Poster Session 2

ABSTRACT:

Background

Two wheel motorized vehicle-related injuries are a common cause of traumatic spine injury in the United States. Off-road motorcycles are predominately used in a recreational manner with crashes being either solo or in a race-type setting with similar vehicles. There is a lack of evidence in current literature describing the specific types of spine injury patterns from off-road motorcycle accidents. Better epidemiologic data can guide improvements in safety equipment, emergency response, and injury diagnosis.

Methods

A retrospective chart review was conducted from the University Medical Center (UMC) Trauma Registry spanning 2010-2019. ICD 9 and 10 injury codes and mechanism codes were used to obtain spinal injury data as well as off-road motorcycle mechanism. Radiology reports were then analyzed to identify specific injury diagnosis and severity. Injury patterns were then identified and reported as observational data.

Results

84 Off-road motorcycle spinal injury patients were identified based on ICD code analysis over the 10 year window. 68 patients met study inclusion criteria with a total of 197 acute traumatic spinal injuries. Cervical spine injuries were present in 18/68, thoracic spine injuries in 37/68, and lumbar spine injuries in 33/68. There were 17 multi-region spine trauma patients. Most common injuries were found to be transverse process fracture (N=71) and compression fracture (N=49).

Conclusions

In a local community-based level 1 trauma center, the predominant spinal injuries from off-road motorcycle crashes were localized to the thoracic and lumbar regions. Further insight to injury patterns can help guide development of future safety equipment as well as emergency response and diagnosis.

ABSTRACT:

Objective: Given the high prevalence of concussion in sports, high school athletes are administered ImPACT at the start of their sport season and again after suspected concussion. Concussion management involves comparison of baseline and post-injury cognitive scores with declines in scores providing evidence for concussive injury. A network framework may identify post-concussive cognitive changes not apparent when composite scores are compared. We examined composite-level and network-level comparisons to determine whether network models could better characterize sport-related concussion (SRC) recovery.

Method: High school athletes (N=1,553) were administered ImPACT at preseason (T1), post-SRC (T2=72 hours of injury), and prior to return-to-play (T3=within two weeks post-injury). Composite scores and networks were calculated and compared across time. Centrality indices were calculated to determine relative importance of cognitive variables within networks.

Results: Composite scores declined acutely post-SRC and improved by T3. Network connectivity increased from T1 to T2 and remained hyperconnected at T3. There was evidence of network reorganization between T1 and T3. Processing speed was central within each network, and visual memory and impulsivity became more central over time.

Conclusions: While composite scores suggested cognitive recovery, cognitive network comparisons revealed lingering network changes at two-weeks post-injury. Network changes may reflect compensatory strategies to achieve task demands due to underlying disruption. Results highlight importance of targeting visual memory and impulsivity during treatment. Network analysis provides nuanced information about cognitive recovery following SRC, thereby providing an effective means of monitoring persistent cognitive symptoms after concussion.

ABSTRACT:

Purpose: To predict the increased risk of severe cardiotoxicities in breast cancer patients receiving chemoradiotherapy could be challenging due to the variations of current methods used to evaluate the cardiotoxicity symptoms, individual susceptibility and early markers of injury. In this study, we developed, for the first time to our knowledge, a Light Gradient Boosting Machine (LightGBM)-enabled predictive model to integrate patients’ chart from electronic medical records (EMRs) for early prediction of severe cardiotoxicities.

Methods: A total of 179 breast cancer patients were included. The patients were randomly partitioned to the training set and the validation set for LightGBM predictive model development and validation. The training features extracted from each patient include 18 different clinical factors. The utility of the LightGBM model constructed in predicting severe cardiotoxicities was evaluated by ROC analysis in the validation set.

Results: The AUC value of the LightGBM predictive model achieved in the validation set was 0.82. The feature important analysis showed that age, the LVEF value before chemoradiotherapy, cancer position, targeted therapy, tumor stage and hormone therapy were the most valuable risk predicting factors.

Conclusion: The LightGBM framework proposed herein affords a means to use EMR data to individualize the prediction of severe cardiotoxicities at point of care of patients with breast cancer receiving chemoradiotherapy, which can facilitate the identification of patients for whom early intervention are warranted before the therapy, thus potentially improving the utility of chemoradiotherapy for breast cancer from a precision treatment perspective.

ABSTRACT:

Introduction

Histology diagnosis of non-small-cell lung cancer (NSCLC) depends on the location and the size of the biopsies. Non-small-cell lung cancer-not otherwise specified (NSCLC-NOS) is a common histologic diagnosis due to uncertain histological subtype. Accurate survival prediction helps clinicians and patients in decision-making about cancer management and improves patient outcomes. We proposed to apply light gradient machine (LightGBM)-enabled radiomics model for survival prediction in NSCLC-NOS.

Methods

Planning CT image sets, contour data sets and clinical data sets of 61 NSCLC-NOS patients from The Cancer Imaging Archive were included. Patients were labeled into three groups according to their survival times into less than 1 year, 1-3 years and more than 3 years. Demographic features and clinical features were extracted from the clinical data sets. Radiomic features were extracted from the CT image of the contoured planning tumor volume using PyRadiomics. LightGBM was trained to predict the survival outcome for individual patients. Feature importance was analyzed based on information gain from features in LightGBM model. Model performance was evaluated using F1 score and average area under the curve (AUC).

Results

A total of 107 radiomic features, 2 demographic features and 11 clinical features were extracted. Among the top 20 features in feature importance analysis, age is the only non-radiomics feature. Results show significantly different performances in predicting survival with an AUC value of 0.97, 0.83 and 0.87, respectively, in the groups of less than 1 year, 1-3 years and more than 3 years.

ABSTRACT:

Horner syndrome is caused by a disruption to any part of the oculosympathetic nerve supply, and is classically characterized by the triad of ptosis, miosis, and facial anhidrosis. The underlying causes of Horner's syndrome vary greatly and may include a tumor, stroke, injury, or underlying disease affecting the areas surrounding the sympathetic nerves. In this interesting case presentation, an 18-year-old male with a history of polysubstance abuse was found unresponsive in his car after a motor vehicle accident. Upon examination on hospitalization day 4, he was found to have new onset anisocoria with a miotic left pupil and left-sided ptosis consistent with Horner’s syndrome. MRI neck soft tissues revealed abnormal increased signal intensity consistent with edema in the left longus colli muscle from C4 to C7 with enhancement, as well as edema in the right longus colli muscle from T6 to T7 with enhancement. This suggests Horner syndrome secondary to edema of the longus colli muscle, as the second-order sympathetic innervation to the eye runs under the longus colli. To the best of our knowledge, this is the first case of bilateral asymmetric longus colli inflammation causing Horner’s syndrome.

ABSTRACT:

Horner syndrome is caused by a disruption to any part of the oculosympathetic nerve supply, and is classically characterized by the triad of ptosis, miosis, and facial anhidrosis. The underlying causes of Horner's syndrome vary greatly and may include a tumor, stroke, injury, or underlying disease affecting the areas surrounding the sympathetic nerves. In this interesting case presentation, an 18-year-old male with a history of polysubstance abuse was found unresponsive in his car after a motor vehicle accident. Upon examination on hospitalization day 4, he was found to have new onset anisocoria with a miotic left pupil and left-sided ptosis consistent with Horner’s syndrome. MRI neck soft tissues revealed abnormal increased signal intensity consistent with edema in the left longus colli muscle from C4 to C7 with enhancement, as well as edema in the right longus colli muscle from T6 to T7 with enhancement. This suggests Horner syndrome secondary to edema of the longus colli muscle, as the second-order sympathetic innervation to the eye runs under the longus colli. To the best of our knowledge, this is the first case of bilateral asymmetric longus colli inflammation causing Horner’s syndrome.