| Popis: |
Osteosarcoma is the most frequent primary malignant bone tumour. Computer-aided detection (CAD) and diagnosis are being used to enhance osteosarcoma detection and diagnosis . The use of machine learning and deep learning algorithms may save up surgeons' time while also improving patient outcomes. An enormous quantity of data must be fed into the classifier for it to become more accurate. Adapted to a public dataset of osteosarcoma histology pictures, a mix of machine and deep learning is used in this work to distinguish between necrotic and healthy tissue images. First, the dataset was preprocessed, and contour based threshold segmentation techniques are applied. Then, Stochastic linear embedding based Feature extraction is used for extracting the abnormal features. Finally, the proposed multilayer grid XG Boost classifier is trained on stained images in order to increase the output accuracy. The experimental findings indicate that the proposed classifier has the greatest accuracy of any illness classification approach currently available. Our finetuned model showed superior performance in identifying osteosarcoma malignancy using H and E stained pictures. |
