|
|
Deep learning model for segmentation of trabecular tissue on CT data of the lumbar spine
Nagyová, Miriam ; Nohel, Michal
This paper focuses on training a deep learning model for vertebral body segmentation of the lumbar spine. The nnU-Net model was trained and tested on a publicly available dataset LumVBCanSeg consisting of 185 lumbar CT scans. Dice coefficient was used to evaluate the accuracy of the trained model. The mean Dice coefficient of the testing dataset was 0.949 with a standard deviation of 0.103. The model was also tested on clinical data containing various abnormalities, such as lytic lesions in multiple myeloma patients and metallic implants. Results were evaluated visually. While the model showed high accuracy on the testing dataset, the results on scans with anomalies showed a decline in accuracy.
|
|
|
Implementation of a deep learning model for segmentation of multiple myeloma in CT data
Gálík, Pavel ; Nohel, Michal
This paper deals with the implementation of a deep learning model for spinal tumor segmentation of multiple myeloma patients in CT data. Deep learning is becoming an important part of developing computer-aided detection and diagnosis systems. In this study, a database of 25 patients who were imaged on spectral CT and for whom different parametric images (conventional CT, virtual monoenergetic images, calcium suppression images) were reconstructed, was used. Three convolutional neural network models based on the nnU-Net framework for lytic lesion segmentation were trained on the selected data. The results were evaluated on a test database and the trained models were compared.
|
|
| |
|
|
Optic nerve head segmentation in retinal image data
Nohel, Michal ; Drahanský, Martin (referee) ; Kolář, Radim (advisor)
This diploma thesis deals with the segmentation of the optic disc and cup in retinal image data. The theoretical part of the thesis describes the optic disc and cup and provides an overview of the current state of the art in using machine learning methods for their segmentation. Furthermore, the basic principles and blocks of convolutional neural networks are described. Convolutional neural networks U-Net and its modification nnU-Net were trained on the created databases. These models were tested and the results obtained were discussed and compared with selected published methods. Finally, the models were evaluated in terms of their potential for practical application.
|
guest ::
