FinalPresentation_Group_4_ComputerAided.pptx

Computer Aided Diagnosis

1. December 28 th , 2024 International University Department of Biomedical Engineering Computer Aided Diagnosis Approaches for Categorizing Symptoms Associated with Lumbar Disc Spondylosis Team Member: Ngo Thi Ngoc Phuong – BEBEIU21263 Do Thanh Nhan - BEBEIU22247 Supervisor: PhD. Nguyen Tan Nhu Subject: Computer Aided Diagnosis 1

2. December 28 th , 2024 International University Department of Biomedical Engineering 2 CURRENT STATUS PART 1

3. December 28 th , 2024 International University Department of Biomedical Engineering Current Status of Lumbar Disc Spine Degradation Lumbar spine degeneration, a major cause of low back pain, affected 619 million people in 2020 and is projected to reach 843 million by 2050. Prevalence varies globally, with higher rates in Europe (1) Ravindra, V. M., Senglaub , S. S., Rattani , A., Dewan, M. C., Härtl , R., Bisson, E., & Park, K. B. (2018). Degenerative lumbar spine disease: Estimating global incidence and worldwide volume. Global Spine Journal, 8(8), 784–794. https://doi.org/10.1177/2192568218770769 Figure 1. Incidence rates of degenerative spine disease/low back pain in World Bank and World Health Organization recognized countries [1].

4. December 28 th , 2024 International University Department of Biomedical Engineering 4 STATE OF THE ART PART 2

5. December 28 th , 2024 International University Department of Biomedical Engineering The Importance of detecting Lumbar Disc Spine Degradation in science and clinical practices *Clinical Needs Clinical Needs Reason for Priority Impact Accurate Diagnosis Clear understanding of the condition Avoiding misdiagnosis and unnecessary treatments. Ensures appropriate interventions and effective management. Early Detection Allows for in-time intervention to slow disease progression. Minimizes long-term complications and reduces the need for invasive treatments. Condition Synchronization Ensures strategies are tailored to the patient's specific condition and severity. Improves outcomes, symptom relief, and overall patient satisfaction. Andersson, G. B. J. (1999). "Epidemiological features of chronic low-back pain." The Lancet, 354(9178), 581-585.

6. December 28 th , 2024 International University Department of Biomedical Engineering The Importance of detecting Lumbar Disc Spine Degradation in science and clinical practices *Scientific needs - Early Intervention and Prevention Slowing Disease Progression Figure 2. The Stages Of Degenerative Disc Disease [1] [1] Yorkville Sports Medicine Clinic. (n.d.). Stages of degenerative disc disease . Retrieved December 26, 2024, from https://www.yorkvillesportsmed.com/blog/stages-of-degenerative-disc-disease

7. December 28 th , 2024 International University Department of Biomedical Engineering The Importance of detecting Lumbar Disc Spine Degradation in science and clinical practices *Scientific needs - Early Intervention and Prevention Minimize Complications Chronic Pain Nerve Compression Reduced Mobility Herniated Disc Localized Pain Radiating Pain Radiculopathy Cauda Equina Syndrome (rare but serious) Bulging or Ruptured Disc Spinal Stenosis Chou, R., Qaseem , A., Snow, V., Casey, D., Cross, J. T., Shekelle , P., & Owens, D. K. (2007). Diagnosis and treatment of low back pain: A joint clinical practice guideline from the American College of Physicians and the American Pain Society. Annals of Internal Medicine, 147(7), 478-491. https://doi.org/10.7326/0003-4819-147-7-200710020-00006

8. December 28 th , 2024 International University Department of Biomedical Engineering The Importance of detecting Lumbar Disc Spine Degradation in science and clinical practices *Scientific needs - Cost-Effectiveness Identify the early stage of lumbar degradation progression Determine the most impactful solution treating specifically for each stage of disease Avoiding utilizing full process of conventional diagnosing and treatment Reduce the squander of money Yorkville Sports Medicine Clinic. (n.d.). Stages of degenerative disc disease . Retrieved December 26, 2024, from https://www.yorkvillesportsmed.com/blog/stages-of-degenerative-disc-disease

9. December 28 th , 2024 International University Department of Biomedical Engineering 9 PREVIOUS STUDIES PART 3

10. December 28 th , 2024 International University Department of Biomedical Engineering SpineOne Framework (due year 2021) A deep learning model with CNNs and attention mechanisms uses keypoint heatmaps to identify landmarks on vertebrae and discs, focusing on critical areas to detect degeneration severity and classify abnormalities in MRI slices. He, J., Liu, W., Wang, Y., Ma, X., & Hua, X. S. (2021, December). SpineOne : A One-Stage Detection Framework for Degenerative Discs and Vertebrae. In 2021 IEEE International Conference on Bioinformatics and Biomedicine (BIBM) (pp. 1331-1334). IEEE.

11. December 28 th , 2024 International University Department of Biomedical Engineering SpineOne Framework (due year 2021) He, J., Liu, W., Wang, Y., Ma, X., & Hua, X. S. (2021, December). SpineOne : A One-Stage Detection Framework for Degenerative Discs and Vertebrae. In 2021 IEEE International Conference on Bioinformatics and Biomedicine (BIBM) (pp. 1331-1334). IEEE. Comprehended Goals Challenges Detects and classifies degenerative discs and vertebrae in MRI slices. Utilizes keypoint heatmaps and attention modules to enhance detection accuracy. Requires high computational resources for training. May struggle with variations in MRI quality across different scanners.

12. December 28 th , 2024 International University Department of Biomedical Engineering SpineNetV2 (due date May 3, 2022) Windsor, R., Jamaludin , A., Kadir, T., & Zisserman, A. (2022). SpineNetV2: Automated Detection, Labelling and Radiological Grading Of Clinical MR Scans.  ArXiv , abs/2205.01683 . The method employs a deep learning-based segmentation and classification pipeline to automatically detect vertebral bodies and intervertebral discs in MRI images. It assigns anatomical labels to detected structures (e.g., L1-L5 vertebrae) and uses trained classifiers to assess degenerative changes such as disc height loss, bulging, or herniation. By integrating multi-sequence MRI data, the model enhances workflow efficiency.

13. December 28 th , 2024 International University Department of Biomedical Engineering Windsor, R., Jamaludin , A., Kadir, T., & Zisserman, A. (2022). SpineNetV2: Automated Detection, Labelling and Radiological Grading Of Clinical MR Scans.  ArXiv , abs/2205.01683 . Comprehended Goals Challenges Automatically detects and labels vertebral bodies in MR scans. Performs radiological grading of intervertebral disc degeneration. Limited in detecting rare or atypical degenerative patterns Performance can vary with low-quality or noisy MRI scans. SpineNetV2 (due date May 3, 2022)

14. December 28 th , 2024 International University Department of Biomedical Engineering 14 OUR RESEARCHING PART 4

15. December 28 th , 2024 International University Department of Biomedical Engineering AI-Assisted MRI Diagnosis (Systematic Review) (due year 2024) Liawrungrueang , W., Park, J. B., Cholamjiak , W., Sarasombath , P., & Riew , K. D. (2024). Artificial Intelligence-Assisted MRI Diagnosis in Lumbar Degenerative Disc Disease: A Systematic Review.  Global Spine Journal , 21925682241274372. This approach uses a combination of machine learning and deep learning techniques, with models trained on labeled MRI datasets. It uses CNNs to analyze cross-sectional and longitudinal MRI images of a sample lumbar spine and outputs the 2D image composite into 3D, generating probabilistic predictions of the severity of degeneration or a diagnostic label (e.g., healthy vs. degenerative).

16. December 28 th , 2024 International University Department of Biomedical Engineering 16 PROJECT REQUIREMENT PART 5

17. December 28 th , 2024 International University Department of Biomedical Engineering Improve the ability to recognize lumbar spondylosis from the first stages of the disease (from spinal nerve stretching) PROJECT REQUIREMENT Expand degenerative diagnosis, recognize each stage of degeneration, assess severity based on universally programmed data processor Integrate code into a common interface, suitable for future applications, when manipulation becomes easier. Requirement 1: Sensitivity Requirement 2: Flexibility Requirement 3: Applicability

18. December 28 th , 2024 International University Department of Biomedical Engineering 18 METHOD PART 6

19. December 28 th , 2024 International University Department of Biomedical Engineering BLOCK DIAGRAM Data Process Technique 1 Expected result Technique n( th ) Technique 3 Technique 2 # Data lấy từ đâu # Process bao gồm bao nhiêu bước ( dựng model, testing, tạo giao diện ,…) …… # Nêu ra những technique ứng dụng trong bài (machine learning, detrasformation ,…) # Đưa ra kết quả dự kiến

20. December 28 th , 2024 International University Department of Biomedical Engineering STEP 1: (# Xác định step đã ghi trong phần process trong Block Diagram) # Mô tả step

21. December 28 th , 2024 International University Department of Biomedical Engineering STEP 2: (# Xác định step đã ghi trong phần process trong Block Diagram) # Mô tả step

22. December 28 th , 2024 International University Department of Biomedical Engineering STEP n( th ): (# Xác định step đã ghi trong phần process trong Block Diagram) # Mô tả step

23. December 28 th , 2024 International University Department of Biomedical Engineering 23 RESULT # Yêu cầu : Mỗi step chạy là được – capture ảnh vào PART 7

24. December 28 th , 2024 International University Department of Biomedical Engineering RESULT 1: (# Nêu đê mục của kết quả (VD: Output của ….., Ảnh …. được chụp từ …. ) # Capture ảnh dán vào

25. December 28 th , 2024 International University Department of Biomedical Engineering RESULT 2: (# Nêu đê mục của kết quả (VD: Output của ….., Ảnh …. được chụp từ …. ) # Capture ảnh dán vào

26. December 28 th , 2024 International University Department of Biomedical Engineering RESULT n( th ): (# Nêu đê mục của kết quả (VD: Output của ….., Ảnh …. được chụp từ …. ) # Capture ảnh dán vào

27. December 28 th , 2024 International University Department of Biomedical Engineering 27 DISCUSSION PART 8

28. December 28 th , 2024 International University Department of Biomedical Engineering DIFFICULTIES - RESTRICTION When using YOLO to detect and assess the degree of lumbar disc spinal degradation, researchers face several significant challenges. First, distinguishing and locating each vertebra on the spine is a difficult task, as the vertebrae have similar sizes and shapes. Additionally, while YOLO has high performance in many object detection tasks, its accuracy may not be sufficient when applied to medical images, as medical objects often have more subtle features. Furthermore, collecting and labeling medical image data , including both normal and pathological cases, is a major challenge. Finally, YOLO requires extensive computational complexity , especially when processing high-resolution medical images, which can lead to issues with speed and efficiency. applications.

29. December 28 th , 2024 International University Department of Biomedical Engineering DISFINISHED GOALS Basically, if we do not take into account the dependencies, this project is really perfect. However, there are still tasks that our team has not completed: First, reducing the cost of hiring a training server is almost impossible because the operation of the built code takes up a large amount of space, the cost can be up to 20 dollars per hour. Second, the development of the project is still in the planning stage, I wonder if in the future I can put this model into a complete interface , similar to the structure of Slicer.