| CT image 2D | 63,890 patients with cancer and 171,345 healthy | Discriminate lung cancer lesions in adenocarcinoma, squamous and small cell carcinoma | CNN | Log-Loss error of 0.66 with a sensitivity of 0.87 | Oncology | [118] |
CT 2D images | 3059 images from several parts of human body | Speed up CT images collection and rebuild the data | Dense Net and a deconvolution model | RMSE of 0.00048 | Various | [11] | |
CT images 3D | 6960 lung nodule regions, 3480 of which were positive samples and rest were negative samples (nonnodule) | CAD to diagnose lung cancer in low-dosage computed tomography | Eye tracking sparse attentional model and convolutional neural network | Accuracy of 0.97 | Oncology | [162] | |
CT images 2D and text (reports) | 9000 training and 1000 testing images | Processing text from CT reports in order to classify their respective images | CNN | Accuracy of 0.95, 0.70 and 0.58 respectively for the three use cases | Various | [12] | |
Computed tomography (CT) | Three datasets: 224,316, 112,120 and 15,783 | Binary classification of posteroanterior chest xray | CNN | 92% accuracy | Radiology | [2] [163] | |
MRI | Diffusion- weighted imaging maps using MRI | 222 patients. 187 treated with rtPA (recombinant tissue-type plasminogen activator) | Decide Acute Ischemic Stroke patients’ treatment through volume lesions prediction | CNN | AUC of 0.88 | Neurology- Psychiatry | [2] ; [81] |
Magnetic resonance images | 474 patients with schizophrenia and 607 healthy subjects | Schizophrenia detection | Deep discriminant autoencoder network | Accuracy over 0.8 | Neurology- Psychiatry | [83] | |
Gadoxetic acid-enhanced 2D MRI | 144,180 images from 634 patients | Staging liver fibrosis through MR | CNN | AUC values of 0.84, 0.84, and 0.85 for each stage | Gastroenterology | [86] | |
Resting state functional magnetic resonance imaging (rs-fMRI), T1 structural cerebral images and phenotypic information | 505 individuals with autism and 520 matched typical controls | Identify different autism spectrum disorders | Denoising AE | Accuracy of 0.70 | Neurology- Psychiatry | [92] | |
3D MRI and PET | 93 Alzheimer Disease, 204 MCI Mild Cognitive Impairment converters and normal control subjects | CAD for early Alzheimer disease stages | Multimodal DBM | Accuracy of 0.95, 0.85 and 0.75 for the three use cases | Neurology- Psychiatry | [2] ; [93] |