0% found this document useful (0 votes)
18 views

PreDefenseSlide - DeepSarkerUpda Ted

This document outlines Deep Sarker's pre-defense presentation for a project on bone fracture detection using X-ray images and machine learning. The presentation covers the objectives to build an ML algorithm for automated fracture detection, curate a diverse medical image dataset, and evaluate the system's performance compared to manual methods. It also summarizes the research methodology, including data collection and preprocessing techniques used to train models like ResNet50, DenseNet121, and VGG16 for fracture classification. Finally, the document discusses interpreting the models' results and the potential impacts on healthcare efficiency and society.

Uploaded by

sademunn
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
18 views

PreDefenseSlide - DeepSarkerUpda Ted

This document outlines Deep Sarker's pre-defense presentation for a project on bone fracture detection using X-ray images and machine learning. The presentation covers the objectives to build an ML algorithm for automated fracture detection, curate a diverse medical image dataset, and evaluate the system's performance compared to manual methods. It also summarizes the research methodology, including data collection and preprocessing techniques used to train models like ResNet50, DenseNet121, and VGG16 for fracture classification. Finally, the document discusses interpreting the models' results and the potential impacts on healthcare efficiency and society.

Uploaded by

sademunn
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 24

Pre-defense

Bone Fracture Detection with XAI


for Advance Medical Technology

Presented by Supervised by

Deep Sarker Ms. Sharun Akter Khushbu

201-15-13863 Lecturer (Senior Scale)


Department of Computer
Science & Engineering

Saturday 30 July 2014


Contents
• Introduction
• Motivation
• Objectives
• Research Questions
• Related Work
• Research Gap
• Data Collection
• Data Processing
• Dataset Split
• Research Methodology
• Result and Analysis
• Interpretations
• Impact on Society and Environment
• Limitations and Future Work
• References

Pre-defense
2
Introduction
• The field of medical imaging has undergone remarkable advancements in recent
years, revolutionizing the diagnosis and treatment of various health conditions
• This research integrates computer science and engineering for advanced bone
fracture detection, addressing challenges in manual methods with a robust
automated system.
• The focus is on diverse dataset collection and preprocessing to train effective
machine learning models.
• The proposed system enhances fracture diagnosis efficiency, optimizing healthcare
resources for improved patient car

Pre-defense
3
Objectives
• Build ML algorithm for automated bone
fracture detection.

• Curate diverse dataset for robust training

• Preprocess data to ensure reliable model


training

• Evaluate system against manual methods


for performance

• Explore healthcare workflow integration


for efficiency

Pre-defense
4
Motivation
• Efficiency Amidst Growing Cases: Addressing rising fractures and imaging demand,
our goal is to boost diagnostic efficiency.
• Automating Time-Intensive Diagnoses: We aim to automate fracture detection,
crucial for speeding up manual radiological interpretations.
• Precision in Complex Fractures: Leveraging tech to precisely identify and
categorize fractures, especially in complex patterns.
• Swift Impact on Patient Care: Timely detection directly improves patient outcomes
by reducing complications and enhancing overall care.
• Innovative Blend of Tradition and Tech: Our motivation lies in merging traditional
diagnostics with cutting-edge technology for streamlined healthcare efficiency

Pre-defense
5
Research Questions
• How does machine learning detect and classify bone fractures in medical images?

• Which preprocessing techniques optimize medical imaging data for a Bone


Fracture Detection system?

• How does integrating the system into clinical workflows impact orthopedic
diagnostics' efficiency?

• How does the automated system improve bone fracture detection compared to
manual methods, enhancing patient outcomes and healthcare efficiency?

Pre-defense
6
Related Work
• Haar wavelet-based compression excels in medical imaging for higher ratios and
improved PSNR values
• SIFT algorithm's resilience enhances X-ray image matching and classification
• Deep learning aids timely bone fracture detection in traumatology
• CrackNet exhibits robust fracture detection with high accuracy
• Hybrid SFNet with Canny edge outperforms in efficient bone fracture diagnosis

Pre-defense
7
Research Gap
• Dataset Limitation: Annotated datasets scarcity hampers accurate bone fracture
detection model training.

• Ethical and Transparency Challenges: Responsible patient data use and algorithmic
transparency pose complex but critical issues.

• Integration Hurdles: Challenges involve user acceptance, health records


integration, and imaging system interoperability.

• Dynamic Healthcare Environment: Adapting to changes in imaging, demographics,


and practices needs ongoing monitoring and refinement

Pre-defense
8
Data Collection

Pre-defense
9
Sample Data

Fractured Not Fractured

Pre-defense
10
Data Preprocessing

• Resize
• Contour

Pre-defense
11
Data Set Split

Pre-defense
12
Research Methodology
Tools Used

Pre-defense
13
Research Methodology

Pre-defense
14
Research Methodology
Used Models

Bone Fracture Detect

ResNet50

DenseNet121

VGG!6

RNN

LSTM

Pre-defense
15
Research Methodology
Research process
• Model Selection
• Data Collection
• Preprocessing Steps
• Training Process
• Evaluation Metrics
• Interpretation Using SHAP
• Interpretation Using Saliency

Pre-defense
16
Final Results
Algorithm Name Accuracy Precision Recall f1-score

ResNet 50 89% 91 89 88

DenseNet 121 97% 99 99 99

VGG 16 73% 73 72 72

LSTM 63% 66 65 65

RNN 60% 64 62 59

InceptionV3 93% 93 93 93

Pre-defense
17
Interpretations
• SHAP: Explains ML model output by quantifying feature contributions.
• Saliency: Measures the impact of each input feature on a model's output for
interpretability.

Pre-defense
18
Final Output

Pre-defense
19
Impact on Society and
Environment
• Improved Diagnoses: Swift and accurate automated diagnoses enhance patient
care.
• Efficient Healthcare: Automation streamlines processes, saving time and
resources.
• Cost-Effective: Quicker diagnoses lead to reduced hospital stays and lower
healthcare costs.
• Green Tech Priority: Focus on energy-efficient algorithms and renewables
minimizes environmental impact.
• Eco-Friendly Practices: Initiatives for sustainable materials and responsible
disposal reduce electronic waste.

Pre-defense
20
Limitations and Future Works
• Data Limitation
• Model Complexity
• Explainability Scope
• Scope for Model Enhancement
• Real-Time Processing Exploration
• Advanced Explainability Techniques
• Integration with Industrial IoT
• Ethical and Socioeconomic Impacts
• Cross-Industry Generalization

Pre-defense
21
References
• [1] Sahin, M. E. (2023). Image processing and machine learning‐based bone fracture detection and classification using X‐ray images. International Journal of Imaging
Systems and Technology, 33(3), 853-865.
• [2] Basha, C. M. A. K., Padmaja, M., & Balaji, G. N. (2018). Computer aided fracture detection system. Journal of Medical Imaging and Health Informatics, 8(3), 526-
531.
• [3] Basha, C. Z., Reddy, M. R. K., Nikhil, K. H. S., Venkatesh, P. S. M., & Asish, A. V. (2020, March). Enhanced computer aided bone fracture detection employing X-ray
images by Harris Corner technique. In 2020 Fourth International Conference on Computing Methodologies and Communication (ICCMC) (pp. 991-995). IEEE.
• [4] Yang, A. Y., Cheng, L., Shimaponda-Nawa, M., & Zhu, H. Y. (2019, December). Long-bone fracture detection using artificial neural networks based on line features of
X-ray images. In 2019 IEEE symposium series on computational intelligence (SSCI) (pp. 2595-2602). IEEE.
• [5] Umadevi, N., & Geethalakshmi, S. N. (2012, July). Multiple classification system for fracture detection in human bone x-ray images. In 2012 Third International
Conference on Computing, Communication and Networking Technologies (ICCCNT'12) (pp. 1-8). IEEE.
• [6] Eksi, Z., & Cakiroglu, M. (2012). Performance evaluation of the popular segmentation algorithms for bone fracture detection. AWERProcedia Information
Technology & Computer Science, 1, 1245-1249.
• [7] Dai, Z., Li, Z., & Han, L. (2021). Bonebert: A bert-based automated information extraction system of radiology reports for bone fracture detection and diagnosis. In
Advances in Intelligent Data Analysis XIX: 19th International Symposium on Intelligent Data Analysis, IDA 2021, Porto, Portugal, April 26–28, 2021, Proceedings 19 (pp.
263-274). Springer International Publishing.
• [8] Santos, K. C., Fernandes, C. A., & Costa, J. R. (2022). Feasibility of Bone Fracture Detection Using Microwave Imaging. IEEE Open Journal of Antennas and
Propagation, 3, 836-847.
• [9] Sinthura, S. S., Prathyusha, Y., Harini, K., Pranusha, Y., & Poojitha, B. (2019, May). Bone Fracture Detection System Using CNN Algorithm. In 2019 International
Conference on Intelligent Computing and Control Systems (ICCS) (pp. 545-549). IEEE.
• [10] Cao, Y., Wang, H., Moradi, M., Prasanna, P., & Syeda-Mahmood, T. F. (2015, April). Fracture detection in x-ray images through stacked random forests feature
fusion. In 2015 IEEE 12th international symposium on biomedical imaging (ISBI) (pp. 801-805). IEEE.
• [11] AlGhaithi, A., & Al Maskari, S. (2021). Artificial intelligence application in bone fracture detection. Journal of Musculoskeletal Surgery and Research, 5, 4.
• [12] Dimililer, K. (2017). IBFDS: Intelligent bone fracture detection system. Procedia computer science, 120, 260-267.
• [13] Meena, T., & Roy, S. (2022). Bone fracture detection using deep supervised learning from radiological images: A paradigm shift. Diagnostics, 12(10), 2420.
• [14] Yadav, D. P., & Rathor, S. (2020, February). Bone fracture detection and classification using deep learning approach. In 2020 International Conference on Power
Electronics & IoT Applications in Renewable Energy and its Control (PARC) (pp. 282-285). IEEE.

Pre-defense
22
Links
• UI
• Google Site
• Github URL
• Video Recording

Pre-defense
23
Thank You

Pre-defense
24

You might also like