The classification and detection of x-ray cancer images
Principal Investigator
Name
Huang Xiansong
Degrees
M.S.
Institution
Department of AI Research, Peng Cheng Laboratory
Position Title
Engineer of Department of Medical Image analysis
Email
About this CDAS Project
Study
PLCO
(Learn more about this study)
Project ID
PLCO-439
Initial CDAS Request Approval
Jan 24, 2019
Title
The classification and detection of x-ray cancer images
Summary
Pathology images can be used for localizing the cancerous area automatically with the method of weekly supervised instance segmentation. To be more specific, it will be realized by exploiting class peak responses to enable a classification network for instance mask extraction. We observed that local maximums, i.e., peaks, in a class response map typically correspond to strong visual cues residing inside each instance. We design a process to stimulate peaks to emerge from a class response map. The emerged peaks are then back-propagated and effectively mapped to highly informative regions of each object instance, such as instance boundaries. We will make ues of this method to circle regions of interest on pathology images automatically, which will facilitate auto-annotation on pathology images.
Besides, the classification of chest X-ray Images will be realized by the supervised deep learning model such as ResNet-101.
Besides, the classification of chest X-ray Images will be realized by the supervised deep learning model such as ResNet-101.
Aims
Pathology images are widely used for diagnosing cancer in clinical application. With the help of deep learning, the cancer area can be detected and segmented from pathology, which can reduce the workload of doctors. The annotation datasets are needed if the supervised learning is applied, which is more accurate compared with unsupervised way. However, the manual annotation on pathology images is really a huge project. Therefore, it is pretty worthwhile of desegning an algorithm to realize auto-annotation on pathology images.
Besides, we want to realize the lung cancer detection from chest X-ray Images with useful deep learning models.
Collaborators
Jie Chen, University of Oulu, Finland