Automatic lesion detection and segmentation of 18F-FET PET in gliomas: A full 3D U-Net convolutional neural network study

• Published in: PloS one 2018-04, Vol.13 (4), p.e0195798-e0195798
• Main Authors: Blanc-Durand, Paul, Van Der Gucht, Axel, Schaefer, Niklaus, Itti, Emmanuel, Prior, John O
• Format: Article
• Language: English
• Subjects: Amino acids; Artificial intelligence; Artificial neural networks; Automation; Biology and Life Sciences; Biopsy; Brain; Brain cancer; Computer and Information Sciences; Computer vision; Decoding; Diagnostic systems; Differential diagnosis; Emission analysis; Feasibility studies; Fluorine isotopes; Glioma; Glioma - diagnostic imaging; Glioma - pathology; Ground truth; Hospitals; Humans; Image acquisition; Image processing; Image Processing, Computer-Assisted; Image segmentation; Imaging, Three-Dimensional; International conferences; Learning algorithms; Lesions; Machine learning; Medical diagnosis; Medical imaging; Medicine and Health Sciences; Neural networks; Neural Networks (Computer); Neuroimaging; Nuclear medicine; Patients; Positron emission; Positron emission tomography; Positron-Emission Tomography - methods; Radiation therapy; Reproducibility of Results; Research and Analysis Methods; Sensitivity; Sensitivity and Specificity; Tomography; Training; Tumors; Tyrosine; Tyrosine - analogs & derivatives; Volumetric analysis
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0195798

Amino-acids positron emission tomography (PET) is increasingly used in the diagnostic workup of patients with gliomas, including differential diagnosis, evaluation of tumor extension, treatment planning and follow-up. Recently, progresses of computer vision and machine learning have been translated for medical imaging. Aim was to demonstrate the feasibility of an automated 18F-fluoro-ethyl-tyrosine (18F-FET) PET lesion detection and segmentation relying on a full 3D U-Net Convolutional Neural Network (CNN). All dynamic 18F-FET PET brain image volumes were temporally realigned to the first dynamic acquisition, coregistered and spatially normalized onto the Montreal Neurological Institute template. Ground truth segmentations were obtained using manual delineation and thresholding (1.3 x background). The volumetric CNN was implemented based on a modified Keras implementation of a U-Net library with 3 layers for the encoding and decoding paths. Dice similarity coefficient (DSC) was used as an accuracy measure of segmentation. Thirty-seven patients were included (26 [70%] in the training set and 11 [30%] in the validation set). All 11 lesions were accurately detected with no false positive, resulting in a sensitivity and a specificity for the detection at the tumor level of 100%. After 150 epochs, DSC reached 0.7924 in the training set and 0.7911 in the validation set. After morphological dilatation and fixed thresholding of the predicted U-Net mask a substantial improvement of the DSC to 0.8231 (+ 4.1%) was noted. At the voxel level, this segmentation led to a 0.88 sensitivity [95% CI, 87.1 to, 88.2%] a 0.99 specificity [99.9 to 99.9%], a 0.78 positive predictive value: [76.9 to 78.3%], and a 0.99 negative predictive value [99.9 to 99.9%]. With relatively high performance, it was proposed the first full 3D automated procedure for segmentation of 18F-FET PET brain images of patients with different gliomas using a U-Net CNN architecture.