Accurate localization of thoracic lymph nodes during endobronchial ultrasound (EBUS) is crucial for lung cancer staging, treatment planning, and prognostication. Artificial intelligence (AI) has the potential to support this process. Deep learning (DL) models often lack transparency but can benefit from explainable AI (XAI) tools like Gradient-weighted Class Activation Mapping (Grad-CAM). However, no prior study has quantitatively assessed whether model attention in EBUS imaging corresponds to relevant anatomy. This study developed a convolutional neural network (CNN) to classify thoracic lymph node stations and evaluated the anatomical relevance of Grad-CAM activations using a structured annotation framework. Applied on 35,527 labeled EBUS images, the CNN achieved 63.1% accuracy, with the highest F1-score in stations 4L, 4R, and 10R. Three expert bronchoscopists independently annotated Grad-CAM maps from 3131 test images. Activations predominantly aligned with lymph nodes and/or blood vessels, yielding an accuracy of 65.9% and an F1-score of 58.4%, with moderate interobserver agreement. These findings indicate that DL can aid lymph node station classification and that XAI offers meaningful insight into model behavior. The proposed framework may enhance anatomical orientation and operator training during EBUS, although further optimization and multicenter validation are required.

Musculoskeletal ultrasound (MSUS) is widely used in clinical medicine and is extensively used in rheumatology. However, acquiring proficiency in MSUS remains challenging and demands substantial training. The advent of machine learning and artificial intelligence (AI) in medical imaging has introduced new possibilities. In this paper we present the automatic segmentation and annotation of 11 anatomical structures in the lower forearm and different possibilities of 3D visualization of the most important anatomical structures of the forearm, demonstrating the potential of this technology as a bedside teaching tool.

Endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) is a cornerstone in minimally invasive thoracic lymph node sampling. In lung cancer staging, precise assessment of lymph node position is crucial for clinical decision-making. This study aimed to demonstrate a new deep learning method to classify thoracic lymph nodes based on their anatomical location using EBUS images. Bronchoscopists labeled lymph node stations in real-time according to the Mountain Dressler nomenclature. EBUS images were then used to train and test a deep neural network (DNN) model, with intraoperative labels as ground truth. In total, 28,134 EBUS images were acquired from 56 patients. The model achieved an overall classification accuracy of 59.5 ± 5.2%. The highest precision, sensitivity, and F1 score were observed in station 4L, 77.6 ± 13.1%, 77.6 ± 15.4%, and 77.6 ± 15.4%, respectively. The lowest precision, sensitivity, and F1 score were observed in station 10L. The average processing and prediction time for a sequence of ten images was 0.65 ± 0.04 s, demonstrating the feasibility of real-time applications. In conclusion, the new DNN-based model could be used to classify lymph node stations from EBUS images. The method performance was promising with a potential for clinical use.

Classification of lung cancer subtypes is a critical clinical step; however, relying solely on H&E-stained histopathology images can pose challenges, and additional immunohistochemical analysis is sometimes required for definitive subtyping. Digital pathology facilitates the use of artificial intelligence for automatic classification of digital tissue slides. Automatic classification of Whole Slide Images (WSIs) typically involves extracting features from patches obtained from them. The aim of this study was to develop a WSI classification framework utilizing an optimizable kernel to encode features from each patch from a WSI into a desirable and adjustable latent space using an evolutionary algorithm. The encoded data can then be used for classification and segmentation while being computationally more efficient. Our proposed framework is compared with a state-of-the-art model, Vim4Path, on an internal and external dataset of lung cancer WSIs. The proposed model outperforms Vim-S16 in accuracy and F1 score at both ×2.5 and ×10 magnification levels on the internal test set, with the highest accuracy (0.833) and F1 score (0.721) at ×2.5. On the external test set, Vim-S16 at ×10 achieves the highest accuracy (0.732), whereas OKEN-DenseNet121 at ×2.5 has the best F1 score (0.772). In future work, finding a dynamic way to tune the output dimensions of the evolutionary algorithm would be of value.

The increased workload in pathology laboratories today means automated tools such as artificial intelligence models can be useful, helping pathologists with their tasks. In this paper, we propose a segmentation model (DRU-Net) that can provide a delineation of human non-small cell lung carcinomas and an augmentation method that can improve classification results. The proposed model is a fused combination of truncated pre-trained DenseNet201 and ResNet101V2 as a patch-wise classifier, followed by a lightweight U-Net as a refinement model. Two datasets (Norwegian Lung Cancer Biobank and Haukeland University Lung Cancer cohort) were used to develop the model. The DRU-Net model achieved an average of 0.91 Dice similarity coefficient. The proposed spatial augmentation method (multi-lens distortion) improved the Dice similarity coefficient from 0.88 to 0.91. Our findings show that selecting image patches that specifically include regions of interest leads to better results for the patch-wise classifier compared to other sampling methods. A qualitative analysis by pathology experts showed that the DRU-Net model was generally successful in tumor detection. Results in the test set showed some areas of false-positive and false-negative segmentation in the periphery, particularly in tumors with inflammatory and reactive changes. In summary, the presented DRU-Net model demonstrated the best performance on the segmentation task, and the proposed augmentation technique proved to improve the results.

To improve the prognosis of patients suffering from pulmonary diseases, such as lung cancer, early diagnosis and treatment are crucial. The analysis of CT images is invaluable for diagnosis, whereas high quality segmentation of the airway tree are required for intervention planning and live guidance during bronchoscopy. Recently, the Multi-domain Airway Tree Modeling (ATM'22) challenge released a large dataset, both enabling training of deep-learning based models and bringing substantial improvement of the state-of-the-art for the airway segmentation task. The ATM'22 dataset includes a large group of COVID'19 patients and a range of other lung diseases, however, relatively few patients with severe pathologies affecting the airway tree anatomy was found. In this study, we introduce a new public benchmark dataset (AeroPath), consisting of 27 CT images from patients with pathologies ranging from emphysema to large tumors, with corresponding trachea and bronchi annotations. Second, we present a multiscale fusion design for automatic airway segmentation. Models were trained on the ATM'22 dataset, tested on the AeroPath dataset, and further evaluated against competitive open-source methods. The same performance metrics as used in the ATM'22 challenge were used to benchmark the different considered approaches. Lastly, an open web application is developed, to easily test the proposed model on new data. The results demonstrated that our proposed architecture predicted topologically correct segmentations for all the patients included in the AeroPath dataset. The proposed method is robust and able to handle various anomalies, down to at least the fifth airway generation. In addition, the AeroPath dataset, featuring patients with challenging pathologies, will contribute to development of new state-of-the-art methods. The AeroPath dataset and the web application are made openly available.

Background and objective Patients suspected to have lung cancer, undergo endobronchial ultrasound bronchoscopy (EBUS) for the purpose of diagnosis and staging. For presumptive curable patients, the EBUS bronchoscopy is planned based on images and data from computed tomography (CT) images and positron emission tomography (PET). Our study aimed to evaluate the feasibility of a multimodal electromagnetic navigation platform for EBUS bronchoscopy, integrating ultrasound and segmented CT, and PET scan imaging data. Methods The proof-of-concept study included patients with suspected lung cancer and pathological mediastinal/hilar lymph nodes identified on both CT and PET scans. Images obtained from these two modalities were segmented to delineate target lymph nodes and then incorporated into the CustusX navigation platform. The EBUS bronchoscope was equipped with a sensor, calibrated, and affixed to a 3D printed click-on device positioned at the bronchoscope’s tip. Navigation accuracy was measured postoperatively using ultrasound recordings. Results The study enrolled three patients, all presenting with suspected mediastinal lymph node metastasis (N1-3). All PET-positive lymph nodes were displayed in the navigation platform during the EBUS procedures. In total, five distinct lymph nodes were sampled, yielding malignant cells from three nodes and lymphocytes from the remaining two. The median accuracy of the navigation system was 7.7 mm. Conclusion Our study introduces a feasible multimodal electromagnetic navigation platform that combines intraoperative ultrasound with preoperative segmented CT and PET imaging data for EBUS lymph node staging examinations. This innovative approach holds promise for enhancing the accuracy and effectiveness of EBUS procedures.

Learning cardiac morphology largely involves spatial abilities and studies indicate benefits from innovative 3D visualization technologies that speed up and increase the learning output. Studies comparing these teaching tools and their educational output are rare and few studies include complex congenital heart defects. This study compared the effects of 3D prints, mixed reality (MR) viewing of 3D meshes and standard cardiac CT slice images on medical students' understanding of complex congenital heart defect morphology, measuring both objective level of understanding and subjective educational experience. The objective of this study was to compare morphological understanding and user experiences of 3D printed models, MR 3D visualization and axial 2D CT slices, in medical students examining morphological details in complex congenital heart defects. Medical students in the median 4th year of study (range 2nd to 6th) examined three of five different complex congenital heart defects by three different modalities: 3D printed model, MR viewed 3D mesh, and cardiac CT slices, answering a questionnaire on morphology and user experience. Time to complete task, diagnostic accuracy, and user experience data were collected and compared on group level. Task times were similar for all modalities. The percentage of correct answers was higher with MR visualization, which was also the preferred modality overall. Medical students both prefer and better understand the morphology of complex congenital heart disease with 3D models viewed using MR, without spending more time than with 3D prints or 2D CT images.

Endobronchial ultrasound (EBUS) is used in the minimally invasive sampling of thoracic lymph nodes. In lung cancer staging, the accurate assessment of mediastinal structures is essential but challenged by variations in anatomy, image quality, and operator-dependent image interpretation. This study aimed to automatically detect and segment mediastinal lymph nodes and blood vessels employing a novel U-Net architecture-based approach in EBUS images. A total of 1161 EBUS images from 40 patients were annotated. For training and validation, 882 images from 30 patients and 145 images from 5 patients were utilized. A separate set of 134 images was reserved for testing. For lymph node and blood vessel segmentation, the mean ± standard deviation (SD) values of the Dice similarity coefficient were 0.71 ± 0.35 and 0.76 ± 0.38, those of the precision were 0.69 ± 0.36 and 0.82 ± 0.22, those of the sensitivity were 0.71 ± 0.38 and 0.80 ± 0.25, those of the specificity were 0.98 ± 0.02 and 0.99 ± 0.01, and those of the F1 score were 0.85 ± 0.16 and 0.81 ± 0.21, respectively. The average processing and segmentation run-time per image was 55 ± 1 ms (mean ± SD). The new U-Net architecture-based approach (EBUS-AI) could automatically detect and segment mediastinal lymph nodes and blood vessels in EBUS images. The method performed well and was feasible and fast, enabling real-time automatic labeling.

Machine learning and deep learning are novel methods which are revolutionizing medical imaging. In our study we trained an algorithm with a U-Net shaped network to recognize ultrasound images of the median nerve in the complete distal half of the forearm and to measure the cross-sectional area at the inlet of the carpal tunnel. Images of 25 patient hands with carpal tunnel syndrome (CTS) and 26 healthy controls were recorded on a video loop covering 15 cm of the distal forearm and 2355 images were manually segmented. We found an average Dice score of 0.76 between manual and automated segmentation of the median nerve in its complete course, while the measurement of the cross-sectional area at the carpal tunnel inlet resulted in a 10.9% difference between manually and automated measurements. We regard this technology as a suitable device for verifying the diagnosis of CTS.

The integration of medical robots is revolutionizing clinical medicine, especially in procedures requiring precision in instrument manipulation and navigation within the body using medical imaging techniques like fluoroscopy, CT, and MRI. This is particularly challenging in peripheral lung lesion examinations, where guiding long, flexible instruments through the lung airways to the target exposes medical professionals and patients to harmful X-ray radiation. Several robotic approaches exist but there are still shortcomings in terms of their large footprint in the operating room and complex and costly mechanical structure. The goal of our research was to develop and test the early feasibility of RoboCath, an innovative robotic platform designed for the adaptable navigation of long, flexible catheter-like medical instruments through the lung airways. RoboCath seamlessly integrates with existing medical devices, such as bronchoscopes, facilitating access to lower lung regions with precision. Its compact design is ideal for crowded operating rooms, ensures sterilization compatibility, and supports a broad range of procedures, including those requiring intricate instrument manipulation and medical imaging for navigation. This technology significantly reduces the reliance on X-ray, thereby minimizing radiation exposure to both healthcare providers and patients. The RoboCath system represents a significant advancement in the field of medical robotics, offering a novel solution to the challenges of lung lesion biopsy and diagnosis, with potential applications in various open orifice procedures.### Competing Interest StatementThe authors have declared no competing interest.

Background Dental caries diagnosis requires the manual inspection of diagnostic bitewing images of the patient, followed by a visual inspection and probing of the identified dental pieces with potential lesions. Yet the use of artificial intelligence, and in particular deep-learning, has the potential to aid in the diagnosis by providing a quick and informative analysis of the bitewing images. Methods A dataset of 13,887 bitewings from the HUNT4 Oral Health Study were annotated individually by six different experts, and used to train three different object detection deep-learning architectures: RetinaNet (ResNet50), YOLOv5 (M size), and EfficientDet (D0 and D1 sizes). A consensus dataset of 197 images, annotated jointly by the same six dental clinicians, was used for evaluation. A five-fold cross validation scheme was used to evaluate the performance of the AI models. Results The trained models show an increase in average precision and F1-score, and decrease of false negative rate, with respect to the dental clinicians. When compared against the dental clinicians, the YOLOv5 model shows the largest improvement, reporting 0.647 mean average precision, 0.548 mean F1-score, and 0.149 mean false negative rate. Whereas the best annotators on each of these metrics reported 0.299, 0.495, and 0.164 respectively. Conclusion Deep-learning models have shown the potential to assist dental professionals in the diagnosis of caries. Yet, the task remains challenging due to the artifacts natural to the bitewing images.

The purpose of this study was to evaluate the accuracy of a method for estimating the tip position of a fiber optic shape-sensing (FOSS) integrated instrument being inserted through a bronchoscope. A modified guidewire with a multicore optical fiber was inserted into the working channel of a custom-made catheter with three electromagnetic (EM) sensors. The displacement between the instruments was manually set, and a point-based method was applied to match the position of the EM sensors to corresponding points on the shape. The accuracy was evaluated in a realistic bronchial model. An additional EM sensor was used to sample the tip of the guidewire, and the absolute deviation between this position and the estimated tip position was calculated. For small displacements between the tip of the FOSS integrated tool and the catheter, the median deviation in estimated tip position was ≤5 mm. For larger displacements, deviations exceeding 10 mm were observed. The deviations increased when the shape sensor had sharp curvatures relative to more straight shapes. The method works well for clinically relevant displacements of a biopsy tool from the bronchoscope tip, and when the path to the lesion has limited curvatures. However, improvements must be made to our configuration before pursuing further clinical testing.

PurposeCancer is among the leading causes of death in the developed world, and lung cancer is the most lethal type. Early detection is crucial for better prognosis, but can be resource intensive to achieve. Automating tasks such as lung tumor localization and segmentation in radiological images can free valuable time for radiologists and other clinical personnel. Convolutional neural networks may be suited for such tasks, but require substantial amounts of labeled data to train. Obtaining labeled data is a challenge, especially in the medical domain. MethodsThis paper investigates the use of a teacher-student design to utilize datasets with different types of supervision to train an automatic model performing pulmonary tumor segmentation on computed tomography images. The framework consists of two models: the student that performs end-to-end automatic tumor segmentation and the teacher that supplies the student additional pseudo-annotated data during training. ResultsUsing only a small proportion of semantically labeled data and a large number of bounding box annotated data, we achieved competitive performance using a teacher-student design. Models trained on larger amounts of semantic annotations did not perform better than those trained on teacher-annotated data. Our model trained on a small number of semantically labeled data achieved a mean dice similarity coefficient of 71.0 on the MSD Lung dataset. ConclusionsOur results demonstrate the potential of utilizing teacher-student designs to reduce the annotation load, as less supervised annotation schemes may be performed, without any real degradation in segmentation accuracy.

Introduction and aims: During an Endovascular Aneurysm Repair (EVAR) procedure a stiff guidewire is inserted from the iliac arteries. This induces significant deformations on the vasculature, thus, affecting the pre-operative planning, and the accuracy of image fusion. The aim of the present work is to predict the guidewire induced deformations using a finite element approach validated through experiments with patient-specific additive manufactured models. The numerical approach herein developed could improve the pre-operative planning and the intra-operative navigation.Material and methods: The physical models used for the experiments in the hybrid operating room, were manufactured from the segmentations of pre-operative Computed Tomography (CT) angiographies. The finite element analyses (FEA) were performed with LS-DYNA Explicit. The material properties used in finite element analyses were obtained by uniaxial tensile tests. The experimental deformed configurations of the aorta were compared to those obtained from FEA. Three models, obtained from Computed Tomography acquisitions, were investigated in the present work: A) without intraluminal thrombus (ILT), B) with ILT, C) with ILT and calcifications.Results and discussion: A good agreement was found between the experimental and the computational studies. The average error between the final in vitro vs. in silico aortic configurations, i.e., when the guidewire is fully inserted, are equal to 1.17, 1.22 and 1.40 mm, respectively, for Models A, B and C. The increasing trend in values of deformations from Model A to Model C was noticed both experimentally and numerically. The presented validated computational approach in combination with a tracking technology of the endovascular devices may be used to obtain the intra-operative configuration of the vessels and devices prior to the procedure, thus limiting the radiation exposure and the contrast agent dose.

Purpose: This study aims to explore training strategies to improve convolutional neural network-based image-to-image deformable registration for abdominal imaging. Methods: Different training strategies, loss functions, and transfer learning schemes were considered. Furthermore, an augmentation layer which generates artificial training image pairs on-the-fly was proposed, in addition to a loss layer that enables dynamic loss weighting. Results: Guiding registration using segmentations in the training step proved beneficial for deep-learning-based image registration. Finetuning the pretrained model from the brain MRI dataset to the abdominal CT dataset further improved performance on the latter application, removing the need for a large dataset to yield satisfactory performance. Dynamic loss weighting also marginally improved performance, all without impacting inference runtime. Conclusion: Using simple concepts, we improved the performance of a commonly used deep image registration architecture, VoxelMorph. In future work, our framework, DDMR, should be validated on different datasets to further assess its value.

ABSTRACT As lung cancer evolves, the presence of potentially malignant lymph nodes must be assessed to properly estimate disease progression and select the best treatment strategy. A method for accurate and automatic segmentation is hence decisive for quantitatively describing lymph nodes. In this study, the use of 3D convolutional neural networks, either through slab-wise schemes or the leveraging of downsampled entire volumes, is investigated. As lymph nodes have similar attenuation values to nearby anatomical structures, we use the knowledge of other organs as prior information to guide the segmentation. To assess the performances, a 5-fold cross-validation strategy was followed over a dataset of 120 contrast-enhanced CT volumes. For the 1178 lymph nodes with a short-axis diameter mm, our best-performing approach reached a patient-wise recall of 92%, a false positive per patient ratio of 5 and a segmentation overlap of 80.5%. Fusing a slab-wise and a full volume approach within an ensemble scheme generated the best performances. The anatomical priors guiding strategy is promising, yet a larger set than four organs appears needed to generate an optimal benefit. A larger dataset is also mandatory given the wide range of expressions a lymph node can exhibit (i.e. shape, location and attenuation).