Volume 10, Issue 8, August – 2025 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25aug1009

Automated Trauma Detection by Using

Machine Learning
Bilal Shabbir Qaisar1; Mahammad Ali Shahid2; Sunil Ashraf3;
Muhammad Adnan4; M. Mudasar Azeem5; Maham Ali6; Muhammad Nauman7
1,2,3,7
Faculty of Comsats University Islamabad, Sahiwal Campus 57000, Pakistan
6
Department of Psychology, Umer Hospital Okara 56300, Pakistan
4;5
Faculty of Computing University of Okara 56300, Pakistan

Publication Date: 2025/09/04

Abstract: Imaging techniques are widely used for medical diagnostics. This can sometimes lead to a real bottleneck when
there is a shortage of medical practitioners, and the images must be manually processed. In such a situation, there is a need
to reduce the amount of manual work by automating part of the analysis. In this study, we investigate the potential of a
machine-learning algorithm for trauma detection in medical image processing. A new method called ResNet50V2 was
developed on the trauma dataset to detect trauma disease. We compare the results of the new method analysis with other
state-of-the-art networks. The proposed base model, ResNet50V2, received a score of 99.40%.

Keywords: Machine Learning; ResNet50V2; Trauma; Medical Images.

How to Cite: Bilal Shabbir Qaisar; Mahammad Ali Shahid; Sunil Ashraf; Muhammad Adnan; M. Mudasar Azeem; Maham Ali;
Muhammad Nauman (2025) Automated Trauma Detection by Using Machine Learning. International Journal of Innovative Science
and Research Technology, 10(8), 2129-2135. https://doi.org/10.38124/ijisrt/25aug1009

I. INTRODUCTION Machine learning is a subfield of AI and computer

science that teaches computers and other automated systems
Post-Traumatic Stress Disorder (PTSD) and acute stress how to learn from data. It falls under the umbrella of AI and
disorder (ASD) are two of the most crippling mental health computer science. Because it is able to handle complicated data
disorders that can develop after being exposed to horrific sets with various distributions and use advanced mathematical
events like war, natural catastrophes, accidents, or acts of mass approaches, machine learning is useful for building complex
murder [1]. It has long been a problem in both clinical practice data models [3]. Both supervised and unsupervised methods are
and academic research to make a diagnosis of ASD or PTSD. commonly used in the "learning" process. In supervised
There are many different ways in which trauma-related learning, the computer learns to convert inputs into desired
disorders might manifest, due to the wide variety of risk factors outputs via methods like regression (where the result is a
and complicated causes. Diagnostic criteria for classification numerical value) or classification (where the result is a
systems such as the DSM-5 have also been refined through category, like "disease" or "no disease"). Data input and desired
research, including chronic populations and those in tertiary results are given to the system by the user. A few examples of
care settings. Nevertheless, the phenomenology of the illness in popular supervised learning algorithms are logistic regression,
its early phases can be quite variable and generic, and these neural networks, and support vector machines. One of the most
criteria might not be able to capture it completely. Cognitive well-known medical applications of supervised learning is the
processing therapy and extended exposure therapy have Framingham Risk Score, which is used to diagnose coronary
garnered the greatest interest among the evidence-based heart disease. Another common use of this method is risk
trauma-focused treatments [2]. It might be tough to create first- calculation and prediction. While supervised learning relies on
line psychotherapies owing to issues like patient burden and predetermined associations and output variables, unsupervised
varied patient characteristics. Since clinical trial subjects do not learning seeks to understand the data structure at its core [4]. It
necessarily reflect the multimorbidity profiles of "real-life" can be done via clustering, which finds groups of related cases,
patients, it is possible that some statistically significant results or by density estimation, which finds the distribution of the
claimed by evidence-based medicine may not apply to available data. Network analysis, which makes use of
individual patients. This is especially important to consider in regression and clustering techniques, visualizes the connection
PTSD, as research does not always fully address the substantial between individual symptoms and groups of related symptoms,
clinical variability that might be present in this disorder. thus providing information on the amount and severity of
correlations between symptoms. An up-to-date explanation of

IJISRT25AUG1009 www.ijisrt.com 2129

Volume 10, Issue 8, August – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25aug1009

the relevant machine learning principles and their limitations is  Developed a deep learning method to classify the trauma
available elsewhere. disease.
 Enhanced the performance of the existing models in terms
Improving disease classification, optimizing of accuracy.
individualized treatment selection, and predicting treatment  Analyzed the issue of an imbalanced dataset in the context
results and risk factors are all possible with the use of machine of classification.
learning methods. Because of the biological and clinical  Developed a method to overcome the issues of high false-
diversity associated with PTSD and ASD, which can impede negative rates.
our capacity to understand their causes and create effective
treatments and diagnostic tools, machine learning presents a II. LITERATURE REVIEW
hopeful avenue for further investigation into these disorders
[5]. Prior studies have concentrated on automating the
assessment of trauma severity and organ segmentation. By
The most popular diagnostic technique for blunt utilising a deep learning-based segmentation technique that
abdominal trauma (BAT) is computed tomography, which has was strengthened by decision tree analysis, Drezin et al. were
a big impact on treatment strategies. Deep learning models, or able to predict severe artery damage in liver trauma with an
DLMs, have demonstrated significant potential to improve a accuracy of 0.84. A similar methodology for quantitative
number of clinical practice elements [6]. evaluation and detection of liver trauma was created by
Farzaneh et al., utilising 77 CT scans. Using active contour
Artificial intelligence technology has the potential to modelling, Chen et al. developed a four-part approach for the
address this issue by expediting the diagnosis of these types of automated grading of spleen damage and automated kidney
injuries and improving patient care and treatment in segmentation in trauma patients. Using a Little dataset and an
emergencies. As a result, the medical field is becoming more external attention and synthetic phase augmentation module,
and more interested in using AI and machine learning (ML) to Zhou et al. enhanced multiphase splenic vascular damage
support physicians. Utilizing AI models as virtual diagnostic segmentation using a DeepLab-v3 baseline. Combining shape
assistants to function as secondary image readers can and statistical data with texture feature extraction has improved
significantly enhance the accuracy and reliability of the diagnosis of traumatic brain injury (TBI).
radiological image interpretation. This gives radiologists more
authority and self-assurance in their diagnostic evaluations. By using entropy to extract nonlinear features,
Using AI's ability to quickly identify images can speed up the Raghavendra et al. were able to identify cerebral haematomas
diagnosis procedure and increase clinical effectiveness. in CT scans with an accuracy of 97.37 percent. He created a
model for classifying aberrant CT slices utilising statistical,
Early and precise injury detection is essential for both GLCM, and wavelet data. To classify haematoma subtypes,
successful treatment and patient survival in trauma care. Sharma and Venugopalan used characteristics depending on
Conventional trauma assessment techniques frequently entail shape, texture, and intensity. Tong et al. achieved an 84.86%
medical professionals' subjective appraisal and manual recall rate using their midline creation technique for identifying
inspection, which can cause delays and irregularities. The haematomas, which involves extracting and comparing LBP
emergence of machine learning holds promise for transforming texture features and histogram data of both hemispheres. He
trauma detection through automation and the use of data-driven came to the conclusion that a Bayesian classifier and a distance
insights. transform using five landmarks could differentiate between
normal and subarachnoid haematomas (SAH), and he
Because they rely on human judgement, the existing suggested a DWT-based paradigm for this purpose in patients
trauma detection technologies have limitations that might cause with traumatic brain injuries.
delays and variability in critical care settings. To deliver precise
and rapid trauma assessments, an automated system that can Our goal in this research was to determine how well a
evaluate different types of medical data is required. prototype system based on deep learning could automatically
detect rib fractures in trauma CT scans. Algorithm performance
To create a machine learning-based automated trauma was on par with that of radiologists, with a sensitivity of 87.4%
detection system that can effectively identify and categorize and specificity of 91.5% at the per-examination level. As a
trauma cases by processing and analyzing medical data, result of breathing artefacts, normal intercostal arteries, and
including imaging scans, patient symptoms, and past medical undamaged ribs, the test generated 0.16 false positives per
records. By increasing trauma diagnosis speed and accuracy, examination. The F1 score came out at 0.85. There were 587
this method seeks to improve patient outcomes. confirmed fractures out of 894 reports (sensitivity: 65.7%).
Fracture displacement and acuteness were the main factors that
By developing and testing a machine learning-based correlated with the algorithm's accuracy.
automated trauma detection system, this study aims to increase
the reliability, efficiency, and timeliness of trauma diagnoses. Since multiple rib fractures are more common in
Specifically, this study aims to: emergencies than single rib fractures, the algorithm worked
better at the per-examination level. Just 9.4 percent of the scans
The measurable Contributions of the study with expected that contained rib fractures in our dataset contained a single
results are as under; fracture. The algorithm's 94.1% NPV proves it is useful as a
supplementary reading tool for the pre-exam level.

IJISRT25AUG1009 www.ijisrt.com 2130

Volume 10, Issue 8, August – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25aug1009

Rib fracture detection algorithms using CT images have Segmentation is typically approached as a pixel/voxel
only been the subject of a single pilot study. Although the classification task or utilises classical picture delineation
number of false positives and false negatives per case was not techniques in machine learning algorithms. After that, post-
stated, Yan et al. utilised a CNN and reported a sensitivity of processing methods including active contours, smoothing,
95.0% and a significantly lower positive predictive value morphological operations, and thresholding are applied to each
(PPV) of 55.7% for rib fracture identification. pixel or voxel to extract relevant characteristics. While
Farzaneh et al. identified each super-pixel as normal or
Our findings are consistent with previous research on the subdural haematoma (SDH) using geometric, textural, and
effectiveness of algorithms in detecting bone fractures using statistical criteria, Scherer et al. classified haematomas voxel-
CT scans. Researchers found 81.3% sensitivity and 2.7 false wise using textural and statistical data. In their work on ICH
positives per case when studying the effectiveness of support identification, Muschelli et al. used intensity-feature-based
vector machines for vertebral body fractures. Our study voxel selection.
improved the usability of clinical process by identifying fewer
false positives per case (0.16). Our results are comparable to Various hybrid segmentation algorithms have been put
those of Bar et al., who also examined a segmentation step and forth, such as active contouring, FCM clustering, region
a patch-based CNN for spinal compression fractures; however, expansion, and thresholding. A wavelet-based thresholding and
they did not provide information regarding false positives. white matter FCM clustering (WMFCM) model was evaluated
Their study also reported a sensitivity of 83.9% and a by Gautam and Raman, while a nonlinear 3D segmentation
specificity of 93.8%. method utilising region growth was reported by Saenz et al.
The use of FCM clustering and active contour modelling for
Also, we found 137 rib fractures that the algorithm had haematoma delineation was successfully demonstrated by
marked several times. Applying this approach at the per- Bhadauria et al. with an accuracy of 99.10%.
examination level has no major consequences; however, if it is
used to precisely count the number of fractures and the results By testing MDRLSE with region growth and adaptive
are then confirmed by a radiologist, workflow efficiency may thresholding, Prakash et al. demonstrated that it effectively
be reduced. False estimates of the frequency of rib fractures segmented haematomas in 3D CT images. Using an
may come from blind acceptance of the detailed results without autoencoder trained to identify haematoma slices, Nag and
verification. colleagues demonstrated a cost-effective haematoma
segmentation approach with a sensitivity of 0.71.
Although 97 acute fractures were missed in the textual
CT reports, the algorithm was able to identify them. Because a Critical findings in CT scans can now be detected and
higher incidence of rib fractures is linked to a higher death rate, classified using deep learning algorithms, according to current
this data highlights the significance of accurate rib fracture research. A deep learning-based approach for key finding
identification. Consistent with the results of Ringl et al., we also identification, such as infarcts and haematomas, was
discovered that the detection rate for anterior fractures was introduced by Prevedello et al. A 3D convolutional neural
lower than other places. This could be because it is difficult to network (CNN) model for CT result triage and classification
diagnose the area where cartilage meets ribs. was created by Titano et al. using ResNet-50. To merge slice-
level classification with context, Grewal et al. developed a
There have been numerous investigations into feature RAD net, which achieved an accuracy of 81.82%. Using a U-
extraction approaches and segmentation methods for Net based architecture for localisation and classification,
haematoma detection. In order to increase classification Chilamkurthy et al. utilised deep learning algorithms to detect
performance, many approaches for haematoma segmentation in and confirm abnormalities in non-contrast head CT scans.
CT scans use bespoke and handcrafted features. As an example,
hematoma regions were segmented using modified DRLSE. III. METHODOLOGY
The regions were further divided into four classes using shape
and texture criteria that were manually created. By analyzing A. Materials and Methods
hematomas’ shapes, Al-Ayoob et al. were able to create a This study introduces a machine-learning approach for
model that was 92% accurate in classifying them into three trauma disease classification. Classification of trauma diseases
categories. Xiao et al. suggested a way to differentiate between affecting the mind using a machine-learning approach is
subdural and epidural hematomas using main and secondary presented in this paper. A first-level augmentation and pre-
characteristics associated with the most extensive area of hyper processing stage is applied to generate distinct records and
density. address the class imbalance. After mechanical features were
extracted to differentiate between trauma diseases from the
dataset, a pre-trained "Machine Learning Method" model was
applied at the second level. Figure 1 shows the steps involved
in the suggested method's process flow.

IJISRT25AUG1009 www.ijisrt.com 2131

Volume 10, Issue 8, August – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25aug1009

The images were rotated by a fixed amount (25 degrees

in this case) using the rotation transformation. With the width
shift parameter set to 0.1, the images were shifted randomly to
the right or left using the width shift range transformation.
Using the height shift range option with a value of 0.1, the
training images were vertically flipped. A shear angle of 0.2
was used in this case; this is a method for image transformation
wherein one axis is kept constant while the other is extended.
A zoom range of 0.2 was utilized to enlarge the image, in
accordance with the rules of random zoom transformation: a
value greater than 1.0 indicates that the images were enlarged,
and a value less than 1.0 indicates that the photos were zoomed
out. Using the flip function, we were able to modify the image's
horizontal orientation. We used a scale where zero is totally
black and one is really light to change the brightness.

We trained ResNet50V2 from the ground up using the

dataset from the last chapter. The RSNA dataset was divided
into three parts: training, testing, and validation. After training
on the training set, the ResNet50V2 model was tested and
evaluated using the validation and test datasets. As a result, we
split the dataset between training, testing, and validation, with
60% going to each.

Scaling, rotating, brightening, height shifting, zooming,

shear-ranging, horizontally flipping, and channel-shifting with
fill mode closest were all used to extend the training set in order
to increase the dataset size and diversity. Eliminating
Fig 1 Flow Chart of the Presented Method. overfitting would guarantee that the model can be applied to
new situations. Model optimization was employed in the study
B. RSNA Abdominal Trauma Detection Dataset process. Training images with 60% image ratios were utilized
The effectiveness of machine learning methods depends in the current study using the RSNA dataset. Twenty percent of
heavily on the availability of a good dataset. The most the remaining forty percent of the actual photos go into testing
extensive collection of high-quality trauma disease images and validation. With the use of ResNet50V2, the model was
made accessible for study can be found in the RSNA taught to classify and forecast the labels of each training image.
Abdominal Trauma Detection Diseases dataset archive [28].
There are 186K images in the dataset. D. Architecture of ResNet50V2
The ResNet model [29] is well-known and widely used in
C. Image Pre-Processing computer vision competitions due to its consistently excellent
Preprocessing is applied to all input photos of the RSNA performance. Many more exist; for example, Inception
dataset to achieve increased consistency in classification results ResNetV2 [30], Mobile Net [31], and Google Net [32]. These
and enhanced feature extraction. A massive image dataset was models are educated using information from numerous photos
needed for the CNN approach's extensive training repetitions across various datasets. To effectively address a wide variety
to avoid the risk of over-fitting. of computer vision problems, transfer learning methods can
take advantage of these pre-trained model weights (dataset and
All photos in the original RSNA dataset can be found in computing resources). The ResNet50 model was pre-trained on
a resolution of 6000 × 4000 pixels. To hasten the process with a small dataset of plant species images for transfer learning.
Python code, the dataset is downsized to 224 x 224. Subsequent paragraphs will delve into the specifics of the
ResNet50V2 model's architecture and the many pre-trained
We have utilized the Image Data Generator function of weights it can draw from. The ResNet50V2 model is a fifty-
the Keras library in Python to enhance the data in multiple layer convolutional neural network (CNN). Figure-2, depicts
ways, preventing overfitting and expanding the training the architecture of the ResNet50V2 model, including the
dataset. Keeping the same range of pixel values allowed us to ResNet50V2 fine-tuning setup.
cut the computing cost by utilizing a scale transformation.
Consequently, a 0–1 range for pixel values was established
using the parameter value (1. /255).

IJISRT25AUG1009 www.ijisrt.com 2132

Volume 10, Issue 8, August – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25aug1009

 Recall
Another critical parameter is called recall, and it refers to
the percentage of input samples that are of a type that the model
can accurately predict. The formula for the recall is as follows:
Recall = T P / (T P+FN)

 F1 Score
The f1 score is a statistic utilised to contrast recall and
precision.
F1 Score = 2 * (Precision * Recall) (Precision + Recall)

IV. RESULTS AND DISCUSSION

A Google Research team evaluated the ResNet50V2

model. This ResNet50V2 method experiment was built using
Tensor-Flow, the free and open-source Keras, and Python.
There was a default learning rate and a binary cross-entropy
loss function used during training by the Adam optimizer.
 Automated Trauma Detection by Using Machine Learning
was observed using the RSNA dataset to assess its efficacy.
 Evaluation of the presented ResNet50V2 model’s
performance on the RSNA dataset using data augmentation
techniques on the training set.
 The results were compared to those of other state-of-the-
art networks.
 To evaluate the results of Trauma disease classification
Fig 2 Architecture of the ResNet50V2 Model. prior studies using machine learning.
E. Evaluation Measures A. Performance of Proposed Model on RSNA Dataset
The testing dataset was used to evaluate the suggested We evaluated and analyzed the performance of the
approach following the training phase. We tested the ResNet50V2 base model on the RSNA dataset. Validation
architecture's efficacy using recall, accuracy, F1 score, and accuracy for the model increased from 99.87% at the end of the
precision. We will examine the performance metrics utilized in first epoch to 99.67% after the most recent epoch. Training
this research in the parts that follow. What follows is a accuracy improves from 96.85% after the first epoch to 99.45%
mathematical definition and representation of the terms "true after the last epoch in Figure 4.1. As seen in Figure-3,
positive," "true negative," "false negative," and "false positive." ResNet50V2’s validation loss drastically decreased from 75%
to 1.34%. Furthermore, similar to the initial loss, the training
 Classification Accuracy loss was 9.78% after the first period and 1.76% after finishing
The accuracy of a classification system can be evaluated training.
by determining what percentage of its predictions were correct
and what percentage were incorrect. On a previously unseen test set, the ResNet50V2 base
model was evaluated. While the model's overall accuracy was
Accuracy = T P+T N / (T P+T N +FP+FN) 99.40% across the board in the test set, ResNet50V2 excelled
in the trauma class, achieving 99% precision, 100% recall, and
 Precision 99% F1-score. The typical class is first-rate, with a flawless
When analysing the effectiveness of a model, 99% recall, 100% accuracy, and 99% f1 score.
classification accuracy may not always be the most appropriate
metric to employ. For instance, this is one of the scenarios Using a confusion matrix, we could visually examine
where there is a considerable gap in socioeconomic status. It’s how well various models classified data. The rows of the
a safe bet to assume that each sample is of the highest possible confusion matrix that are not on the diagonal represent the
quality. If the model isn’t picking up any new information, it inaccurate predictions. Classification accuracy in the related
would be irrational to infer that all components belong to the ResNet50V2 base model was shown by darker colours,
best class. Therefore, when we talk about accuracy, we refer to whereas misclassified data was shown by lighter colours.
the fluctuation in findings you receive while measuring the Figure-4 shows the confusion matrices from the test set that
same object several times with the same tools. The term will be used to evaluate the overall effectiveness of
"precision" refers to one of these statistics and can be defined ResNet50V2. The confusion matrix shows that the
as follows: ResNet50V2 baseline model's predictions were correct for all
picture categories. Confusion analysis using the default
Precision = T P / (T P+FP) ResNet50V2 model settings reveals a data identification
success rate of 99.40% and a false positive rate of 0.60%. The
ResNet50V2 base model does a fantastic job when comparing
the trauma and normal samples' confusion matrices.

IJISRT25AUG1009 www.ijisrt.com 2133

Volume 10, Issue 8, August – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25aug1009

V. CONCLUSION

The research described in this manuscript explored using

Convolutional Neural Networks to recognize trauma in real-
time using machine learning techniques. For trauma detection,
this method is both reliable and quick. The test findings
demonstrate a remarkable accuracy rate in identifying people
who are either trauma disease or normal. The trained model
completed the task using the ResNet50V2 model, with
individual accuracy results of 99.40%.

The best way to train a CNN model to identify and

recognize trauma diseases in humans is to integrate multiple
models and evaluate their performance accuracy. In addition,
the authors recommend an improved optimizer, more precise
parameter values, enhanced tuning, and models for adaptive
transfer learning.

Fig 4 The ResNet50V2 Base Model Confusion Matrix on ACKNOWLEDGMENT

Test Set.
Special thank goes to Comsats University Islamabad,
Sahiwal Campus, for providing the IT services, Lab, and server
provisions for smoothly writing the manuscript.

Fig 3 The ResNet50V2 Base Model: (a) Accuracy (b) Loss Graph.

REFERENCES [4]. Ghosh, P., et al., Efficient prediction of cardiovascular

disease using machine learning algorithms with relief
[1]. Goldenberg, M.N., D. Benedek, and R.J. Ursano, and LASSO feature selection techniques. IEEE
Disaster victims and the response to trauma, in Access, 2021. 9: p. 19304-19326.
Textbook of Community Psychiatry: American [5]. Ramos-Lima, L.F., et al., The use of machine learning
Association for Community Psychiatry. 2022, techniques in trauma-related disorders: a systematic
Springer. p. 443-455. review. Journal of psychiatric research, 2020. 121: p.
[2]. Shifrin, A., et al., Posttraumatic stress disorder 159-172.
treatment preference: Prolonged exposure therapy, [6]. Bharde, P., R.V. Sai, and S. Sripriya, Role of
cognitive processing therapy, or medication therapy? computed tomography imaging in the diagnosis of
Psychological Services, 2022. blunt and penetrating abdominal trauma injuries.
[3]. Tyagi, A.K. and P. Chahal, Artificial intelligence and International Journal, 2023. 10(1): p. 10.
machine learning algorithms, in Challenges and
applications for implementing machine learning in
computer vision. 2020, IGI Global. p. 188-219.

IJISRT25AUG1009 www.ijisrt.com 2134

Volume 10, Issue 8, August – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25aug1009

[7]. Sharma, A.K., et al., Medical image classification [21]. Gudigar, A., et al., Automated detection and screening
techniques and analysis using deep learning networks: of traumatic brain injury (TBI) using computed
a review. Health informatics: a computational tomography images: a comprehensive review and
perspective in healthcare, 2021: p. 233-258. future perspectives. International journal of
[8]. Chen, Y.-H. and M. Sawan, Trends and challenges of environmental research and public health, 2021.
wearable multimodal technologies for stroke risk 18(12): p. 6499.
prediction. Sensors, 2021. 21(2): p. 460. [22]. Pitt, J., Y. Pitt, and J. Lockwich, Clinical and cellular
[9]. Le Glaz, A., et al., Machine learning and natural aspects of traumatic brain injury, in Handbook of
language processing in mental health: systematic Toxicology of Chemical Warfare Agents. 2020,
review. Journal of medical Internet research, 2021. Elsevier. p. 745-765.
23(5): p. e15708. [23]. Chen, Q., et al., Traumatic axonal injury:
[10]. Yeo, M., et al., Review of deep learning algorithms for neuropathological features, postmortem diagnostic
the automatic detection of intracranial hemorrhages on methods, and strategies. Forensic Science, Medicine
computed tomography head imaging. Journal of and Pathology, 2022. 18(4): p. 530-544.
neurointerventional surgery, 2021. 13(4): p. 369-378. [24]. Morell-Hofert, D., et al., Validation of the revised
[11]. Hussain, S., et al., Modern diagnostic imaging 2018 AAST-OIS classification and the CT severity
technique applications and risk factors in the medical index for prediction of operative management and
field: a review. BioMed research international, 2022. survival in patients with blunt spleen and liver injuries.
2022(1): p. 5164970. European radiology, 2020. 30: p. 6570-6581.
[12]. Brunner, M., et al., Social media and people with [25]. Hamid, S., et al., Dual-energy CT: a paradigm shift in
traumatic brain injury: a metasynthesis of research acute traumatic abdomen. Canadian Association of
informing a framework for rehabilitation clinical Radiologists Journal, 2020. 71(3): p. 371-387.
practice, policy, and training. American journal of [26]. Cheng, C.-T., et al., Deep Learning for automated
speech-language pathology, 2021. 30(1): p. 19-33. detection and localization of traumatic abdominal
[13]. Buhrmester, V., D. Münch, and M. Arens, Analysis of solid organ injuries on CT scans. Journal of Imaging
explainers of black box deep neural networks for Informatics in Medicine, 2024: p. 1-11.
computer vision: A survey. Machine Learning and [27]. Tiwari, A., M. Poduval, and V. Bagaria, Evaluation of
Knowledge Extraction, 2021. 3(4): p. 966-989. artificial intelligence models for osteoarthritis of the
[14]. Ponnusamy, V., et al., AI‐Driven Information and knee using deep learning algorithms for orthopedic
Communication Technologies, Services, and radiographs. World Journal of Orthopedics, 2022.
Applications for Next‐Generation Healthcare System. 13(6): p. 603.
Smart Systems for Industrial Applications, 2022: p. 1- [28]. https://www.kaggle.com/datasets/theoviel/rsna-
32. abdominal-trauma-detection-png-pt1
[15]. Gipson, J., et al., Diagnostic accuracy of a [29]. Balakrishnan, J., & David, D. (2019). Melanoma
commercially available deep-learning algorithm in classification and birthmark mole detection on clinical
supine chest radiographs following trauma. The images. Paper presented at the 2019 International
British Journal of Radiology, 2022. 95(1134): p. Conference on Vision Towards Emerging Trends in
20210979. Communication and Networking (ViTECoN).
[16]. Choi, J., et al., The impact of trauma systems on [30]. Xie, F., Fan, H., Li, Y., Jiang, Z., Meng, R., & Bovik,
patient outcomes. Current problems in surgery, 2021. A. J. I. t. o. m. i. (2016). Melanoma classification on
58(1): p. 100849. dermoscopy images using a neural network ensemble
[17]. Cheng, C.-T., et al., A scalable physician-level deep model. 36(3), 849-858.
learning algorithm detects universal trauma on pelvic [31]. Alquran, H., Qasmieh, I. A., Alqudah, A. M.,
radiographs. Nature communications, 2021. 12(1): p. Alhammouri, S., Alawneh, E., Abughazaleh, A., &
1066. Hasayen, F. (2017). The melanoma skin cancer
[18]. Sundkvist, J., et al., Epidemiology, classification, detection and classification using support vector
treatment, and mortality of adult femoral neck and machine. Paper presented at the 2017 IEEE Jordan
basicervical fractures: an observational study of Conference on Applied Electrical Engineering and
40,049 fractures from the Swedish Fracture Register. Computing Technologies (AEECT).
Journal of orthopaedic surgery and research, 2021. 16: [32]. Filali, Y., Khoukhi, H. E., Sabri, M. A., Aarab, A. J.
p. 1-10. M. T., & Applications. (2020). Efficient fusion of
[19]. Puttagunta, M. and S. Ravi, Medical image analysis handcrafted and pre-trained CNNs features to classify
based on deep learning approach. Multimedia tools melanoma skin cancer. 79(41), 31219-31238.
and applications, 2021. 80(16): p. 24365-24398.
[20]. Tulbure, A.-A., A.-A. Tulbure, and E.-H. Dulf, A
review on modern defect detection models using
DCNNs–Deep convolutional neural networks. Journal
of Advanced Research, 2022. 35: p. 33-48.

IJISRT25AUG1009 www.ijisrt.com 2135