Hematologic Cancer Detection Using White Blood Cancerous Cells Empowered with Transfer Learning and Image Processing.

Area of Science:

• Oncology
• Medical Imaging
• Artificial Intelligence

Background:

• Lymphoma and leukemia are fatal blood cancers with high mortality rates.
• Early detection of blood cancer is crucial for improving patient survival.
• Manual analysis of white blood cell images for cancer prediction is time-consuming and prone to errors.

Purpose of the Study:

• To develop an accurate and efficient deep learning model for early blood cancer prediction.
• To enhance the prediction accuracy of cancerous white blood cells using image processing and transfer learning.
• To overcome limitations of existing manual and machine learning methods for blood cancer diagnosis.

Main Methods:

• A deep learning model integrating transfer learning and image processing techniques was developed.
• Multiple transfer learning models (AlexNet, MobileNet, ResNet) were evaluated with varying parameters.
• Cloud techniques were used to select the optimal prediction model.
• Performance was assessed using various metrics, including accuracy and misclassification rate.

Main Results:

• The proposed model, specifically AlexNet with stochastic gradient descent momentum and image processing, achieved a prediction accuracy of 97.3%.
• The misclassification rate for the best-performing model was 2.7%.
• The model demonstrated superior performance compared to models without image processing.

Conclusions:

• The developed deep learning model shows significant promise for the smart diagnosis of blood cancer.
• The integration of transfer learning and image processing enhances the accuracy of white blood cell analysis.
• This approach can aid in the early and accurate detection of conditions like lymphoma and leukemia.