Leveraging Convolutional Neural Networks for Classifying Lymphoma Using Histopathological Images

The lymphatic system serves as the origin of lymphoma, a cancerous condition in which lymphocytes undergo a malignant transformation and begin to divide uncontrollably. As the medical experts need more time to determine the type of lymphoma based on the histopathologic images. We used CNNs for the classification of lymphomas. In this work, we used the Multi Cancer Dataset for image classification. The Multi Cancer Dataset contains lymphoma images categorized into 3 classes and contains 15000 images with a size of
$\mathbf{512}\times \mathbf{512}$
pixels. Here we used a customized pre trained model to classify the images. We achieved accuracy of 96.2% and sensitivity of 96.2% in classifying lymphomas on histopathology images using the customized pre trained model. The preliminary results show how CNNs can be used to achieve a strong and correct classification of lymphomas, making them quite useful as an aid in histopathologic diagnosis.