeleton-based 3D patches, semantic segmentation, and auxiliary classification for overcoming data imbalance in brain TOF-MRA

Accurate and reliable detection of intracranial aneurysms is vital for subsequent treatment to prevent bleeding. However, the detection of intracranial aneurysms can be time-consuming and even challenging, and there is great variability among experts, especially in the case of small aneurysms. This study aimed to detect intracranial aneurysms accurately using a convolutional neural network (CNN) with 3D time-of-flight magnetic resonance angiography (TOF-MRA). A total of 154 3D TOF-MRA datasets with intracranial aneurysms were acquired, and the gold standards were manually drawn by neuroradiologists. We also obtained 113 subjects from a public dataset for external validation. These angiograms were pre-processed by using skull-stripping, signal intensity normalization, and N4 bias correction. The 3D patches along the vessel skeleton from MRA were extracted. Values of the ratio between the aneurysmal and the normal patches ranged from 1:1 to 1:5. The semantic segmentation on intracranial aneurysms was trained using a 3D U-Net with an auxiliary classifier to overcome the imbalance in patches. The proposed method achieved an accuracy of 0.910 in internal validation and external validation accuracy of 0.883 with a 2:1 ratio of normal to aneurysmal patches. This multi-task learning method showed that the aneurysm segmentation performance was sufficient to be helpful in an actual clinical setting.

Diagnosis of osteoporotic vertebral compression fractures and fracture level detection using multitask learning with U-Net in lumbar spine lateral radiographs

Recent studies of automatic diagnosis of vertebral compression fractures (VCFs) using deep learning mainly focus on segmentation and vertebral level detection in lumbar spine lateral radiographs (LSLRs). Herein, we developed a model for simultaneous VCF diagnosis and vertebral level detection without using adjacent vertebral bodies. In total, 1102 patients with VCF, 1171 controls were enrolled. The 1865, 208, and 198 LSLRS were divided into training, validation, and test dataset. A ground truth label with a 4-point trapezoidal shape was made based on radiological reports showing normal or VCF at some vertebral level. We applied a modified U-Net architecture, in which decoders were trained to detect VCF and vertebral levels, sharing the same encoder. The multi-task model was significantly better than the single-task model in sensitivity and area under the receiver operating characteristic curve. In the internal dataset, the accuracy, sensitivity, and specificity of fracture detection per patient or vertebral body were 0.929, 0.944, and 0.917 or 0.947, 0.628, and 0.977, respectively. In external validation, those of fracture detection per patient or vertebral body were 0.713, 0.979, and 0.447 or 0.828, 0.936, and 0.820, respectively. The success rates were 96 % and 94 % for vertebral level detection in internal and external validation, respectively. The multi-task-shared encoder was significantly better than the single-task encoder. Furthermore, both fracture and vertebral level detection was good in internal and external validation. Our deep learning model may help radiologists perform real-life medical examinations.

Subregional pharyngeal changes after orthognathic surgery in skeletal Class III patients analyzed by convolutional neural networks-based segmentation

OBJECTIVES

To evaluate the accuracy of fully automatic segmentation of pharyngeal volume of interests (VOIs) before and after orthognathic surgery in skeletal Class III patients using a convolutional neural network (CNN) model and to investigate the clinical applicability of artificial intelligence for quantitative evaluation of treatment changes in pharyngeal VOIs.

METHODS

310 cone-beam computed tomography (CBCT) images were divided into a training set (n=150), validation set (n=40), and test set (n=120). The test datasets comprised matched pairs of pre- and posttreatment images of 60 skeletal Class III patients (mean age 23.1±5.0 years; ANB<-2⁰) who underwent bimaxillary orthognathic surgery with orthodontic treatment. A 3D U-Net CNNs model was applied for fully automatic segmentation and measurement of subregional pharyngeal volumes of pretreatment (T0) and posttreatment (T1) scans. The model's accuracy was compared to semi-automatic segmentation outcomes by humans using the dice similarity coefficient (DSC) and volume similarity (VS). The correlation between surgical skeletal changes and model accuracy was obtained.

RESULTS

The proposed model achieved high performance of subregional pharyngeal segmentation on both T0 and T1 images, representing a significant T1-T0 difference of DSC only in the nasopharynx. Region-specific differences among pharyngeal VOIs, which were observed at T0, disappeared on the T1 images. The decreased DSC of nasopharyngeal segmentation after treatment was weakly correlated with the amount of maxillary advancement. There was no correlation between the mandibular setback amount and model accuracy.

CONCLUSIONS

The proposed model offers fast and accurate subregional pharyngeal segmentation on both pretreatment and posttreatment CBCT images in skeletal Class III patients.

CLINICAL SIGNIFICANCE

We elucidated the clinical applicability of the CNNs model to quantitatively evaluate subregional pharyngeal changes after surgical-orthodontic treatment, which offers a basis for developing a fully integrated multiclass CNNs model to predict pharyngeal responses after dentoskeletal treatments.

CheSS: Chest X-Ray Pre-trained Model via Self-supervised Contrastive Learning

Training deep learning models on medical images heavily depends on experts' expensive and laborious manual labels. In addition, these images, labels, and even models themselves are not widely publicly accessible and suffer from various kinds of bias and imbalances. In this paper, chest X-ray pre-trained model via self-supervised contrastive learning (CheSS) was proposed to learn models with various representations in chest radiographs (CXRs). Our contribution is a publicly accessible pretrained model trained with a 4.8-M CXR dataset using self-supervised learning with a contrastive learning and its validation with various kinds of downstream tasks including classification on the 6-class diseases in internal dataset, diseases classification in CheXpert, bone suppression, and nodule generation. When compared to a scratch model, on the 6-class classification test dataset, we achieved 28.5% increase in accuracy. On the CheXpert dataset, we achieved 1.3% increase in mean area under the receiver operating characteristic curve on the full dataset and 11.4% increase only using 1% data in stress test manner. On bone suppression with perceptual loss, we achieved improvement in peak signal to noise ratio from 34.99 to 37.77, structural similarity index measure from 0.976 to 0.977, and root-square-mean error from 4.410 to 3.301 when compared to ImageNet pretrained model. Finally, on nodule generation, we achieved improvement in Fréchet inception distance from 24.06 to 17.07. Our study showed the decent transferability of CheSS weights. CheSS weights can help researchers overcome data imbalance, data shortage, and inaccessibility of medical image datasets. CheSS weight is available at https://github.com/mi2rl/CheSS .

Screening of adolescent idiopathic scoliosis using generative adversarial network (GAN) inversion method in chest radiographs

OBJECTIVE

Conventional computer-aided diagnosis using convolutional neural networks (CNN) has limitations in detecting sensitive changes and determining accurate decision boundaries in spectral and structural diseases such as scoliosis. We devised a new method to detect and diagnose adolescent idiopathic scoliosis in chest X-rays (CXRs) employing the latent space's discriminative ability in the generative adversarial network (GAN) and a simple multi-layer perceptron (MLP) to screen adolescent idiopathic scoliosis CXRs.

MATERIALS AND METHODS

Our model was trained and validated in a two-step manner. First, we trained a GAN using CXRs with various scoliosis severities and utilized the trained network as a feature extractor using the GAN inversion method. Second, we classified each vector from the latent space using a simple MLP.

RESULTS

The 2-layer MLP exhibited the best classification in the ablation study. With this model, the area under the receiver operating characteristic (AUROC) curves were 0.850 in the internal and 0.847 in the external datasets. Furthermore, when the sensitivity was fixed at 0.9, the model's specificity was 0.697 in the internal and 0.646 in the external datasets.

CONCLUSION

We developed a classifier for Adolescent idiopathic scoliosis (AIS) through generative representation learning. Our model shows good AUROC under screening chest radiographs in both the internal and external datasets. Our model has learned the spectral severity of AIS, enabling it to generate normal images even when trained solely on scoliosis radiographs.

Improved Performance of Organic Thermoelectric Generators Through Interfacial Energetics

The interfacial energetics are known to play a crucial role in organic diodes, transistors, and sensors. Designing the metal-organic interface has been a tool to optimize the performance of organic (opto)electronic devices, but this is not reported for organic thermoelectrics. In this work, it is demonstrated that the electrical power of organic thermoelectric generators (OTEGs) is also strongly dependent on the metal-organic interfacial energetics. Without changing the thermoelectric figure of merit (ZT) of polythiophene-based conducting polymers, the generated power of an OTEG can vary by three orders of magnitude simply by tuning the work function of the metal contact to reach above 1000 µW cm^-2 . The effective Seebeck coefficient (S_eff ) of a metal/polymer/metal single leg OTEG includes an interfacial contribution (V_inter /ΔT) in addition to the intrinsic bulk Seebeck coefficient of the polythiophenes, such that S_eff  = S + V_inter /ΔT varies from 22.7 µV K^-1 [9.4 µV K^-1 ] with Al to 50.5 µV K^-1 [26.3 µV K^-1 ] with Pt for poly(3,4-ethylenedioxythiophene):p-toluenesulfonate [poly(3,4-ethylenedioxythiophene):poly(4-styrenesulfonate)]. Spectroscopic techniques are used to reveal a redox interfacial reaction affecting locally the doping level of the polymer at the vicinity of the metal-organic interface and conclude that the energetics at the metal-polymer interface provides a new strategy to enhance the performance of OTEGs.

An interactive and realistic phantom for cricothyroidotomy simulation of a patient with obesity through a reusable design using 3D-printing and Arduino

BACKGROUND AND OBJECTIVES

Proper airway management during emergencies can prevent serious complications. However, cricothyroidotomy is challenging in patients with obesity. Since this technique is not performed frequently but at a critical time, the opportunity for trainees is rare. Simulators for these procedures are also lacking. Therefore, we proposed a realistic and interactive cricothyroidotomy simulator.

METHODS

All anatomical structures were modeled based on computed tomography images of a patient with obesity. To mimic the feeling of incision during cricothyroidotomy, the incision site was modeled to distinguish between the skin and fat. To reinforce the educational purpose, capacitive touch sensors were attached to the artery, vein, and thyroid to generate audio feedback. The tensile strength of the silicone-cast skin was measured to verify the similarity of the mechanical properties between humans and our model. The fabrication and assembly accuracies of the phantom between the Standard Tessellation Language and the fabricated model were evaluated. Audio feedback through sensing the anatomy parts and utilization was evaluated.

RESULTS

The body, skull, clavicle, artery, vein, and thyroid were fabricated using fused deposition modeling (FDM) with polylactic acid. A skin mold was fabricated using FDM with thermoplastic polyurethane. A fat mold was fabricated using stereolithography apparatus (SLA) with a clear resin. The airway and tongue were fabricated using SLA with an elastic resin. The tensile strength of the skin using silicone with and without polyester mesh was 2.63 ± 0.68 and 2.46 ± 0.21 MPa. The measurement errors for fabricating and assembling parts of the phantom between the STL and the fabricated models were -0.08 ± 0.19 mm and 0.13 ± 0.64 mm. The measurement errors internal anatomy embodied surfaces in fat part were 0.41 ± 0.89 mm. Audio feedback was generated 100% in all the areas tested. The realism, understanding of clinical skills, and intention to retrain were 7.1, 8.8, and 8.3 average points.

CONCLUSIONS

Our simulator can provide a realistic simulation experience for trainees through a realistic feeling of incision and audio feedback, which can be used for actual clinical education.

Improvement of semantic segmentation through transfer learning of multi-class regions with convolutional neural networks on supine and prone breast MRI images

Semantic segmentation of breast and surrounding tissues in supine and prone breast magnetic resonance imaging (MRI) is required for various kinds of computer-assisted diagnoses for surgical applications. Variability of breast shape in supine and prone poses along with various MRI artifacts makes it difficult to determine robust breast and surrounding tissue segmentation. Therefore, we evaluated semantic segmentation with transfer learning of convolutional neural networks to create robust breast segmentation in supine breast MRI without considering supine or prone positions. Total 29 patients with T1-weighted contrast-enhanced images were collected at Asan Medical Center and two types of breast MRI were performed in the prone position and the supine position. The four classes, including lungs and heart, muscles and bones, parenchyma with cancer, and skin and fat, were manually drawn by an expert. Semantic segmentation on breast MRI scans with supine, prone, transferred from prone to supine, and pooled supine and prone MRI were trained and compared using 2D U-Net, 3D U-Net, 2D nnU-Net and 3D nnU-Net. The best performance was 2D models with transfer learning. Our results showed excellent performance and could be used for clinical purposes such as breast registration and computer-aided diagnosis.

Deep Learning of Time-Signal Intensity Curves from Dynamic Susceptibility Contrast Imaging Enables Tissue Labeling and Prediction of Survival in Glioblastoma

BACKGROUND AND PURPOSE

An autoencoder can learn representative time-signal intensity patterns to provide tissue heterogeneity measures using dynamic susceptibility contrast MR imaging. The aim of this study was to investigate whether such an autoencoder-based pattern analysis could provide interpretable tissue labeling and prognostic value in isocitrate dehydrogenase (IDH) wild-type glioblastoma.

MATERIALS AND METHODS

Preoperative dynamic susceptibility contrast MR images were obtained from 272 patients with IDH wild-type glioblastoma (training and validation, 183 and 89 patients, respectively). The autoencoder was applied to the dynamic susceptibility contrast MR imaging time-signal intensity curves of tumor and peritumoral areas. Representative perfusion patterns were defined by voxelwise K-means clustering using autoencoder latent features. Perfusion patterns were labeled by comparing parameters with anatomic reference tissues for baseline, signal drop, and percentage recovery. In the validation set (n = 89), a survival model was created from representative patterns and clinical predictors using Cox proportional hazard regression analysis, and its performance was calculated using the Harrell C-index.

RESULTS

Eighty-nine patients were enrolled. Five representative perfusion patterns were used to characterize tissues as high angiogenic tumor, low angiogenic/cellular tumor, perinecrotic lesion, infiltrated edema, and vasogenic edema. Of these, the low angiogenic/cellular tumor (hazard ratio, 2.18; P = .047) and infiltrated edema patterns (hazard ratio, 1.88; P = .009) in peritumoral areas showed significant prognostic value. The combined perfusion patterns and clinical predictors (C-index, 0.72) improved prognostication when added to clinical predictors (C-index, 0.55).

CONCLUSIONS

The autoencoder perfusion pattern analysis enabled tissue characterization of peritumoral areas, providing heterogeneity and dynamic information that may provide useful prognostic information in IDH wild-type glioblastoma.

Draft Genome Sequences of 27 Rhizogenic Agrobacterium Biovar 1 Strains, the Causative Agent of Hairy Root Disease

Rhizogenic Agrobacterium biovar 1 strains are important plant pathogens that cause hairy root disease in Cucurbitaceae and Solanaceae crops cultivated under hydroponic conditions. In contrast to tumorigenic agrobacteria, only a few genome sequences of rhizogenic agrobacteria are currently available. Here, we report the draft genome sequences of 27 rhizogenic Agrobacterium strains.