Detecting liver cancer from MRI using deep learning

Patients' chances of survival can be increased by early detection and treatment of liver cancer. The most comprehensive information for the differential diagnosis of liver cancers comes from dynamic contrast-enhanced MRI. However, as MRI diagnosis is influenced by personal experience, deep learning might offer a fresh approach to diagnosis. In order to categorize liver cancers based on improved MR images, unenhanced MR images, clinical data including language and laboratory test results, we employed convolutional neural networks (CNNs) to create a deep learning system (DLS). The solution that has been suggested makes use of AlexNet in conjunction with the MATLAB software and a dataset image including 1030 malignant and 868 Benign. When applied to the dataset, the proposed model achieved the following results: 98.3 % sensitivity, 98.2 % specificity, 98.2 %accuracy, 98.7 % sub-curve area (AUC). The model's performance was similar to that of doctors' diagnoses when measured across a total of 1898 images. In the not-too-distant future, as additional data is given to the model, it will grow more accurate and precise, ultimately resulting in the maximum potential detection rate.