China
Africa
Explore chapters and articles related to this topic
Machine Learning Applications In Medical Image Processing
Published in Sanjay Saxena, Sudip Paul, High-Performance Medical Image Processing, 2022
Tanmay Nath, Martin A. Lindquist
The imaging data is stored slice-wise and contains large amounts of meta-data, including patient information and information about image acquisition settings. Usually, there are multiple DICOM files for each data acquisition and therefore, it is necessary to convert to other file format like NIFTI (neuroimaging informatics technology initiative) for data analysis. Brain imaging data structure (BIDS) [48] is a standard for organizing and describing the collected MRI dataset. It helps to organize DICOM images in a standard directory structure and describe the data collected during the experiment, making it easier to share the data and associated analysis pipelines. Recently, BIDS format has been extended to organize magneto-encephalography (MEG) [49], intracranial electroencephalography (iEEG) [50] and electroencephalography (EEG) [51] dataset.
Brain Imaging Data
Published in Atsushi Kawaguchi, Multivariate Analysis for Neuroimaging Data, 2021
There are several formats for storing images measured by MRI as electronic files on a computer. The digital imaging and communication in medicine (digital imaging and communications in medicine: DICOM) format is common for medical images. By saving images captured by different devices in this format, the user can see images without being conscious of the difference in the devices used. “Analyze” is a format consisting of a data file (.img) that records pixel information and a header file (.hdr) that records meta information of the origin information as well as the size of the image. However, in the Analyze format, it was not possible to record the relation between the voxel address and the spatial information of the MRI machine. This limitation is overcome in the neuroimaging informatics technology initiative (NIfTI) format, an extension of the Analyze format that has become mainstream in recent years. As with Analyze, NIfTI can store data files and header files separately, and these two files can be saved together (.nii).
A 3D image segmentation for lung cancer using V.Net architecture based deep convolutional networks
Published in Journal of Medical Engineering & Technology, 2021
Kamel K. Mohammed, Aboul Ella Hassanien, Heba M. Afify
In total, 3D CT scans for lung cancer were gathered from patients with NSCLC at Stanford University, Stanford, CA, USA [28]. These lung images are called Task06_Lung that are available from TCIA and previously used to produce a radiogenic signature [29]. The area of the tumour was identified by a specialist thoracic radiologist on a cross-section of CT using OsiriX software [30]. Figure 2 describes the labelled image with blue colour for the tumour region. Each image has a variable size from 512 × 512 × 100 to 512 × 512 × 600 pixels. The database is one of the 2018 Medical Segmentation Decathlon challenges [31] that comprised 96 images including 64 training images and 32 testing images. All images were reformatted from standard DICOM to Neuroimaging Informatics Technology Initiative (NIfTI) format created by the National Institutes of Health [32]. The NIfTI file held the 3D image matrix and diverse metadata. Figure 3 shows a 3D image for volumetric measurements of lung cancer across three coordinates.