Multitask deep learning for prediction of microvascular invasion and recurrence-free survival in hepatocellular carcinoma based on MRI images.

BACKGROUND AND AIMS: Accurate preoperative prediction of microvascular invasion (MVI) and recurrence-free survival (RFS) is vital for personalised hepatocellular carcinoma (HCC) management. We developed a multitask deep learning model to predict MVI and RFS using preoperative MRI scans. METHODS: Utilising a retrospective dataset of 725 HCC patients from seven institutions, we developed and validated a multitask deep learning model focused on predicting MVI and RFS. The model employs a transformer architecture to extract critical features from preoperative MRI scans. It was trained on a set of 234 patients and internally validated on a set of 58 patients. External validation was performed using three independent sets (n = 212, 111, 110). RESULTS: The multitask deep learning model yielded high MVI prediction accuracy, with AUC values of 0.918 for the training set and 0.800 for the internal test set. In external test sets, AUC values were 0.837, 0.815 and 0.800. Radiologists' sensitivity and inter-rater agreement for MVI prediction improved significantly when integrated with the model. For RFS, the model achieved C-index values of 0.763 in the training set and ranged between 0.628 and 0.728 in external test sets. Notably, PA-TACE improved RFS only in patients predicted to have high MVI risk and low survival scores (p < .001). CONCLUSIONS: Our deep learning model allows accurate MVI and survival prediction in HCC patients. Prospective studies are warranted to assess the clinical utility of this model in guiding personalised treatment in conjunction with clinical criteria.

OA-GAN: organ-aware generative adversarial network for synthesizing contrast-enhanced medical images.

Contrast-enhanced computed tomography (CE-CT) images are vital for clinical diagnosis of focal liver lesions (FLLs). However, the use of CE-CT images imposes a significant burden on patients due to the injection of contrast agents and extended shooting. Deep learning-based image synthesis models offer a promising solution that synthesizes CE-CT images from non-contrasted CT (NC-CT) images. Unlike natural images, medical image synthesis requires a specific focus on certain organs or localized regions to ensure accurate diagnosis. Determining how to effectively emphasize target organs poses a challenging issue in medical image synthesis. To solve this challenge, we present a novel CE-CT image synthesis model called, Organ-Aware Generative Adversarial Network (OA-GAN). The OA-GAN comprises an organ-aware (OA) network and a dual decoder-based generator. First, the OA network learns the most discriminative spatial features about the target organ (i.e. liver) by utilizing the ground truth organ mask as localization cues. Subsequently, NC-CT image and captured feature are fed into the dual decoder-based generator, which employs a local and global decoder network to simultaneously synthesize the organ and entire CECT image. Moreover, the semantic information extracted from the local decoder is transferred to the global decoder to facilitate better reconstruction of the organ in entire CE-CT image. The qualitative and quantitative evaluation on a CE-CT dataset demonstrates that the OA-GAN outperforms state-of-the-art approaches for synthesizing two types of CE-CT images such as arterial phase and portal venous phase. Additionally, subjective evaluations by expert radiologists and a deep learning-based FLLs classification also affirm that CE-CT images synthesized from the OA-GAN exhibit a remarkable resemblance to real CE-CT images.

Force Transmission Analysis and Optimization of Bowden Cable on Body in a Flexible Exoskeleton.

The Bowden cable is a significant force transmission equipment for a flexible exoskeleton. However, the previous researches of Bowden cable had emphasized on the data from experimenting test board, instead of on human body, which produced the inaccurate assisting analysis of the flexible exoskeleton. In this paper, a flexible exoskeleton for assisting knee extension was proposed, which provided an on-body condition. Then, the friction force and its influencing factors between the wire rope and sheath of the Bowden cable from the motor to the anchor of knee have been analyzed. The segment models of force transmission with the concern of three kinds of friction modes were established, and the relationship between various lengths and bending angles of Bowden cable was fitted to the equations of curve. Furthermore, the association rule between the force transmission and the lengths of Bowden cable was obtained, based on which, the optimal force transmission efficiency was 78.68% when the length value of the Bowden cable was 475 mm. A flexible exoskeleton prototype was assembled; then, the experiments with force transmission and metabolic cost have been developed. The results showed that the force transmission efficiency had strong association with the lengths of Bowden cable, instead of the transmission velocities. Furthermore, this knee assistance exoskeleton reduced the net metabolic cost of the testees during walking. These experiments results corroborated the force transmission modeling and simulation of the Bowden cable on body we proposed in this paper.

A Hierarchical Stabilization Control Method for a Three-Axis Gimbal Based on Sea-Sky-Line Detection.

Obtaining a stable video sequence for cameras on surface vehicles is always a challenging problem due to the severe disturbances in heavy sea environments. Aiming at this problem, this paper proposes a novel hierarchical stabilization method based on real-time sea-sky-line detection. More specifically, a hierarchical image stabilization control method that combines mechanical image stabilization with electronic image stabilization is adopted. With respect to the mechanical image stabilization method, a gimbal with three degrees of freedom (DOFs) and with a robust controller is utilized for the primary motion compensation. In addition, the electronic image stabilization method based on sea-sky-line detection in video sequences accomplishes motion estimation and compensation. The Canny algorithm and Hough transform are utilized to detect the sea-sky line. Noticeably, an image-clipping strategy based on prior information is implemented to ensure real-time performance, which can effectively improve the processing speed and reduce the equipment performance requirements. The experimental results indicate that the proposed method for mechanical and electronic stabilization can reduce the vibration by 74.2% and 42.1%, respectively.

A Docking Mechanism Based on a Stewart Platform and Its Tracking Control Based on Information Fusion Algorithm.

Aiming at the problem of unmanned reconfiguration and docking of ground vehicles under complex working conditions, we designed a piece of docking equipment composed of an active mechanism based on a six-degree-of-freedom platform and a locking mechanism with multi-sensors. Through the proposed control method based on laser and image sensor information fusion calculation, the six-dimensional posture information of the mechanism during the docking process is captured in real time so as to achieve high-precision docking. Finally, the effectiveness of the method and the feasibility of the 6-DOF platform are verified by the established model. The results show that the mechanism can meet the requirements of smooth docking of ground unmanned vehicles.