1
|
Cheng CT, Lin HH, Hsu CP, Chen HW, Huang JF, Hsieh CH, Fu CY, Chung IF, Liao CH. Deep Learning for Automated Detection and Localization of Traumatic Abdominal Solid Organ Injuries on CT Scans. JOURNAL OF IMAGING INFORMATICS IN MEDICINE 2024; 37:1113-1123. [PMID: 38366294 PMCID: PMC11169164 DOI: 10.1007/s10278-024-01038-5] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Subscribe] [Scholar Register] [Received: 11/25/2023] [Revised: 01/31/2024] [Accepted: 02/01/2024] [Indexed: 02/18/2024]
Abstract
Computed tomography (CT) is the most commonly used diagnostic modality for blunt abdominal trauma (BAT), significantly influencing management approaches. Deep learning models (DLMs) have shown great promise in enhancing various aspects of clinical practice. There is limited literature available on the use of DLMs specifically for trauma image evaluation. In this study, we developed a DLM aimed at detecting solid organ injuries to assist medical professionals in rapidly identifying life-threatening injuries. The study enrolled patients from a single trauma center who received abdominal CT scans between 2008 and 2017. Patients with spleen, liver, or kidney injury were categorized as the solid organ injury group, while others were considered negative cases. Only images acquired from the trauma center were enrolled. A subset of images acquired in the last year was designated as the test set, and the remaining images were utilized to train and validate the detection models. The performance of each model was assessed using metrics such as the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, specificity, positive predictive value, and negative predictive value based on the best Youden index operating point. The study developed the models using 1302 (87%) scans for training and tested them on 194 (13%) scans. The spleen injury model demonstrated an accuracy of 0.938 and a specificity of 0.952. The accuracy and specificity of the liver injury model were reported as 0.820 and 0.847, respectively. The kidney injury model showed an accuracy of 0.959 and a specificity of 0.989. We developed a DLM that can automate the detection of solid organ injuries by abdominal CT scans with acceptable diagnostic accuracy. It cannot replace the role of clinicians, but we can expect it to be a potential tool to accelerate the process of therapeutic decisions for trauma care.
Collapse
Affiliation(s)
- Chi-Tung Cheng
- Department of Trauma and Emergency Surgery, Chang Gung Memorial Hospital, Linkou, Chang Gung University, Taoyuan, Taiwan
| | - Hou-Hsien Lin
- Department of Trauma and Emergency Surgery, Chang Gung Memorial Hospital, Linkou, Chang Gung University, Taoyuan, Taiwan
| | - Chih-Po Hsu
- Department of Trauma and Emergency Surgery, Chang Gung Memorial Hospital, Linkou, Chang Gung University, Taoyuan, Taiwan
| | - Huan-Wu Chen
- Department of Medical Imaging & Intervention, Chang Gung Memorial Hospital, Linkou, Chang Gung University, Taoyuan, Taiwan
| | - Jen-Fu Huang
- Department of Trauma and Emergency Surgery, Chang Gung Memorial Hospital, Linkou, Chang Gung University, Taoyuan, Taiwan
| | - Chi-Hsun Hsieh
- Department of Trauma and Emergency Surgery, Chang Gung Memorial Hospital, Linkou, Chang Gung University, Taoyuan, Taiwan
| | - Chih-Yuan Fu
- Department of Trauma and Emergency Surgery, Chang Gung Memorial Hospital, Linkou, Chang Gung University, Taoyuan, Taiwan
| | - I-Fang Chung
- Institute of Biomedical Informatics, National Yang Ming Chiao Tung University, Taipei, Taiwan
| | - Chien-Hung Liao
- Department of Trauma and Emergency Surgery, Chang Gung Memorial Hospital, Linkou, Chang Gung University, Taoyuan, Taiwan.
| |
Collapse
|
2
|
Emergency whole-body CT scans in pediatric patients with trauma: patterns of injuries, yield of dual-phase scanning, and influence of second read on detection of injuries. Eur Radiol 2022; 32:8473-8484. [PMID: 35687137 DOI: 10.1007/s00330-022-08878-1] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/03/2021] [Revised: 04/12/2022] [Accepted: 05/12/2022] [Indexed: 11/04/2022]
Abstract
OBJECTIVES To describe injury patterns in children with multiple trauma (MT), evaluate the yield of dual-phase whole-body CT (WBCT), and quantify missed injuries detected on second reading. METHODS Remotely analyzed WBCT performed between 2011 and 2020 in 63 emergency departments on children admitted for MT were included. Second reading occurred within 24 h. Collected data included age, sex, mechanism, Injury Severity Score (ISS), radiologists' experience, time and duration of first reading, conclusion of both readings, and dosimetry. Melvin score assessed the clinical impact of missed injuries. RESULTS Overall, 1114 patients were included, 1982 injuries were described in 662 patients (59.4%), 452/1114 (40.6%) WBCT were negative, and 314 (28.2%) patients had MT (≥ 2 body parts injured). The most frequent injuries were pulmonary contusions (8.3%), costal fractures (6.2%), and Magerl A1 vertebral fractures (4.9%). Overall, 151 injuries were missed in 92 (8.3%) patients. Independent predictors for missed injuries were age ≤ 4 years (p = 0.03), number of injured body parts ≥ 2 (p = 0.01), and number of injuries ≥ 3 (p < 0.001). Melvin score grade 3 lesions were found in 16/92 (17.4%) patients with missed injuries (1.4% of all WBCT), where only prolonged follow-up was necessary. Thirteen active bleeding or pseudoaneurysms were detected (0.7% of injuries). CONCLUSION Injuries were diagnosed in 59.4% of patients. Double-reading depicted additional injuries in 8.3% of patients, significantly more in children ≤ 4 years, with ≥ 3 injuries or ≥ 2 injured body parts. As 28 % of patients had MT and 1.1% had active extravasation or pseudoaneurysm, indication for WBCT should be carefully weighted. KEY POINTS • When performed as a first-line imaging evaluation, approximately 41% of WBCT for MT children were considered normal. • The three most common injuries were pulmonary contusions, costal fractures, and Magerl A1 vertebral fractures, but the patterns of traumatic injuries on WBCT depended on the children's age and the trauma mechanism. • The independent predictors of missed injuries were age ≤ 4 years, number of body parts involved ≥ 2, and total number of injuries ≥ 3.
Collapse
|
3
|
Ozcan A, Ahn T, Akay B, Menoch M. Imaging for Pediatric Blunt Abdominal Trauma With Different Prediction Rules: Is the Outcome the Same? Pediatr Emerg Care 2022; 38:e654-e658. [PMID: 33616315 DOI: 10.1097/pec.0000000000002346] [Citation(s) in RCA: 3] [Impact Index Per Article: 1.5] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Indexed: 11/26/2022]
Abstract
BACKGROUND Computerized tomography (CT) of the abdomen and pelvis is the standard imaging modality to diagnose intra-abdominal injury (IAI). Clinicians must weigh the risk-benefit of CT compared with the degree of clinical suspicion for an IAI. Pediatric Emergency Care Applied Research Network (PECARN), Streck, and blunt abdominal trauma in children (BATiC) prediction rules have been published to help guide evaluation of these patients. Pediatric Emergency Care Applied Research Network uses history and physical examination findings, whereas Streck and BATiC use examination plus laboratory and imaging findings. At the time of the study, there was not a protocol that was more routinely sited. Our goal was to compare these different prediction rules. METHODS This was a retrospective electronic chart review of all children younger than 18 years presenting for either level 1 or 2 trauma activations at our pediatric emergency department (ED) between June 1, 2015, to June 30, 2017. Charts were manually reviewed for a mechanism concerning for abdominal trauma, and demographic data, history and physical examination findings, laboratory and imaging results per prediction rules, and revisits in 7 days were collected.The prediction rules were applied to all charts that had all data necessary. For study purposes, a score of zero for PECARN and Streck, and score of ≤5 for modified BATiC (mBATiC) were defined as "low risk." Patients with no CT, negative CT, and no new injury found on revisit were classified as "no IAI identified," and patients with positive CT or revisit with injury found as "IAI identified." The results were compared via Fisher exact test. RESULTS A total of 249 patients met the inclusion criteria with a median age of 12 years. Of the low-risk patients, 119 (98.7%) of 121 in PECARN group, 21 (100%) of 21 in Streck, and 48 (85.7%) of 56 in mBATiC group had no IAI identified. None of the low-risk patients required any intra-abdominal intervention. No missed IAI was identified during revisit review. Negative predictive values of all 3 rules were significant for PECARN, Streck, and mBATiC (98.35%, 100%, and 85.71%, respectively). Overall, 27 patients had positive CT results for IAI. CONCLUSIONS The PECARN and Streck rules have high negative predictive values to predict low-risk patients who do not require CT. When laboratory studies are not obtained, PECARN is an effective means of excluding IAI for low-risk patients. When laboratory tests were obtained, the Streck rule performed well. Overall, the results are similar to the past individual studies done on each individual rule. History and physical examination findings are of high importance in pediatric trauma. This study supports limited imaging when no abnormal findings are present in children with blunt torso trauma. This is the only study found in the literature that has compared 3 different prediction rules.
Collapse
Affiliation(s)
- Ali Ozcan
- From the Pediatric Emergency Department, Loma Linda University Medical Center, Loma Linda, CA
| | - Terrie Ahn
- Pediatric Allergy and Immunology Department, UCLA Mattel Children's Hospital, Los Angeles, CA
| | | | - Margaret Menoch
- Pediatric Emergency Department,William Beaumont Hospital, Royal Oak, MI
| |
Collapse
|
4
|
Leeper CM, Nasr I, Koff A, McKenna C, Gaines BA. Implementation of clinical effectiveness guidelines for solid organ injury after trauma: 10-year experience at a level 1 pediatric trauma center. J Pediatr Surg 2018. [PMID: 28625692 DOI: 10.1016/j.jpedsurg.2017.05.025] [Citation(s) in RCA: 10] [Impact Index Per Article: 1.7] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Indexed: 11/19/2022]
Abstract
BACKGROUND Diagnostic imaging of pediatric blunt abdominal trauma is evolving in light of increased attention to radiation exposure. We hypothesize that the implementation of imaging guidelines has reduced total CT scans without missing clinically significant injury. METHODS We retrospectively reviewed blunt trauma patients age 0-17 with solid organ injury who underwent CT scan at our academic level 1 pediatric trauma center between 2005 and 2014. Variables including total annual trauma admissions and CT scans, demographics, injury characteristics, and procedures were recorded. Descriptive statistics, Fisher exact and rank sum testing were performed. p<0.05 defined significance. RESULTS Overall percentage of abdominal CT scans decreased significantly after protocol implementation. There were 498 solid organ injuries in 403 subjects. There was a significant decrease in the median percentage of low grade injuries (1.3% versus 0.6%; p=0.019) but no difference in high grade injuries (1.3% versus 1.1%; p=0.394). No patient had death, readmission or delayed diagnosis of injury requiring intervention. CONCLUSION Implementation of imaging guidelines for blunt abdominal trauma decreased the incidence of low grade solid organ injuries at our institution, but did not inhibit diagnosis and safe management of high grade injuries. Selective imaging of trauma patients decreases childhood radiation exposure and does not result in delayed bleeding or death. LEVEL OF EVIDENCE Level III, retrospective study.
Collapse
Affiliation(s)
- Christine M Leeper
- Division of General Surgery and Trauma, Department of Surgery, University of Pittsburgh Medical Center. 200 Lothrop Street, Pittsburgh, PA 15213, USA; Children's Hospital of Pittsburgh of UPMC. 7th Floor, Faculty Pavilion, One Children's Hospital Drive, 4401 Penn Avenue, Pittsburgh, PA 15224, USA.
| | - Isam Nasr
- The Johns Hopkins Department of Surgery, 1800 Orleans Street Pediatric Surgery Bloomberg 7323, Baltimore, MD 2128.
| | - Abigail Koff
- Division of General Surgery and Trauma, Department of Surgery, University of Pittsburgh Medical Center. 200 Lothrop Street, Pittsburgh, PA 15213, USA.
| | - Christine McKenna
- Children's Hospital of Pittsburgh of UPMC. 7th Floor, Faculty Pavilion, One Children's Hospital Drive, 4401 Penn Avenue, Pittsburgh, PA 15224, USA.
| | - Barbara A Gaines
- Children's Hospital of Pittsburgh of UPMC. 7th Floor, Faculty Pavilion, One Children's Hospital Drive, 4401 Penn Avenue, Pittsburgh, PA 15224, USA.
| |
Collapse
|
5
|
Savatmongkorngul S, Wongwaisayawan S, Kaewlai R. Focused assessment with sonography for trauma: current perspectives. Open Access Emerg Med 2017; 9:57-62. [PMID: 28794661 PMCID: PMC5536884 DOI: 10.2147/oaem.s120145] [Citation(s) in RCA: 36] [Impact Index Per Article: 5.1] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/26/2022] Open
Abstract
Focused assessment with sonography for trauma (FAST) is a part of resuscitation of trauma patients recommended by international panel consensus. The purpose of FAST is to identify free fluid, which necessarily means blood in acute trauma patients. In this article, the authors focused on various aspects of FAST in the emergency department, prehospital care, pediatric setting, training and general pearls/pitfalls. Detailed techniques and interpretation of FAST are beyond the scope of this article.
Collapse
Affiliation(s)
| | - Sirote Wongwaisayawan
- Division of Emergency Radiology, Department of Diagnostic and Therapeutic Radiology, Ramathibodi Hospital, Mahidol University, Bangkok, Thailand
| | - Rathachai Kaewlai
- Division of Emergency Radiology, Department of Diagnostic and Therapeutic Radiology, Ramathibodi Hospital, Mahidol University, Bangkok, Thailand
| |
Collapse
|
6
|
Berona K, Kang T, Rose E. Pelvic Free Fluid in Asymptomatic Pediatric Blunt Abdominal Trauma Patients: A Case Series and Review of the Literature. J Emerg Med 2016; 50:753-8. [DOI: 10.1016/j.jemermed.2016.01.003] [Citation(s) in RCA: 4] [Impact Index Per Article: 0.5] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/18/2015] [Accepted: 01/07/2016] [Indexed: 10/22/2022]
|
7
|
Zundel S, Szavay P. Konservative vs. chirurgische Therapie des Bauchtraumas. Monatsschr Kinderheilkd 2015. [DOI: 10.1007/s00112-015-3472-y] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/29/2022]
|
8
|
Izquierdo Hernández B, Gutiérrez Alonso C, González Esgueda A, Alconchel Lagranja A. [Intestinal injury due to blunt abdominal trauma. Surgical indication based on diagnostic imaging]. An Pediatr (Barc) 2014; 81:e60-1. [PMID: 24874525 DOI: 10.1016/j.anpedi.2014.04.022] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 03/26/2014] [Accepted: 04/22/2014] [Indexed: 11/16/2022] Open
Affiliation(s)
- B Izquierdo Hernández
- Servicio de Radiodiagnóstico, Sección de Radiología Pediátrica, Hospital Universitario Miguel Servet, Zaragoza, España.
| | - C Gutiérrez Alonso
- Servicio de Radiodiagnóstico, Sección de Radiología Pediátrica, Hospital Universitario Miguel Servet, Zaragoza, España
| | - A González Esgueda
- Servicio de Cirugía Pediátrica, Hospital Universitario Miguel Servet, Zaragoza, España
| | - A Alconchel Lagranja
- Servicio de Radiodiagnóstico, Sección de Radiología Pediátrica, Hospital Universitario Miguel Servet, Zaragoza, España
| |
Collapse
|