Skip to content

Latest commit

 

History

History
58 lines (47 loc) · 2.95 KB

README.md

File metadata and controls

58 lines (47 loc) · 2.95 KB

PNEUNOMIA-DETECTION-USING-CNN

An automated system using machine learning techniques to accurately and efficiently detect pneumonia and other lung diseases(Tuberculosiand Covid-19) from chest X-ray images, enabling faster and more reliable diagnoses for timely medical intervention.

METHODOLOGY

  • Gather lung disease datasets and select the best available dataset for each class.
  • The images of the gathered dataset are preprocessed, rescaled, and augmented as a part of preprocessing.
  • The system is trained using the newly compiled dataset of labeled chest X-ray images, where each image is annotated into pneumonia, normal, tuberculosis, or COVID-19.
  • Adjustments are made to the model based on the analysis post-training.
  • After training, a CXR image image is fed into the network to generate a prediction indicating the likelihood of lung disease.

RESULTS

  • After training the model for 100 epochs, the model achieved a training loss of 0.0063 and a training accuracy of 99.90 percent.
  • The validation loss at the end of the training was 0.0081, with a validation accuracy of 99.84 percentage.
  • After evaluating the model on the test dataset, it achieved a test loss of 0.0241 and test accuracy of 99.36 percentage.

CONCLUSION

  • The developed model demonstrated promising accuracy in classifying chest X-ray images of Pneumonia, Normal, Tuberculosis, and Covid using Deep learning techniques, indicating its potential for accurate diagnosis of various chest conditions.
  • Convolutional neural networks (CNNs) proved effective in extracting relevant features and patterns from the X-ray images, contributing to the model’s successful classification performance.
  • Accurate classification of chest X-ray images has the potential to significantly assist in early detection and diagnosis, ultimately benefiting patient outcomes.
  • This project highlights the valuable role of Deep learning in automating image analysis and aiding decision-making in the medical field. However, certain limitations, such as dataset constraints, and opportunities for further model enhancements should be considered for future research and development.

Project v2 (PNCT)

1. v2 Dataset 5.1 (TB replaced with Belarus) Link: https://drive.google.com/drive/folders/12eadNrEtuxmn6bkytmNitqG8-IAbePrL?usp=drive_link

2. v2 Dataset 5.0 (Imbalanced) Link: https://drive.google.com/drive/folders/1ARY8yYrXgVA1MKbKmQwXyG0KVgSPl9TY?usp=drive_link

3. v2 Dataset 5.1.1 (replaced with COVID-19 Radiography Dataset) Link: https://drive.google.com/drive/folders/1Xq-tITMcUxYRsEjxw939XPKeGRs-661e?usp=drive_link

4. Shenzen Dataset Link: https://drive.google.com/drive/folders/1HTaWkGav3cvYTn4ebUOD9ymlfcL1I3Su?usp=drive_link

MODELS

1. Model of Jv2.12 [99.51% acc] Dataset 5.1.1, Arch 1, 100 epoch, 16 batch size.h5: https://drive.google.com/file/d/10fZwFbmg-WL52h6_JxK6O25m0iDFw9Bw/view?usp=sharing