fhberger@gmail.com

1) To be familiar with currently worldwide accepted protocols in polytrauma CT imaging. 2) To know clinical conditions requiring whole-body CT. 3) To comprehend the selection of trauma patients for targeted CT examinations.

In the western world, polytrauma is the major cause of mortality in people under 45 years of age. Furthermore, it is a major contributor to loss of quality of life and ability to work. The setting of polytrauma is almost always chaotic, not a favourable environment to come to timely diagnosis and treatment. To decrease morbidity and mortality, time is everything. It is our job as radiologist to contribute to the trauma team and help facilitate timely diagnosis - and in many cases, also timely treatment by interventional radiology. To reach the best treatment strategy for the patient as quickly and accurately as safely possible, is the goal. In this update on imaging of polytrauma patients, the focus is on the role of CT to achieve this goal. With the progress in CT scanner development, different protocol options arise. Which CT protocols are being used and what factors do they depend upon? In addition, there is a widespread increase in use of whole body CT internationally, is this a good thing or should we be more selective? What is the current evidence to select patients for targeted CT examinations in polytrauma? A lot of these questions have not been definitively resolved. This lecture aims to provide an update of the current insights into the use of CT for trauma care, with the goal to choose wisely on how to investigate the polytrauma patient in a timely and meaningful fashion.

http://abstract.rsna.org/uploads/2019/19000911/Active RC608-01.pdf

jshyu@bwh.harvard.edu

Indications for whole-body trauma CT are unclear. This study evaluates patients transferred to a level 1 trauma center, who had selective CT at the originating hospital and completion whole-body CT at the accepting hospital, to determine if additional CT imaging detects clinically significant injury.

This was a single center study at a level 1 trauma center with a dedicated Emergency Radiology division. 243 consecutive trauma patients transferred from outside hospitals were included from 9/6/2015 to 12/20/2015. A review of the patient's acute traumatic injuries was obtained from chart reviews, radiology reports, and abbreviated injury scale (AIS). Whole-body CT was defined as CTs of the head, cervical spine, chest, abdomen, and pelvis. A patient is considered to have had 'completion' CT imaging if she or he obtained some of the whole-body CT components at the outside institution, and the rest at the accepting institution. Injuries that were detectable with radiographs (such as extremity fractures) were excluded.

35 received whole-body CT at the outside institution, and 45 received completion CT at the accepting institution. Of those who received completion CT, 13 (29%) had additional injuries on completion CTs that were not detected on CTs or radiographs from the outside institution. An additional 9 patients had indeterminate injuries in the radiology report that were not given a corresponding AIS. The additional injuries with AIS scores were subdural hemorrhage (1 patient), rib fractures (5), clavicle fracture (1), and thoracic (4) and lumbar (5) spine fractures. One patient who died in the trauma completion group had a lumbar spine fracture found on completion imaging, not considered to be the primary cause of death. Average ISS of transfer patients who received whole-body CTs at the outside institution was 13.9, compared to 10.6 for the completion group. A statistically significant difference between ISS was found between the transfer whole-body group and completion CT group (p = 0.044).

Completion whole-body CT for trauma transfer patients detects additional injuries in 29% of patients. Rib and spinal fractures are the most commonly detected injuries. Further work is needed to determine if this increase in diagnostic yield translated into patient management changes.

This study clarifies the role of whole-body completion CT for patients with major trauma.

foryourmiracle@gmail.com

Supervised learning has limitation in that it requires a large amount of annotated data. The purpose of this study is to determine if anomaly detection with generative adversarial networks (AnoGANs) are useful for detecting multiple various traumatic lesions on whole-body CT (WBCT) in an unsupervised manner.

We trained a Progressive Growing of GAN (PGGAN) to generate realistic artificial CT images, using the training set of 11,775 normal chest and/or abdominopelvic CT scans (172,249 chest slices and 301,584 abdominopelvic slices). Test set consisted of total 200 axial slices of WBCT images (100 abnormal and 100 normal images) in trauma patients. Using our simplified AnoGAN model, PGGAN-trained generator yields a corresponding realistic fake image to a given test image by minimizing mean square error between the fake and the test images. The differences between the fake and the test image on attention maps can detect and localize abnormal findings. For evaluation of the detection performance, we defined 7 clinically significant traumatic lesions (hemothorax, hemomediastinum, pneumothorax, pneumomediastinum, hemoperitoneum, hemoretroperitoneum and pneumoperitoneum). If the attention map partially included the traumatic lesions, it was considered a positive detection.

Total sensitivity per slice was 95.0% (95/100) and total sensitivity per lesions was 94.4% (135/143). For each traumatic lesion, sensitivity was 100% for hemothorax, 95.2% for hemomediastinum, 95.5% for pneumothorax, 93.3% for hemoperitoneum, 84.6% for hemoretroperitoneum, and 100% for pneumoperitoneum. Evaluation of other parameters of performance was limited due to difficult quantification and calculation of non-pathologic false positives.

We suggest that unsupervised learning of GANs using healthy dataset can be used to detect multiple traumatic lesions on unseen data and has high sensitivity to detect anomalies.

We propose that this model can be useful to develop deep learning algorithm to screen emergency or traumatic patients with multiple various lesions.

ssteenbu@iuhealth.org

Near hanging injuries are included as a high risk mechanism for the development of blunt cerebrovascular injuries (BCVI), despite there being a paucity of evidence-based data in support of this practice. As a result, this group of patients has been coalesced under the BCVI group which includes a myriad of different mechanisms of injury. The purpose of this study was to determine the incidence of BCVI in a large series of self-inflicted hanging patients who received neck CTA, and to guide appropriate diagnostic imaging in this specific group.

A 10-year retrospective review of self-inflicted hanging patients who received neck CTA at two urban Level 1 trauma centers was performed. The medical record was used to confirm self-inflicted hanging mechanism of injury, as well as key demographic data, airway status, physical exam findings, neurological status and deficits, drug screen results, and mortality. Neck CTA were evaluated for neck arterial injuries, cervical spine fractures and signs of ligamentous injury. CT Head and/or MRI brain exams performed during hospitalization were evaluated for infarction and ischemic brain injury. A Fisher's exact test was used to compare variables associated with patients with positive versus negative neck CTA exams with p < 0.05 considered statistically significant.

A total of 151 patients (mean age 31.6 years) of which 113 were male (74.8%) were included for analysis. Five patients (3.3%) were diagnosed with BCVI. A total of 74% had abnormal neck examination, 64% had abnormal drug screen, 63% had GCS <15, 33% were intubated, 30% had abnormal neurologic examination, 15.2% had anoxic brain injury resulting in death, and 0% had cervical spine fracture or ligamentous injury. Neurological deficit (p=0.027), and mortality (p=0.03) were significantly higher in CTA positive patients, while abnormal neck examination (p=1.0), positive drug screen (p=1.0) and intubation (p=0.33) were not significantly different.

The incidence of BCVI among patients with self-inflicted hanging was 3.3%. A total of 15% of patients died due to anoxic brain injury.

The incidence of BCVI in the setting of self-inflicted hanging is similar to that seen in other high risk mechanisms of injury. Thus including hanging injuries as a high risk mechanism for screening neck CTA remains prudent. Death due to anoxic brain injury poses a greater risk than that of BCVI.

The purpose of the study is to determine the value of liver injury CT grade in predicting the potential for subacute/late complications, and to determine the ideal timing of follow-up (FU) CT imaging to detect complications.

From August 2017 to July 2018, 58 major trauma patients (Pts) were diagnosed with liver injury. In this retrospective observational study, the admission CT and relevant clinical data were available for 53 Pts (43 male, 10 female; mean age 37.2 years ±18.2). Hepatic injuries detected on the admission CT were graded by two trauma radiologists using the AAST grading system. Mechanism of injury, liver-related subacute/late complications, and timing of follow-up CT imaging were reviewed.

The mechanisms of injury were as follows: vehicle incident/collision (n=25), fall >2 m (n=16), fall <2 m (n=1), penetrating trauma (n=10), rugby injury (n=1). There were 6 grade I liver injuries, 14 grade II, 14 grade III, 15 grade IV, and 4 grade V. Two Patients died within 30 days from presentation. Liver-related complications were observed in 10 patients (see Table) and included bilomas, biliary stricture and vascular complications. A statistically significant correlation between penetrating trauma and the occurrence of complications was observed (p<0.014). No correlation was observed between the injury grade and the trauma mechanism or the occurrence of complications. In 50% of cases, the complication was identified at FU CT within 7 days from the trauma (mean 6 days, range 5-7), in 50% of cases it was identified at further FU CT (mean 14 days, range 9-55).

Independent of the CT injury grade, a higher incidence of liver related complications occured with penetrating than a blunt mechanism of trauma. An initial follow-up CT between 5 and 7 days after the trauma is adequate to reveal early liver-related complications, but a subsequent FU CT within 15 days is recommended to detect complications in those patients with high grade liver injury.

A follow-up CT 5-7 days after traumatic liver injury is adequate to reveal early complications, a FU CT within 15 days is recommended in patients with high-grade injury and in penetrating liver trauma

patlas@hhsc.ca

1) To review the radiological and surgical literature of the potential pitfalls in diagnosis of diaphragmatic injuries. 2) To describe direct and indirect signs of blunt and penetrating diaphragmatic injury. 3) To highlight factors affection detection of diaphragmatic injuries.

dr.meokeeffe@gmail.com

1) Review imaging pearls and pitfalls in the assessment of mesenteric injury in trauma patients. 2) Focus on the anatomy of the small and large bowel mesentery, patterns of mesenteric injury, and their appearance on MDCT. 3) Review specific CT appearance of isolated mesenteric injury and polytrauma cases.

The small and large bowel mesentery are all too frequently underestimated as potential sites of significant injury in the trauma patient. In fact many would now argue that the mesentery itself has enough individual anatomical components and physiological roles to be considered a separate organ within the human body . As such we need to review the mesentery as a unique anatomical entity . It demonstrates a recognizable pattern of injury on CT imaging. These "fingerprints of trauma" can be searched or in every case and provide a valuable guide to potentially serious bowel and vascular injury.

CT plays an important role in the workup of stable patients after blunt trauma. Suspected bowel or mesenteric injuries (BBMI) often present with subtle and inconsistent imaging findings. Various radiographic signs have been used to predict the presence of these injuries. However, the optimal predictor for BBMI remains controversial. It is our contention that one of the best predictors is the overall impression of the reviewing radiologist. Thus, the purpose of this study was to identify radiographic predictors of therapeutic operative intervention in patients after blunt abdominal trauma.

Patients with a discharge diagnosis of a mesenteric injury after blunt trauma were identified over a 5-year period. Admission CT scans were reviewed for potential predictors of BBMI, including mesenteric hematoma, acute arterial extravasation, bowel wall hematoma, bowel devascularization, fecalization of small bowel, free air, fat pad injury. In addition, the overall impression of the scan by the reviewing radiologist was recorded. Patients were then stratified by therapeutic laparotomy and compared. Multivariable logistic regression (MLR) was then used to identify predictors of therapeutic laparotomy.

Over the study, 114 patients underwent operative intervention: 75 patients (66%) underwent therapeutic laparotomy. After adjusting for the above predictors including the overall impression of the radiologist, MLR identified the impression of the radiologist (OR 3.14; 95%CI 1.19-8.27, p=0.021), fat pad injury (OR 3.5; 95%CI 1.24-9.99, p=0.018) and bowel devascularization (OR 8.2; 95%CI 0.962-9.91, p=0.054) as independent predictors of therapeutic laparotomy. Interestingly, the overall impression of the radiologist had a positive predictive value of 82.1%.

CT remains vital in the evaluation of patients suspected of having bowel and mesenteric injuries after blunt trauma. An experienced radiologist remains invaluable in assessing often subtle signs of BBMI. A simplified scoring system utilizing these predictors could potentially aid the radiologist and surgeons in identifying those patients that would benefit from early operative intervention.

CT helps identify stable patients suspected of mesenteric/bowel injuries who would benefit from early operative intervention.

zohaib.ahmad@bmc.org

To evaluate the incidence in diagnosis and misses of surgically relevant abdominopelvic injuries on computed tomography (CT) imaging in the Damage Control (DC) patient.

This retrospective study was IRB approved and HIPAA compliant. Informed consent was waived. Patients aged 18 and older who sustained blunt or penetrating trauma requiring DC surgery without a prior CT at Boston Medical Center 2/21/2005 - 9/26/2018 were included. 59 patients met inclusion criteria (52 male, 4 female, mean age of 29). A CT was obtained 24 hours after the initial surgery. Each study was assessed by a single blinded fellowship trained radiologist. Outcomes were evaluated through failed surgical repair warranting surgical intervention, a clinically significant injury discovered on CT in a surgically explored area, a clinically significant injury discovered on CT in a surgically unexplored area, and a clinically significant injury missed on the initial CT but found on later surgery/imaging. These categorical variables were evaluated by percentages.

In a cohort of 57 patients, a total of 7 (12.5%) patients had a failed surgical repair discovered on initial CT (12.3%); of those 7 patients, 3 (42.8%) had failed repair of the liver. 6 (10.7%) patients had a clinically significant injury discovered on CT in a surgically explored area; of those 6 patients, 2 (33.3%) had injury of the kidney. 6 (10.7%) patients had a clinically significant injury discovered on CT in a surgically unexplored area. 9 (16.1%) patients who had a clinically significant injury that was missed on the initial CT; of those 9 patients, 3 (33.3%) had a missed injury to the large bowel.

As a staged surgical process in a critically traumatic injured patient, Damage Control (DC) surgery is a burgeoning life-saving method to address both traumatic and metabolic derangements in a timely manner. Further knowledge of common surgically and radiographically missed injuries is important to provide accurate diagnoses in these patients especially in the retroperitoneum and gastrointestinal system.

Accurate interpretation of computed tomography (CT) imaging during this process is vital to assessing for any surgically missed injury or assessment of repair in the critically ill DC patient.

fpaes@med.miami.edu

Selecting low risk penetrating trauma patients to forego laparotomy can be challenging. Bowel injury may prevent nonoperative management. Our goal is to compare the diagnostic performance of triple-contrast (oral, rectal, and IV) against IV contrast only CT in detecting bowel injury from penetrating abdominopelvic trauma, using surgical diagnosis during exploratory laparotomy as standard.

997 patients who underwent CT for penetrating trauma between 2009-2016 were enrolled in this IRB-approved retrospective cohort study. A total of 143 patients, including 15 females (ages 16-41), and 123 males (ages 14-83) underwent preoperative CT followed by exploratory laparotomy. Of these, 56 patients received triple-contrast CT. CT examinations were reviewed by 2 attending radiologists, blinded to surgical outcome and clinical presentation. Direct and indirect signs of bowel injury were documented. Results were stratified by contrast type and mechanism of injury and subsequently compared based upon diagnostic performance indicators of sensitivity, specificity, NPV, and PPV. AUCs were analyzed for determination of diagnostic accuracy.

Bowel injury was present in 45 out of 143 patients. Specificity and accuracy were higher with triple-contrast CT (98% specific [0.95, 1.00]), 97-99% accurate) compared to IV contrast only CT (66% specific [0.56, 0.75], 78-79% accurate). Sensitivity was highest with IV contrast only CT (91% sensitive [0.85, 0.98]) compared with triple-contrast CT (75% sensitive [0.56, 0.94]), although not statistically significant. Triple contrast CT increased diagnostic accuracy for both reviewers regardless of mechanism of injury. For reader 1, diagnostic accuracy with triple contrast CT versus IV contrast only CT was (99% [0.98, 1.00]) vs. 80% [0.62, 0.97]) for stab wounds and (100% vs. 76%[0.61, 0.91]) for gunshot wounds. For reader 2, diagnostic accuracy with triple-contrast CT versus IV contrast only CT was (99% [0.98, 1.00] vs. 74%, [0.55, 0.92]) for stab wounds and (95% [0.85, 1.00] vs. 79% [0.66, 0.92]) for gunshot wounds.

In our retrospective study, triple-contrast CT had greater accuracy, specificity, and NPV when compared to IV contrast only CT in evaluating for bowel injury from penetrating wounds.

Triple-contrast CT has greater accuracy, specificity, and NPV when compared to IV contrast only CT in evaluating for bowel injury from penetrating trauma.

jtlee3@uky.edu

Report trends for treating splenic injuries from 8 US trauma centers over 7 years Evaluate the frequency of reported splenic vascular injuries Evaluate factors influencing surgeon's decision to invasively treat (surgery or embolization) or conservatively manage

IRB approved, retrospective review of splenic injuries recorded from Level 1 trauma registries over 7 years from 8 institutions. Inclusion: Adults (>=18) with blunt splenic trauma, CT within 12 hours of admission Exclusion: penetrating trauma to the abdomen/pelvis, splenectomy prior to CT, left AMA, CT <16 detector, and death before splenic treatment. Descriptive statistics as well as regression analysis was performed, adjusting for multiple covariates.

918 subject were identified, 776 met inclusion criteria. 268 female. Original CT reports indictated active splenic hemorrhage (ASH) in 25%. 36% received invasive treatment (14% IR, 22% OR) and 64% were managed conservatively. A steady increase in IR management of splenic injuries and respective decrease in operative and conservative management over the study period. Multinomial logistic regression was performed for multiple outcomes including odds of recieving embolization or operative treatment and length of stay. Not surprisingly, AIS spleen, AIS Head/Neck and ISS scores showed significant increase in odds for invasive treatment. Presence of ASH on CT report was extremely predictive of invasive treatment when compared to conservative observation: Odds ratios for embolization: 22.063 and for operative 9.374 (while controlling for gender, age, synchronous major organ injury, vital signs, hemoglobin, INR, Platelets and if blood products received at admission). Regarding length of stay, on average, for every one unit increase in ISS, the length of stay increases by 1.031 days. Interestingly, on average, ASH demonstrated a 0.933 days longer than those without ASH; however this was not statistically significant (p=0.961).

We observed changing trends in treatment of splenic vascular injuries over the study period, as well as institutional differences in utilization of embolization versus operative management. Radiologic description of active splenic extravasation was highly predictive of embolization

Radiologist detection of active hemorrhage or contained vascular injury is highly predictive of invasive treatment of blunt splenic injury

jorge.soto@bmc.org

1) Review key direct and indirect CT findings of blunt pancreatic trauma. 2) Highlight potential pitfalls in diagnosis of pancreatic trauma. 3) Understand proper utilization of MR in patients with suspected pancreatic injuries.