AccuLearning
Radiotherapy AI model training platform
OAR/CTV/GTV auto-delineation models on multi-modality images;
dose prediction models
Manteia AccuLearningTM AI model training platform provides
visualized solutions for radiotherapy AI applications. High
efficiency small sampling training, automatic data cleansing,
multi-parameter and multi-layer adjustable neural network
architecture, no-coding workflow and visualized model evaluation,
all these features are guaranteed to bring significant values to
clinical and research work by achieving high quality
auto-segmentation and dose prediction AI models.
Major Modules：
OAR/CTV/GTV auto-delineation models on multi-modality images; dose
prediction models
[
Workflow
]
1
Dataset creation
selected data will be cleaned and imported as training data
set
2
Model training and evaluation
one-click model training using default neural network
settings, automatic model evaluation using testing data set
3
Model application
trained model will be exported to application software, e.g.
AccuContour, to be used as the new delineation model
Advantages
Small-sampling data set
satisfactory result can be achieved from as few as 30 data
sets
No-coding operation
users do not need coding knowledge to program neural network.
Visualized parameter tuning helps adjusting model training
algorithms
Accurate models
automatically trained models using default settings are as
much accurate as manually trained models
Advantages
Data safety
importing data sets and exporting models within local network
to avoid data breaching
Open architecture
connect with all Manteia products for data importing and model
exporting
Translational research support
improve clinical practice with technologies and translational
research