presentation based on research paper informstion

this is presentation

1. Security Enhancing Pneumonia Detection in Chest X-rays: A Deep Learning and Hybrid Model Approach for Improved Medical Image Analysis 15-04-2024 KIRSET ’ 24 1 Name of Author/s:Sukhpreet Singh, Jaspreet Kaur Assistant Professor Guru Kashi University Talwandi Sabo, Bathinda ABSTRACT ID:1077

2. Abstract Medical image analysis is crucial for accurate diagnosis and treatment planning. This paper introduces an innovative approach that combines deep learning methodologies with hybrid models to revolutionize the analysis of medical X-ray images. By integrating the strengths of deep learning and hybrid models, our research aims to overcome the limitations of conventional methods and enhance the precision, efficiency, and interpretability of medical image analysis tasks. Specifically, we focus on pneumonia detection in chest X-rays, a challenging yet critical task in medical imaging. Through extensive experimentation and evaluation, our proposed approach demonstrates exceptional performance, achieving a remarkable accuracy rate of 90% in distinguishing normal cases from pneumonia-afflicted ones. Additionally, we elucidate the intricate mechanisms underlying the collaborative prowess of deep learning and hybrid models, providing profound insights into their integrated effectiveness in medical image analysis. This research opens new avenues for advancing medical imaging technologies and improving healthcare outcomes. 15-04-2024 KIRSET ’ 24 2

3. Literature Review 15-04-2024 KIRSET ’ 24 3 S.No Authors Title of the paper Journal name Vol No, Page No & Year of publication Key findings 1 Cherian, T.; Mulholland, E.K.; Carlin, J.B.; Ostensen, H.; Amin, R.; Campo, M.D.; Greenberg, D.; Lagos, R.; Lucero, M.; Madhi, S.A.; et al. Standardized interpretation of pediatric chest radiographs for the diagnosis of pneumonia in epidemiological studies National Library of Medicine 2005 May;83(5):353-9.   Epub 2005 Jun 24. Out of 208 interpretable images, 43% were categorized as pneumonia by reference reading, with reader sensitivity and specificity > 0.70 in 14 out of 20 cases. Intra-observer variability showed kappa index > 0.6 in 19 out of 20 readers for 92 images. 2 Tahir, A.M.; Chowdhury, M.E.; Khandakar, A.; Al-Hamouz, S.; Abdalla, M.; Awadallah, S.; Reaz, M.B.I.; Al-Emadi, N. A Systematic Approach to the Design and Characterization of a Smart Insole for Detecting Vertical Ground Reaction Force ( vGRF ) in Gait Analysis National Library of Medicine 2020 Feb; 20(4): 957. The study introduces low-cost FSR and piezoelectric sensor calibration setups, effective for 1D force but limited for flexible piezoelectric sensors; it also outlines smart insole design, highlighting FSR-based vGRF acquisition efficacy and suggesting alternative bio-sensing uses for calibrated piezoelectric sensors. 3 Chowdhury, M.E.; Alzoubi, K.; Khandakar, A.; Khallifa, R.; Abouhasera, R.; Koubaa, S.; Ahmed, R.; Hasan, A Wearable Real-Time Heart Attack Detection and Warning System to Reduce Road Accidents National Library of Medicine 2019 Jun 20;19(12):2780.  doi: 10.3390/s19122780. (SVM) algorithm with polynomial kernel with extended time-frequency features using extended modified B-distribution (EMBD) showed highest accuracy and was able to detect 97.4% and 96.3% of ST-elevation myocardial infarction (STEMI) and non-ST-elevation MI (NSTEMI), respectively. 4 Chowdhury, M.E.H.; Khandakar, A.; Alzoubi, K.; Mansoor, S.; Tahir, A.M.; Reaz, M.B.I.; Al-Emadi, N Real-Time Smart-Digital Stethoscope System for Heart Diseases Monitoring (Multidisciplinary Digital Publishing Institute) s 2019, 19, 2781; doi:10.3390/s19122781 The study introduces a portable heart sound capturing system with optimized ensemble algorithm achieving high accuracy (94.63%) for anomaly detection, offering potential for noninvasive at-home cardiac monitoring and future integration with smartphones for real-time classification and display of results. 5 Liu, N.; Wan, L.; Zhang, Y.; Zhou, T.; Huo, H.; Fang, T. Liu, N.; Wan, L.; Zhang, Y.; Zhou, T.; Huo , H.; Fang, T. Exploiting convolutional neural networks with deeply local description for remote sensing image classification. IEEE Access 2018, 6, 11215–11228. IEEE VOLUME 6, 2018 The study explores local descriptions using CNN models ( CaffeNet and VGG-VD16), applying Hellinger kernel, PCA transformation, and two aggregate strategies for global representation formation, with experiments demonstrating effectiveness on RSI datasets.

4. Motivation / Objective The primary objective of this study is to develop a deep learning and hybrid model approach for pneumonia detection in chest X-rays. Specific objectives include: Collecting a diverse dataset of chest X-ray images annotated for pneumonia. Designing a deep learning model architecture optimized for medical image analysis. Integrating hybrid models to combine the strengths of different algorithms. Training and evaluating the proposed approach on the dataset to assess its performance in terms of accuracy, sensitivity, and specificity. Investigating the interpretability of model decisions to gain insights into the underlying mechanisms of pneumonia detection. 15-04-2024 KIRSET ’ 24 4

5. Proposed Methodology Data Collection and Preprocessing We collected a dataset of chest X-ray images from publicly available repositories, including the NIH Chest X-ray dataset and the RSNA Pneumonia Detection Challenge dataset. The images were annotated by expert radiologists for the presence or absence of pneumonia. We performed preprocessing steps such as resizing, normalization, and augmentation to standardize the images and increase the robustness of the model. Deep Learning Model Architecture For the deep learning component of our approach, we designed a convolution neural network (CNN) architecture tailored for medical image analysis. The CNN consists of multiple layers of convolution, pooling, and fully connected layers, followed by softmax activation for classification. We fine-tuned the model parameters using gradient descent optimization and evaluated its performance on the validation set. Training and Evaluation Strategies We split the dataset into training, validation, and testing sets to train and evaluate the models. We employed cross-validation techniques to mitigate over fitting and ensure robustness of the models. Performance metrics such as accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC-ROC) were used to quantify the models' performance. 15-04-2024 KIRSET ’ 24 5

6. Block Diagram / Flow Chart 15-04-2024 KIRSET ’ 24 6

7. Results Obtained Performance Metrics We evaluated the performance of the proposed approach using standard metrics such as accuracy, sensitivity, specificity, and AUC-ROC. These metrics provide insights into the models' ability to correctly classify normal and pneumonia cases and assess their overall diagnostic performance. 15-04-2024 KIRSET ’ 24 7

8. Results Obtained 15-04-2024 KIRSET ’ 24 8

9. Results Obtained 15-04-2024 KIRSET ’ 24 9

10. Results Obtained Enhanced Security : The implemented system achieved enhanced security for cloud storage through cryptographic techniques such as encryption and blockchain double hashing. This ensured data confidentiality and integrity. Efficient Auditing : The checksum algorithm facilitated efficient dynamic auditing, allowing for the detection of unauthorized changes to stored data. This provided users with assurance regarding the integrity of their data in the cloud. Improved Accessibility : The token-based key generation technique enhanced user control and accessibility to stored data. Users could securely access and retrieve their data from the cloud server with the generated access tokens. Robustness : The system demonstrated robustness against data loss or corruption by employing redundancy through blockchain splitting. This distributed storage approach increased data availability and resilience. 15-04-2024 KIRSET ’ 24 10

11. Conclusion / Future Scope In conclusion, we have introduced a novel approach that combines deep learning methodologies with hybrid models for pneumonia detection in chest X-rays. Our research demonstrates exceptional performance in terms of accuracy, efficiency, and interpretability, surpassing existing methods and opening new avenues for advancing medical imaging technologies. Recommendations for Future Research Future research directions include addressing the aforementioned limitations, refining the model architecture, incorporating domain-specific knowledge, and conducting prospective clinical studies to validate the approach in real-world settings. 15-04-2024 KIRSET ’ 24 11

12. References World Health Organization. Standardization of Interpretation of Chest Radiographs for the Diagnosis of Pneumonia in Children; Technical Report; World Health Organization: Geneva, Switzerland, 2001. Cherian , T.; Mulholland, E.K.; Carlin, J.B.; Ostensen , H.; Amin , R.; Campo, M.D.; Greenberg, D.; Lagos, R.; Lucero, M.; Madhi , S.A.; et al. Standardized interpretation of paediatric chest radiographs for the diagnosis of pneumonia in epidemiological studies. Bull. World Health Organ. 2005, 83, 353–359. Tahir , A.M.; Chowdhury , M.E.; Khandakar , A.; Al- Hamouz , S.; Abdalla , M.; Awadallah , S.; Reaz , M.B.I.; Al- Emadi , N. A systematic approach to the design and characterization of a smart insole for detecting vertical ground reaction force ( vGRF ) in gait analysis. Sensors 2020, 20, 957. Chowdhury , M.E.; Alzoubi , K.; Khandakar , A.; Khallifa , R.; Abouhasera , R.; Koubaa , S.; Ahmed, R.; Hasan , A. Wearable real-time heart attack detection and warning system to reduce road accidents. Sensors 2019, 19, 2780. Chowdhury , M.E.H.; Khandakar , A.; Alzoubi , K.; Mansoor , S.; Tahir , A.M.; Reaz , M.B.I.; Al- Emadi , N. Real-time smart-digital stethoscope system for heart diseases monitoring. Sensors 2019, 19, 2781. Kallianos , K.; Mongan , J.; Antani , S.; Henry, T.; Taylor, A.; Abuya , J.; Kohli , M. How far have we come? Artificial intelligence for chest radiograph interpretation.  Clin . Radiol . 2019, 74, 338–345. Liu, N.; Wan, L.; Zhang, Y.; Zhou, T.; Huo , H.; Fang, T. Exploiting convolutional neural networks with deeply local description for remote sensing image classification. IEEE Access 2018, 6, 11215–11228. Sun, C.; Yang, Y.; Wen , C.; Xie , K.; Wen , F. Voiceprint identification for limited dataset using the deep migration hybrid model based on transfer learning. Sensors 2018, 18, 2399. Chen, Z.; Zhang, Y.; Ouyang , C.; Zhang, F.; Ma, J. Automated landslides detection for mountain cities using multi-temporal remote sensing imagery. Sensors 2018, 18, 821. Razzak , M.I.; Naz , S.; Zaib , A. Deep learning for medical image processing: Overview, challenges and the future. In Classification in BioApps ; Springer: Cham, Switzerland, 2018; pp. 323–350.  15-04-2024 KIRSET ’ 24 12

13. Acknowledgement We extend our heartfelt gratitude to all the experts in the field of medical imaging and deep learning whose valuable insights and contributions have greatly enriched this research. We also acknowledge the support and resources provided by Guru Kashi University, Talwandi Sabo, Bathinda , which have been instrumental in conducting this study. 15-04-2024 KIRSET ’ 24 13

14. Thank You 15-04-2024 KIRSET ’ 24 14