Aims and objectives
This study sought to develop an AI approach to accurately and efficiently differentiate primary intracranial tumors.
Artificial Intelligence (AI) offers a way to aid clinicians in diagnostic accuracy.
AI machine learning algorithms for Computer Assisted Diagnosis are software packages that intake labeled images of different classes (i.e.
cancerous images and healthy images).
AI uses learning theories and training algorithms which mimic techniques used in clinical training.
Convolutional Neural Networks (CNN's) are a theory found in machine learning that entail a feed forward deep class approach...
Methods and materials
MRI images of ten primary CNS tumors were obtained from the Radiopaedia (copyright) nervous system library. Single 3T MRI images were chosen to train tumor identification.
The system analyzed 2190 images depicting Craniopharyngioma,
Ependymoma,
Ganglioglioma,
Medullablastoma,
Meningioma,
Pylocytic Astrocytoma,
Acoustic Neuroma,
and Schwannoma.
An AI system using a Convolutional Neural Network (CNN) was built through Python scripting language to accurately identify specific imaging characteristics.
Accuracy efficiency was determined in seconds per slice.
Using this data,
an AI system using a CNN was built through Python...
Results
Using a neural network AI we classified 8 primary CNS tumors with an overall validation accuracy of 88.05%.
Individual binary accuracy (healthy vs specific pathology) ranged from 95.5 – 99.8%.
Pylocytic Astrocytoma was most readily detected at 99.8%.
The AI classifies images at a rate of 2.28 seconds/slice,
and automatically iterates through each slice of a volume.
Conclusion
CNN offers a novel way to accurately distinguish primary CNS tumors using MRI images.
This may have important clinical applications for non-invasive CNS tumor detection,
diagnosis and assessment,
as well as differentiation of tumor sub-types,
facilitating optimized treatment decisions and optimizing treatment planning for the individual patient.
AI algorithms can be trained to detect individual tumor characteristics,
which will allow clinicians to identify essential aspects of the tumor and its individual behavior prior to treatment and post treatment for disease monitoring
References
Abadi,
M.,
Agarwal,
A.,
Barham,
P.,
Brevdo,
E.,
Chen,
Z.,
Citro,
C.,
...
&
Ghemawat,
S.
(2016).
Tensorflow: Large-scale machine learning on heterogeneous distributed systems.
arXiv preprint arXiv:1603.04467.
Dalmis,
M.
U.,
Litjens,
G.,
Holland,
K.,
Setio,
A.,
Mann,
R.,
Karssemeijer,
N.,
& Gubern-Nerida,
A.;(2017).
Using deep learning to
segment breast and fibroglandular tissue in MRI volumes.
Medical
Physics,
533-546.
Kleesiek,
J.,
Biller,
A.,
Urban,
G.,
Kothe,
U.,
Bendszus,
M.,
&
Hamprecht,
F.
A.
(2014).
ilastik for Multi-modal Brain Tumor
Segmentation.
Proceedings MICCAI BraTS (Brain Tumor...