An Interpretable Deep Hierarchical Semantic Convolutional Neural Network for Lung Nodule Malignancy Classification.

While deep learning methods have demonstrated performance comparable to human readers in tasks such as computer-aided diagnosis, these models are difficult to interpret, do not incorporate prior domain knowledge, and are often considered as a "black-box." The lack of model interpretability hinders them from being fully understood by end users such as radiologists. In this paper, we present a novel interpretable deep hierarchical semantic convolutional neural network (HSCNN) to predict whether a given pulmonary nodule observed on a computed tomography (CT) scan is malignant. Our network provides two levels of output: 1) low-level semantic features; and 2) a high-level prediction of nodule malignancy. The low-level outputs reflect diagnostic features often reported by radiologists and serve to explain how the model interprets the images in an expert-interpretable manner. The information from these low-level outputs, along with the representations learned by the convolutional layers, are then combined and used to infer the high-level output. This unified architecture is trained by optimizing a global loss function including both low- and high-level tasks, thereby learning all the parameters within a joint framework. Our experimental results using the Lung Image Database Consortium (LIDC) show that the proposed method not only produces interpretable lung cancer predictions but also achieves significantly better results compared to using a 3D CNN alone.

M3 receptor is involved in the effect of penehyclidine hydrochloride reduced endothelial injury in LPS-stimulated human pulmonary microvascular endothelial cell.

LPS has been recently shown to induce muscarinic acetylcholine 3 receptor (M3 receptor) expression and penehyclidine hydrochloride (PHC) is an anticholinergic drug which could block the expression of M3 receptor. PHC has been demonstrated to perform protective effect on cell injury. This study is to investigate whether the effect of PHC on microvascular endothelial injury is related to its inhibition of M3 receptor or not. HPMVECs were treated with specific M3 receptor shRNA or PBS, and randomly divided into LPS group (A group), LPS+PHC group (B group), LPS + M3 shRNA group (C group) and LPS + PHC + M3 shRNA group (D group). Cells were collected at 60 min after LPS treatment to measure levels of LDH, endothelial permeability, TNF-α and IL-6 levels, NF-κB p65 activation, I-κB protein expression, p38MAPK, and ERK1/2 activations as well as M3 mRNA expression. PHC could decrease LDH levels, cell permeability, TNF-α and IL-6 levels, p38 MAPK, ERK1/2, NF-κB p65 activations and M3 mRNA expressions compared with LPS group. When M3 receptor was silence, the changes of these indices were much more obvious. These findings suggest that M3 receptor plays an important role in LPS-induced pulmonary microvascular endothelial injury, which is regulated through NF-κB p65 and MAPK activation. And knockout of M3 receptor could attenuate LPS-induced pulmonary microvascular endothelial injury. Regulative effects of PHC on pulmonary microvascular permeability and NF-κB p65 as well as MAPK activations are including but not limited to inhibition of M3 receptor.

Increased ACL direct insertion coverage provided more positive biomechanical effects on graft and bone tunnel during knee flexion: a simulation study.

PURPOSE: Flattened femoral tunnels were recently applied in anatomical single-bundle anterior cruciate ligament (ACL) reconstruction. Little is known about the biomechanical effect of such changes during knee flexion. The aim of the present simulation study was to assess the effect of altered ACL direct insertion coverage on the biomechanics of the graft and bone tunnel. METHODS: Five finite element (FE) models, including a round femoral tunnel and four progressively flattened rounded rectangular femoral tunnels, were established to represent the ACL reconstructions. In vivo knee kinematics data obtained from the validated dual fluoroscopic imaging techniques controlled the FE models to simulate lunge motions. The maximal principal stress of the graft and the volume of equivalent strain within 1000-3000 microstrain (V1000-3000) of the cancellous bone were subsequently calculated at 0°, 30°, 60° and 90° of knee flexion. RESULTS: A lower stress state on the graft and a more beneficial strain state on the cancellous bone were observed when the femoral tunnel better covered the ACL direct insertion. The average maximal principal stress of each model were 3.93 ± 0.60 MPa, 3.82 ± 0.54 MPa, 3.43 ± 0.44 MPa, 3.45 ± 0.44 MPa and 3.05 ± 0.43 MPa, respectively. The average V1000-3000 of the cancellous bone of each model were 179.06 ± 89.62 mm3, 221.40 ± 129.83 mm3, 247.57 ± 157.78 mm3, 282.74 ± 178.51 mm3 and 295.71 ± 162.59 mm3, respectively. Both the stress and strain values exhibited two peaks during the flexion simulation. The highest value occurred at 30° of flexion, and the second highest value occurred at 90° of flexion. CONCLUSIONS: Increased ACL direct insertion coverage provided more positive biomechanical effects after anatomical single-bundle ACL reconstruction during knee flexion.

Research on the Variable-Temperature Cracking Mechanism of CRTS I Type Double-Block Ballastless Track on a Bridge.

The CRTS I type double-block ballastless track (CRTS I TDBBT) has the advantages of convenient construction and low cost, but it has low crack resistance and the temperature field distribution of the railway on the bridge is uneven and frequently changes, so it is necessary to study the mechanical properties of the CRTS I TDBBT under the load of a temperature field. The temperature field model of the CRTS I TDBBT on the bridge is established by finite element software, the real-time temperature field of the track bed slab is brought into the coupled model as a load, and the variation laws of the temperature stress of the CRTS I TDBBT under different schemes are compared. The temperature gradient in the CRTS I TDBBT track bed slab has the largest fluctuation range, and the positive and negative temperature gradient range can reach 93.34 °C. For the temperature longitudinal stress around the sleeper block of the track bed slab, the edge is the largest; the temperature longitudinal stress is reduced by at most 5.27% after the anti-cracking diagonal bars are added. When the expansion joint is added, the temperature stress can be reduced by up to 80.29%. The fluctuation range of the temperature gradient of the track bed is basically consistent with the fluctuation range of the local air temperature. The huge temperature difference leads to the occurrence of cracks in the track structure, and cracks are more likely to occur at the corners of the sleeper block. The addition of both anti-crack diagonal bars and expansion joints has an anti-crack effect, but the effect of adding expansion joints is better.

Path R-CNN for Prostate Cancer Diagnosis and Gleason Grading of Histological Images.

Prostate cancer is the most common and second most deadly form of cancer in men in the United States. The classification of prostate cancers based on Gleason grading using histological images is important in risk assessment and treatment planning for patients. Here, we demonstrate a new region-based convolutional neural network framework for multi-task prediction using an epithelial network head and a grading network head. Compared with a single-task model, our multi-task model can provide complementary contextual information, which contributes to better performance. Our model is achieved a state-of-the-art performance in epithelial cells detection and Gleason grading tasks simultaneously. Using fivefold cross-validation, our model is achieved an epithelial cells detection accuracy of 99.07% with an average area under the curve of 0.998. As for Gleason grading, our model is obtained a mean intersection over union of 79.56% and an overall pixel accuracy of 89.40%.

[Effects of 16 week aerobic exercise on baPWV and ABI of middle-aged and elderly patients].

OBJECTIVES: To investigate the interventional effects of 16-week aerobic exercises on the elderly's arteriosclerosis and its mechanism. METHODS: Twenty-seven elderly people with the average age of 62. 70 ±3. 26 joined a 16-week square dance/taijiquan exercise program that conducted 60 minutes each time, six times per week. Arterial stiffness and its related indexes such as systolic pressure(SBP), diastolic pressure(DBP), left brachial-ankle pulse wave velocity (L-baPWV), right brachial-ankle pulse wave velocity(R-baPWV), left ankle brachial index (L-ABI), right ankle brachial index(R-ABI), serum triglyceride(TG), total cholesterol(TC), high density lipoprotein cholesterol(HDL-c), low density lipoprotein cholesterol(LDL-c), superoxide dismutase(SOD), malondialdehyde(MDA) and glutathione peroxidase (GSH-Px) were detected at 3 time points including before exercise program, by the end of exercise for 8 weeks and 16 weeks. RESULTS: ① Compared with pre-exercise, the R-baPWV and R-ABI of the elderly people were decreased at the end of the 8th week, and the L-baPWV, RbaPWV, R-ABI and L-ABI were decreased significantly at the end of the 16th week. ②Compared with pre-exercise, SBP and DBP were declined markedly (P<0.01, P<0.05) at the end of the 8th week, SBP, DBP and pulse pressure were decreased significantly (P<0.01, P<0.05) at the end of the 16th week. ③Compared with pre-exercise, TC and LDL-c were declined markedly (P<0.01) at the end of the 8th and the 16th week, and there was no difference of the level of TG and LDL-c between pre-exercise and post-exercise. ④There was no evident difference of serum level of SOD, GSH-Px, MDA between pre-exercise and post-exercise at the end of the 8th week. Compared with pre-exercise, the level of serum SOD, GSH-Px was increased evidently while the content of serum MDA was decreased significantly (P<0.01). CONCLUSIONS: Sixteen-week aerobic exercises could reduce baPWV and ABI levels, regulate blood pressure, blood lipids and lipid peroxides levels of the elderly evidently, thus improve the controlling quality of atherosclerosis.

A Bayesian model for estimating multi-state disease progression.

A growing number of individuals who are considered at high risk of cancer are now routinely undergoing population screening. However, noted harms such as radiation exposure, overdiagnosis, and overtreatment underscore the need for better temporal models that predict who should be screened and at what frequency. The mean sojourn time (MST), an average duration period when a tumor can be detected by imaging but with no observable clinical symptoms, is a critical variable for formulating screening policy. Estimation of MST has been long studied using continuous Markov model (CMM) with Maximum likelihood estimation (MLE). However, a lot of traditional methods assume no observation error of the imaging data, which is unlikely and can bias the estimation of the MST. In addition, the MLE may not be stably estimated when data is sparse. Addressing these shortcomings, we present a probabilistic modeling approach for periodic cancer screening data. We first model the cancer state transition using a three state CMM model, while simultaneously considering observation error. We then jointly estimate the MST and observation error within a Bayesian framework. We also consider the inclusion of covariates to estimate individualized rates of disease progression. Our approach is demonstrated on participants who underwent chest x-ray screening in the National Lung Screening Trial (NLST) and validated using posterior predictive p-values and Pearson's chi-square test. Our model demonstrates more accurate and sensible estimates of MST in comparison to MLE.

An automated lung segmentation approach using bidirectional chain codes to improve nodule detection accuracy.

Computer-aided detection and diagnosis (CAD) has been widely investigated to improve radiologists׳ diagnostic accuracy in detecting and characterizing lung disease, as well as to assist with the processing of increasingly sizable volumes of imaging. Lung segmentation is a requisite preprocessing step for most CAD schemes. This paper proposes a parameter-free lung segmentation algorithm with the aim of improving lung nodule detection accuracy, focusing on juxtapleural nodules. A bidirectional chain coding method combined with a support vector machine (SVM) classifier is used to selectively smooth the lung border while minimizing the over-segmentation of adjacent regions. This automated method was tested on 233 computed tomography (CT) studies from the lung imaging database consortium (LIDC), representing 403 juxtapleural nodules. The approach obtained a 92.6% re-inclusion rate. Segmentation accuracy was further validated on 10 randomly selected CT series, finding a 0.3% average over-segmentation ratio and 2.4% under-segmentation rate when compared to manually segmented reference standards done by an expert.