Automatically classify Brain MRI series by pulse sequence types: FLAIR, T1C, T2, ADC, DWI, TOF and OTHER
Project description
Brain MRI Pulse Sequences Classification
About Project
This is official code of our package brainmri_ps. We provide a machine learning based tool to automatically classify Brain MRI series into different pulse sequence types:
- FLAIR
- T1C
- T2
- ADC
- DWI
- TOF
- OTHER
Installation
Install via pip:
pip install brainmri_ps
Usage
Load pretrained models:
from brainmri_ps import PulseSequenceClassifier
classifier = PulseSequenceClassifier("mobilenet_v2").from_pretrained()
Name | Input Resolution | #Params (M) | MACs (G) | Test Accuracy | Pretrained |
---|---|---|---|---|---|
MobileNet V2 | 256 | 2.23 | 0.42 | 100.0 | ✓ |
Example - predict from a study:
In : classifier.predict_study("*/1.2.840.113619.6.388.6361536015762131135133837693432843617")
Out :
{
"1.2.840.113619.2.5.1821162425615901145251590114525252000": "ADC",
"1.2.840.113619.2.388.57473.14165493.12954.1590103413.819": "T2",
"1.2.840.113619.2.388.57473.14165493.12954.1590103413.822": "DWI",
"1.2.840.113619.2.388.57473.14165493.12954.1590103413.823": "T1C",
"1.2.840.113619.2.388.57473.14165493.12954.1590103413.821": "FLAIR"
}
Function predict_study
does the following steps:
- Read all dicom files in a study folder and group them into series by SeriesInstanceUID field
- Determine the orientation plane (axial, sagittal, coronal) of the series by using the ImageOrientationPatient field
- Predict and return the pulse sequence types of axial series (ignore the non-axial ones)
Contact
Issues should be raised directly in the repository. For further support please email us at:
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
brainmri_ps-1.1.1.tar.gz
(5.9 kB
view hashes)
Built Distribution
Close
Hashes for brainmri_ps-1.1.1-py2.py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | eea8eac0b18f11579e9092d84d359f1d5d373b036e8d98c9b72fc514d552e33a |
|
MD5 | 19ae4dd290914991106f14e9bd01013d |
|
BLAKE2b-256 | 576b58a4dbfb82f8715228d860cb9517408314f73b15521addd41021d7e0797c |