Deep Learning Algorithm for Traumatic Splenic Injury Detection and Sequential Localization

The Three-dimensional Weakly Supervised Deep Learning Algorithm for Traumatic Splenic Injury Detection and Sequential Localization

Spleen laceration is a lethal abdominal trauma and usually be diagnosed by medical images such as computed tomography. Deep learning had been proved its usage in detect abnormalities in medical images.

In this trial, we used de-identified registry databank to develop a novel deep-learning based algorithm to detect the spleen trauma and to identify the injury locations.

Study Overview

• Status: Completed
• Conditions: Machine Learning; Spleen Injury
• Intervention / Treatment: Diagnostic test: Deep learning algorithm

Detailed Description

Background

Splenic injury is the most common solid visceral injury in blunt abdominal trauma, and high-resolution abdominal computed tomography (CT) can adequately detect the injury. However, these lethal injuries sometime have been overlooked in current practice. Deep learning algorithms have proven their capabilities in detecting abnormal findings in medical images. The aim of this study is to develop a three-dimensional, unsupervised deep learning algorithm for detecting splenic injury on abdominal CT using a sequential localization and classification approach.

Material and Methods

We retrospectively collected data from patients who underwent contrast-enhanced abdominal CT in the emergency department of Chang Gung Memorial Hospital, Linko, due to trauma and acute abdomen from Jul 2008 to Dec 2017. All patients were registered in the trauma and acute abdomen registries. Demographic information, including age, sex, disease diagnosis, trauma mechanism, Injury Severity Score, New Injury Severity Score , Abbreviated Injury Scale, and spleen injury grade, was collected. Scans showing splenic injury were identified as positive, and the remaining scans were defined as negative controls. We identified 300 venous phase scans with splenic injury and randomly selected 300 additional venous phase scans from the negative controls. CT scans with abdominal trauma injuries other than splenic injury were not excluded to reduce the selection bias. All data were split by age, sex, the presence of splenic injury, and injury severity score using stratified sampling into the developmental dataset and the test set at a ratio of 8:2. One-eighth of the developmental dataset was further reserved as the validation set during model construction.

Image preprocessing and labeling

The CT scan images were acquired in the original Digital Imaging and Communications in Medicine (DICOM) format. The images were then converted to the Neuroimaging Informatics Technology Initiative format, producing 3D voxel-based images. Our algorithm was then developed based on the venous axial slices, the most common imaging direction in abdominal trauma surveys. During the training process, image augmentation by translation, rotation, scaling, and elastic distortions was applied to increase model generalizability.

A trauma surgeon with 10 years of experience confirmed all the positive and negative scans. In all scans, the spleen with its surrounding background was covered with a manually drawn 3D bounding box.

Spleen localization

The localization model was designed based on 3D Faster RCNN with Resnet-101as the backbone structure and trained on the development dataset. We used cross-entropy, focal loss as the class loss, and L1 loss, distance intersection over union (DIOU) as box regression loss, and adopted intersection over union (IOU) and DIOU in non-maximum suppression (NMS) for training of the object detection algorithm.

Spleen injury identification and visualization

The cropped 3D images were used to develop the splenic injury classification model. We modified the block architecture to improve the interpretability of the reasoning process of the learned network. The output of the model was the probability of splenic injury.

Model performance was evaluated using the area under the receiver operating characteristic curve (AUROC), accuracy, sensitivity, specificity, positive predictive value ,and negative predictive value.

• Study Type: Observational
• Enrollment (Actual): 600

Contacts and Locations

Study Locations

• Taiwan
  • Taoyuan, Taiwan, 333
    • Chang Gung Memorial Hospital

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: 18 years and older (Adult, Older Adult)
• Accepts Healthy Volunteers: No
• Genders Eligible for Study: All

• Sampling Method: Probability Sample

Study Population

We enrolled images of the patients who underwent abdominal computed tomography in emergency department for trauma and acute abdominal survey from Jul 2008 to Dec 2017. Then we selected 300 images with splenic injury and 300 images without.

Description

Inclusion Criteria:

• patients who underwent abdominal computed tomography in emergency department for trauma and acute abdominal survey from Jul 2008 to Dec 2017.

Exclusion Criteria:

• poor quality images
• no contrast series of computed tomography images.
• images from other hospitals without proper evaluation

Study Plan

How is the study designed?

Design Details

• Observational Models: Other
• Time Perspectives: Retrospective

Number of groups / cohorts

2

Cohorts and Interventions

Group / Cohort	Intervention / Treatment
splenic injury group; We retrospectively collected data from patients who underwent contrast-enhanced abdominal CT in the emergency department of Chang Gung Memorial Hospital, Linko, due to trauma and acute abdomen from Jul 2008 to Dec 2017. We identified 300 venous phase scans with splenic injury.	Diagnostic test: Deep learning algorithm; A sequential two-stage 3D spleen injury detection framework to identify splenic injury in the CT scans
control group; We retrospectively collected data from patients who underwent contrast-enhanced abdominal CT in the emergency department of Chang Gung Memorial Hospital, Linko, due to trauma and acute abdomen from Jul 2008 to Dec 2017. We randomly selected 300 additional venous phase scans without splenic injury	Diagnostic test: Deep learning algorithm; A sequential two-stage 3D spleen injury detection framework to identify splenic injury in the CT scans

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Diagnostic accuracy	Diagnostic accuracy of the deep learning algorithm to detect splenic injury	3 days

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Comparison of difference backbone of algorithm	The difference of diagnostic accuracy between different infrastructure of deep learning algorithm	3 days

Collaborators and Investigators

• Sponsor: Chang Gung Memorial Hospital
• Investigators: Principal Investigator: Chien-Hung Liao, MD., Chang Gung Memorial Hospital

Study record dates

Study Major Dates

• Study Start (Actual): February 1, 2022
• Primary Completion (Actual): November 1, 2022
• Study Completion (Actual): November 1, 2022

Study Registration Dates

• First Submitted: November 30, 2022
• First Submitted That Met QC Criteria: December 7, 2022
• First Posted (Estimate): December 9, 2022

Study Record Updates

• Last Update Posted (Estimate): December 9, 2022
• Last Update Submitted That Met QC Criteria: December 7, 2022
• Last Verified: November 1, 2022

More Information

Terms related to this study

• Keywords: artificial intelligence; deep learning; spleen trauma; spleen injury
• Additional Relevant MeSH Terms: Wounds and Injuries
• Other Study ID Numbers: SpleenTrNet

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: No
• IPD Plan Description: Because we extract the de-identified data from registry databank, we could not offer individual participant data.

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No