Image_Processing_-_Transfer_Learning_For_Dectection_of_Leukemia

Blood cancer is an uprising issue and doing physical medical procedures is too sensitive and time-consuming to detect any blast cell. Manual testing includes blood tests, spinal fluid tests, bone marrow tests, imaging tests, etc. A solution to this is to use modern methods in health care that help to detect diseases faster and increase the cure rate.ssing and Transfer Learning for Detection of Types of Leukemia: In image processing, data preparation and image preprocessing are done where we have rescaled the image and adjusted the brightness to improve the image quality. Data augmentation is performed to increase the image count by flipping it horizontally and vertically. Images are converted to grayscale to reduce the matrix calculation.The images in the dataset are: AML has 935 images, ALL has 858, CML has 623 and CLL has 510. Transfer learning is used. I have used different pre-trained CNN models such as ResNet-50, VGG16, Inception V3, and MobileNet for feature extraction and classification.VGG16, InceptionV3 and MobileNet - all three models give 100% accuracy, while ResNet50 gives 85% accuracy.