Can AI outperform a junior resident? Comparison of deep neural network to first-year radiology residents for identification of pneumothorax.

PURPOSE: To (1) develop a deep learning system (DLS) using a deep convolutional neural network (DCNN) for identification of pneumothorax, (2) compare its performance to first-year radiology residents, and (3) evaluate the ability of a DLS to augment radiology residents by detecting missed pneumothoraces. METHODS: This was a retrospective study performed in September 2018. We obtained 112,120 chest radiographs (CXRs) from the NIH ChestXray14 database, of which 4360 cases (4%) were labeled as pneumothorax by natural language processing. We utilized 111,518 CXRs to train and validate the ResNet-152 DCNN pretrained on ImageNet to identify pneumothorax. DCNN testing was performed on a hold-out set of 602 CXRs, whose groundtruth was determined by a cardiothoracic radiologist. Two first-year radiology residents evaluated the test CXRs for presence of pneumothorax. Receiver operating characteristic (ROC) curves were generated for each evaluator with area under the curve (AUC) compared using the DeLong parametric method. RESULTS: The DCNN achieved AUC of 0.841 for identification of pneumothorax at a rate of 1980 images/min. In contrast, both first-year residents achieved significantly higher AUCs of 0.942 and 0.905 (p < 0.01 for both compared to DCNN), but at a slower rate of two images/min. The DCNN identified 3 of 31 (9.7%) additional pneumothoraces missed by at least one of the residents. CONCLUSION: A DLS for pneumothorax identification had lower AUC than 1st-year radiology residents, but interpreted images > 1000× as fast and identified 3 additional pneumothoraces missed by the residents. Our findings suggest that DLS could augment radiologists-in-training to identify potential urgent findings.

Unsuspected pheochromocytoma incidentally found on chest CT.

A 51-year-old woman with history of migraine headaches and intermittent nausea, vomiting, palpitations, and diaphoresis presented to the emergency department with hypertensive emergency 1 month after starting a beta blocker for migraine prophylaxis. Contrast-enhanced computed tomography of the chest incidentally revealed a large abdominal mass in the area of the left adrenal gland. Iodine-123 metaiodobenzylguanidine scan imaging showed localized uptake into the left adrenal gland. Along with imaging results, laboratory testing confirmed the diagnosis of pheochromocytoma. The patient was treated with blood pressure control, specifically alpha blockade, and surgical excision of the mass. This case displays a typical clinical presentation of pheochromocytoma coupled with atypical radiographic size and appearance.

Full Step Cycle Kinematic and Kinetic Comparison of Barefoot Walking and a Traditional Shoe Walking in Healthy Youth: Insights for Barefoot Technology.

OBJECTIVE: Barefoot technology shoes are becoming increasingly popular, yet modifications are still needed. The present study aims to gain valuable insights by comparing barefoot walking to neutral shoe walking in a healthy youth population. METHODS: 28 healthy university students (22 females and 6 males) were recruited to walk on a 10-meter walkway both barefoot and in neutral running shoes at their comfortable walking speed. Full step cycle kinematic and kinetic data were collected using an 8-camera motion capture system. RESULTS: In the early stance phase, the knee extension moment (MK1), the first peak absorbed joint power at the knee joint (PK1), and the flexion angle of knee/dorsiflexion angle of the ankle were significantly reduced when walking in neutral running shoes. However, in the late stance, barefoot walking resulted in decreased hip joint flexion moment (MH2), second peak extension knee moment (MK3), hip flexors absorbed power (PH2), hip flexors generated power (PH3), second peak absorbed power by knee flexors (PK2), and second peak anterior-posterior component of joint force at the hip (APFH2), knee (APFK2), and ankle (APFA2). CONCLUSIONS: These results indicate that it should be cautious to discard conventional elements from future running shoe designs and rush to embrace the barefoot technology fashion.

Neuroplasticity in post-stroke gait recovery and noninvasive brain stimulation.

Gait disorders drastically affect the quality of life of stroke survivors, making post-stroke rehabilitation an important research focus. Noninvasive brain stimulation has potential in facilitating neuroplasticity and improving post-stroke gait impairment. However, a large inter-individual variability in the response to noninvasive brain stimulation interventions has been increasingly recognized. We first review the neurophysiology of human gait and post-stroke neuroplasticity for gait recovery, and then discuss how noninvasive brain stimulation techniques could be utilized to enhance gait recovery. While post-stroke neuroplasticity for gait recovery is characterized by use-dependent plasticity, it evolves over time, is idiosyncratic, and may develop maladaptive elements. Furthermore, noninvasive brain stimulation has limited reach capability and is facilitative-only in nature. Therefore, we recommend that noninvasive brain stimulation be used adjunctively with rehabilitation training and other concurrent neuroplasticity facilitation techniques. Additionally, when noninvasive brain stimulation is applied for the rehabilitation of gait impairment in stroke survivors, stimulation montages should be customized according to the specific types of neuroplasticity found in each individual. This could be done using multiple mapping techniques.

Dance recognition system using lower body movement.

The current means of locating specific movements in film necessitate hours of viewing, making the task of conducting research into movement characteristics and patterns tedious and difficult. This is particularly problematic for the research and analysis of complex movement systems such as sports and dance. While some systems have been developed to manually annotate film, to date no automated way of identifying complex, full body movement exists. With pattern recognition technology and knowledge of joint locations, automatically describing filmed movement using computer software is possible. This study used various forms of lower body kinematic analysis to identify codified dance movements. We created an algorithm that compares an unknown move with a specified start and stop against known dance moves. Our recognition method consists of classification and template correlation using a database of model moves. This system was optimized to include nearly 90 dance and Tai Chi Chuan movements, producing accurate name identification in over 97% of trials. In addition, the program had the capability to provide a kinematic description of either matched or unmatched moves obtained from classification recognition.

Activity classification in users of ankle foot orthoses.

A framework for activity classification using inertial sensors mounted on ankle foot orthoses (AFOs) is presented. A decision tree-nearest neighbor algorithm classifies activities using subject-specific training. Eight volunteer subjects wore modified bilateral AFOs with shank and foot mounted triaxial accelerometers and gyroscopes. The AFOs were fitted with hardware to induce different gait perturbations: free rotation of the ankle, plantarflexion or "equinus" gait, and locked ankle joint. For each condition, the subject performed eight gait activities at varied slopes and standing, sitting, and lying postures. Using video for ground truth, the algorithm had an overall mean sensitivity of 95% using 50% of the data (≈ 140 s) for training and demonstrated upwards of 90% sensitivity with 25% of the data (≈ 70 s) for training. High sensitivities (≥ 87%) and PPV (≥ 90%) were achieved for all annotated gait patterns for all perturbations, excluding stair climbing (63%, 77%) and descending (80%, 78%). Postures were classified with less sensitivity and PPV than gait activities: lying (98%, 93%), standing (80%, 84%) and sitting (64%, 75%). Non-annotated walking (68%) and standing (73%) were classified with less sensitivity than were corresponding annotated events. Our results indicate that AFOs are a suitable sensor platform for future research in activity classification and gait monitoring in AFO users with perturbed gait using limited training data.

Changes in kinematics, metabolic cost and external work during walking with a forward assistive force.

We examined how the application of a forward horizontal force applied at the waist alters the metabolic cost, kinematics, and external work of gait. Horizontal assist forces of 4%, 8% and 12% of a subject's body weight were applied via our testing apparatus while subjects walked at comfortable walking speed on a level treadmill. Kinematic and metabolic parameters were measured using motion capture and ergospirometry respectively on a group of 10 healthy male subjects. Changes in kinematic and metabolic parameters were quantified and found similar to walking downhill at varying grades. A horizontal assist force of 8% resulted in the greatest reduction of metabolic cost. Changes in recovery factor, external work, and center of mass (COM) movement did not correlate with changes in metabolic rate and therefore were not driving the observed reductions in cost. The assist force may have performed external work by providing propulsion as well as raising the COM as it pivots over the stance leg. Assist forces may decrease metabolic cost by reducing the concentric work required for propulsion while increasing the eccentric work of braking. These findings on the effects of assist forces suggest novel mobility aids for individuals with gait disorders and training strategies for athletes.

The effects of ankle foot orthoses on energy recovery and work during gait in children with cerebral palsy.

BACKGROUND: Studies suggest that 50% of children with cerebral palsy are prescribed ankle foot orthoses. One of the aims of ankle foot orthosis use is to aid in walking. This research examined the effects that ankle foot orthoses have on the energy recovery and the mechanical work performed by children with cerebral palsy during walking. METHODS: Twenty-one children with spastic diplegia walked with and without their prescribed bilateral ankle foot orthoses. Ten of the subjects wore articulated (hinged) orthoses and 11 subjects wore solid orthoses. Three dimensional kinematic data were collected and between and within group repeated measures ANOVAs were applied to the dependent measures. FINDINGS: The results were similar for both groups. There was an increase in stride length, energy recovery, and potential energy and the kinetic energy variation. There was no change in the mechanical work performed to walk or the normalized center of mass vertical excursion. Unfortunately, the increase in energy recovery did not alter the external work, as it was offset by increased variation in the potential and kinetic energies of the center of mass. There was a great deal of variability in the measured work, with both large increases and decreases in the work of individual subjects when wearing orthoses. INTERPRETATION: These results suggest that current ankle foot orthoses can reduce the work to walk, but do not do so for many children with cerebral palsy. This research suggests that ankle foot orthosis prescription could be aided by measuring the mechanical work during walking.

The gait of children with and without cerebral palsy: work, energy, and angular momentum.

This paper describes a method to characterize gait pathologies like cerebral palsy using work, energy, and angular momentum. For a group of 24 children, 16 with spastic diplegic cerebral palsy and 8 typically developed, kinematic data were collected at the subjects self selected comfortable walking speed. From the kinematics, the work-internal, external, and whole body; energy-rotational and relative linear; and the angular momentum were calculated. Our findings suggest that internal work represents 53% and 40% respectively of the whole body work in gait for typically developed children and children with cerebral palsy. Analysis of the angular momentum of the whole body, and other subgroupings of body segments, revealed a relationship between increased angular momentum and increased internal work. This relationship allows one to use angular momentum to assist in determining the kinetics and kinematics of gait which contribute to increased internal work. Thus offering insight to interventions which can be applied to increase the efficiency of bipedal locomotion, by reducing internal work which has no direct contribution to center of mass motion, in both normal and pathologic populations.

Angular momentum of walking at different speeds.

Recently, researchers in robotics have used regulation of the angular momentum of body segments about the total body center of mass (CoM) to develop control strategies for bipedal gait. This work was spurred by reports finding that for a "large class" of human movement tasks, including standing, walking, and running the angular momentum is conserved about the CoM. However, there is little data presented to justify this position. This paper describes an analysis of 11 male adults walking overground at 0.7, 1.0, and 1.3 times their comfortable walking speed (CWS). The normalized angular momenta about the body CoM of 12 body segments were computed about all three coordinate axes. The normalized angular momenta were both small (<0.03) and highly regulated for all subjects and walking speed with extrema that negatively correlated with walking speeds. It was found that the angular momentum of the body about its CoM during walking could be described by a small number of principal components. For the adult walkers the first three principal components accounted for more than 97% of the variability of the angular momentum about each of the three principal axes at all walking speeds. In addition, it was found that the orthogonal principal components at each speed and for each subject were similar, i.e., the vectors of the principal components at each speed and for each subject were co-linear.

Angular momentum synergies during walking.

We studied the coordination of body segments during treadmill walking. Specifically, we used the uncontrolled manifold hypothesis framework to quantify the segmental angular momenta (SAM) synergies that stabilize (i.e., reduce the across trials variability) the whole body angular momentum (WBAM). Seven male subjects were asked to walk over a treadmill at their comfortable walking speed. A 17-segment model, fitted to the subject's anthropometry, was used to reconstruct their kinematics and to compute the SAM and WBAM in three dimensions. A principal component analysis was used to represent the 17 SAM by the magnitudes of the first five principal components. An index of synergy (DeltaV) was used to quantify the co-variations of these principal components with respect to their effect on the WBAM. Positive values of DeltaV were observed in the sagittal plane during the swing phase. They reflected the synergies among the SAM that stabilized (i.e., made reproducible from stride to stride) the WBAM. Negative values of DeltaV were observed in both frontal and sagittal plane during the double support phase. They were interpreted as "anti-synergies", i.e., a particular organization of the SAM used to adjust the WBAM. Based on these results, we demonstrated that the WBAM is a variable whose value is regulated by the CNS during walking activities, and that the nature of the WBAM control changed between swing phase and double support phase. These results can be linked with humanoid gait controls presently employed in robotics.

Effects of balance training on gait parameters in patients with chronic ankle instability: a randomized controlled trial.

OBJECTIVE: To examine the effects of a four-week balance training programme on ankle kinematics during walking and jogging in those with chronic ankle instability. A secondary objective was to evaluate the effect of balance training on the mechanical properties of the lateral ligaments in those with chronic ankle instability. DESIGN: Randomized controlled trial. SETTING: Laboratory. SUBJECTS/PATIENTS: Twenty-nine participants (12 males, 17 females) with self-reported chronic ankle instability were randomly assigned to a balance training group or a control group. INTERVENTION: Four weeks of supervised rehabilitation that emphasized dynamic balance stabilization in single-limb stance. The control group received no intervention. MAIN OUTCOME MEASURES: Kinematic measures of rearfoot inversion/eversion, shank rotation, and the coupling relationship between these two segments throughout the gait cycle during walking and jogging on a treadmill. Instrumented ankle arthrometer measures were taken to assess anterior drawer and inversion talar tilt laxity and stiffness. RESULTS: No significant alterations in the inversion/eversion or shank rotation kinematics were found during walking and jogging after balance training. There was, however, a significant decrease in the shank/rearfoot coupling variability during walking as measured by deviation phase after balance training (balance training posttest: 13.1 degrees +/- 6.2 degrees , balance training pretest: 16.2 degrees +/- 3.3 degrees , P = 0.03), indicating improved shank/rearfoot coupling stability. The control group did not significantly change. (posttest: 16.30 degrees +/- 4.4 degrees , pretest: 18.6 degrees +/- 7.1 degrees , P40.05) There were no significant changes in laxity measures for either group. CONCLUSIONS: Balance training significantly altered the relationship between shank rotation and rearfoot inversion/eversion in those with chronic ankle instability.

Balance training improves function and postural control in those with chronic ankle instability.

PURPOSE: The purpose of this randomized controlled trial was to determine the effect of a 4-wk balance training program on static and dynamic postural control and self-reported functional outcomes in those with chronic ankle instability (CAI). METHODS: Thirty-one young adults with self-reported CAI were randomly assigned to an intervention group (six males and 10 females) or a control group (six males and nine females). The intervention consisted of a 4-wk supervised balance training program that emphasized dynamic stabilization in single-limb stance. Main outcome measures included the following: self-reported disability on the Foot and Ankle Disability Index (FADI) and the FADI Sport scales; summary center of pressure (COP) excursion measures including area of a 95% confidence ellipse, velocity, range, and SD; time-to-boundary (TTB) measures of postural control in single-limb stance including the absolute minimum TTB, mean of TTB minima, and SD of TTB minima in the anteroposterior and mediolateral directions with eyes open and closed; and reach distance in the anterior, posteromedial, and posterolateral directions of the Star Excursion Balance Test (SEBT). RESULTS: The balance training group had significant improvements in the FADI and the FADI Sport scores, in the magnitude and the variability of TTB measures with eyes closed, and in reach distances with the posteromedial and the posterolateral directions of the SEBT. Only one of the summary COP-based measures significantly changed after balance training. CONCLUSIONS: Four weeks of balance training significantly improved self-reported function, static postural control as detected by TTB measures, and dynamic postural control as assessed with the SEBT. TTB measures were more sensitive at detecting improvements in static postural control compared with summary COP-based measures.

Center of mass motion and the effects of ankle bracing on metabolic cost during submaximal walking trials.

The goal of this research was to examine the relationship between center of mass (CoM) motion and metabolic cost and to assess secondarily the effects of equinus gait on metabolic cost during walking trials at various velocities. Twelve (n=12) healthy male adults walked in four brace conditions, including fixed equinus, at three different walking speeds. Metabolic measures and 3D kinematic data were collected during each trial. Significant main effects for velocity were observed on both dependent measures, net O2 cost (p<0.001), and CoM vertical excursion (CoMz) (p<0.01). Correlation analysis showed significant positive correlations between net O2 cost and CoMz that were strongest at fast velocities. Further, analysis revealed the cost of walking at a comfortable speed in the equine position was 28% greater than walking in braces without springs or springs in neutral position and 64% greater than walking in shoes alone. CoMz does correlate with metabolic cost as measured by net O2 cost. Also, normal subjects walking in equinus have an increased net O2 cost and increased CoMz when compared to walking with shoes alone or the other brace conditions at all velocities.

Center of mass movement and energy transfer during walking in children with cerebral palsy.

OBJECTIVE: To gain insight into the mechanical inefficiencies of gait patterns used by children with spastic diplegia by analysis of center of mass (COM) movement and energy recovery. DESIGN: Prospective study using between-group measures to analyze differences between children with cerebral palsy (CP) and age-matched controls without CP. SETTING: Assessments were performed in a gait laboratory. PARTICIPANTS: Fifteen children with spastic diplegia and 6 age-matched controls without CP with a mean age of 9.7 years. INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: Gait data assessed included temporal-distance factors, COM vertical excursion, work done on the COM, and the percentage of energy transferred and relative phase between the potential and kinetic energy. RESULTS: Children with CP had a 33% smaller energy recovery factor than the controls (P<.001). They also had 60% greater COM vertical excursion (P<.02) and a poorer phasic relation between potential and kinetic energies (P<.02), both of which contributed to greater mechanical work performed (P<.003). CONCLUSIONS: Compared with the age-matched controls without CP, the children with CP were mechanically less efficient in their gait. Interventions that promote heel contact and roll over and greater knee stability to better utilize the kinetic energy of push-off could improve walking efficiency.

Biomechanical changes in gait following selective dorsal rhizotomy.

OBJECT: The purpose of this study was to evaluate the effects of selective dorsal rhizotomy (SDR) on the walking pattern of children with cerebral palsy (CP). The authors hypothesized that SDR would reduce the overactive stretch reflex and result in improved joint positions, greater joint angular velocities and motion, and greater strides. The authors also expected that the changes from the SDR would be reflected in the electrical activities of the muscles. METHODS: Twenty children with CP participated in the study. Ten underwent SDR and 10 who met the inclusion criteria but who did not undergo SDR formed a control group. Gait analysis was performed twice in both groups at comparable time intervals. Within-group (pre- and post-SDR) assessments and between-group (SDR compared with no SDR) assessments were performed. The results indicated that strides increased 1.4 cm on average (p < 0.001), hip and knee excursion increased as a result of greater hip and knee flexion in swing, and ankle dorsiflexion in stance improved. Hip-flexion and knee-extension velocities were increased in swing. Electromyography changes were consistent with these gait changes and support the authors' hypothesis. CONCLUSIONS: By attenuating the stretch response, SDR alters gait mechanics. The effects of SDR include greater strides, joint excursions, and joint angular velocity.

Reports of body pain in a dental student population.

BACKGROUND: Dentists experience more neck, shoulder and lower back pain than do practitioners in other occupational groups. The authors examined the prevalence of musculoskeletal pain in dental students, by sex. METHODS: The authors investigated the body distribution and severity of reported musculoskeletal pain in a population of dental students, considering increased exposure to clinical experience with years in dental school. A total of 271 dental students in all four school years completed a questionnaire focusing on pain reported in five general body regions. RESULTS: Forty-six to 71 percent of students reported body pain, with the percentage generally increasing with years in dental school. Women reported having the worst pain in their neck/shoulder region (chi2, P = .004); men reported having the worst pain in their mid- to lower back regions (chi2, P = .015). Frequency and daily duration of the worst pain were higher in the third year of dental school than in the first year (Bonferroni test, P = .014 and P = .001, respectively), as was the persistence (in months) of the most symptomatic body pain (P = .001). Pain intensity was higher for women than for men (two-way analysis of variance, P < .05). The perception of how performance of dental procedures affects pain increased significantly with number of years in dental school (P = .001). CONCLUSIONS: Chronic musculoskeletal pain appears early in dental careers, with more than 70 percent of dental students of both sexes reporting pain by their third year. CLINICAL IMPLICATIONS: Musculoskeletal pain is a common complaint of dental professionals that may lead to serious physical disability. Since this type of pain occurs early in dental training, dentistry is obligated to further examine the mental, physical and ergonomic factors that may be contributory.

Low-back biomechanics and static stability during isometric pushing.

Pushing and pulling tasks are increasingly prevalent in industrial workplaces. Few studies have investigated low-back biomechanical risk factors associated with pushing, and we are aware of none that has quantified spinal stability during pushing exertions. Data recorded from 11 healthy participants performing isometric pushing exertions demonstrated that trunk posture, vector force direction of the applied load, and trunk moment were influenced (p < .01) by exertion level, elevation of the handle for the pushing task, and foot position. A biomechanical model was used to analyze the posture and hand force data gathered from the pushing exertions. Model results indicate that pushing exertions provide significantly (p < .01) less stability than lifting when antagonistic cocontraction is ignored. However, stability can be augmented by recruitment of muscle cocontraction. Results suggest that cocontraction may be recruited to compensate for the fact that equilibrium mechanics provide little intrinsic trunk stiffness and stability during pushing exertions. If one maintains stability by means of cocontraction, additional spinal load is thereby created, increasing the risk of overload injury. Thus it is important to consider muscle cocontraction when evaluating the biomechanics of pushing exertions. Potential applications of this research include improved assessment of biomechanical risk factors for the design of industrial pushing tasks.