Purpose Task automation is essential for efficient and consistent image segmentation in radiation oncology. We report on a deep learning architecture, comprising a U‐Net and a variational autoencoder (VAE) for automatic contouring of the prostate gland incorporating interobserver variation for radiotherapy treatment planning. The U‐Net/VAE generates an ensemble set of segmentations for each image CT slice. A novel outlier mitigation (OM) technique was implemented to enhance the model segmentation accuracy. Methods The primary source dataset (source_prim) consisted of 19 200 CT slices (from 300 patient planning CT image datasets) with manually contoured prostate glands. A smaller secondary source dataset (source_sec) comprised 640 CT slices (from 10 patient CT datasets), where prostate glands were segmented by 5 independent physicians on each dataset to account for interobserver variability. Data augmentation via random rotation (<5 degrees), cropping, and horizontal flipping was applied to each dataset to increase sample size by a factor of 100. A probabilistic hierarchical U‐Net with VAE was implemented and pretrained using the augmented source_prim dataset for 30 epochs. Model parameters of the U‐Net/VAE were fine‐tuned using the augmented source_sec dataset for 100 epochs. After the first round of training, outlier contours in the training dataset were automatically detected and replaced by the most accurate contours (based on Dice similarity coefficient, DSC) generated by the model. The U‐Net/OM‐VAE was retrained using the revised training dataset. Metrics for comparison included DSC, Hausdorff distance (HD, mm), normalized cross‐correlation (NCC) coefficient, and center‐of‐mass (COM) distance (mm). Results Results for U‐Net/OM‐VAE with outliers replaced in the training dataset versus U‐Net/VAE without OM were as follows: DSC = 0.82 ± 0.01 versus 0.80 ± 0.02 (p = 0.019), HD = 9.18 ± 1.22 versus 10.18 ± 1.35 mm (p = 0.043), NCC = 0.59 ± 0.07 versus 0.62 ± 0.06, and COM = 3.36 ± 0.81 versus 4.77 ± 0.96 mm over the average of 15 contours. For the average of 15 highest accuracy contours, values were as follows: DSC = 0.90 ± 0.02 versus 0.85 ± 0.02, HD = 5.47 ± 0.02 versus 7.54 ± 1.36 mm, and COM = 1.03 ± 0.58 versus 1.46 ± 0.68 mm (p < 0.03 for all metrics). Results for the U‐Net/OM‐VAE with outliers removed were as follows: DSC = 0.78 ± 0.01, HD = 10.65 ± 1.95 mm, NCC = 0.46 ± 0.10, COM = 4.17 ± 0.79 mm for the average of 15 contours, and DSC = 0.88 ± 0.02, HD = 7.00 ± 1.17 mm, COM = 1.58 ± 0.63 mm for the average of 15 highest accuracy contours. All metrics for U‐Net/VAE trained on the source_prim and source_sec datasets via pretraining, followed by fine‐tuning, show statistically significant improvement over that trained on the source_sec dataset only. Finally, all metrics for U‐Net/VAE with or without OM showed statistically significant improvement over those for the standard U‐Net. Conclusions A VAE combined with a hierarchical U‐Net and an OM strategy (U‐Net/OM‐VAE) demonstrates promise toward capt...

Introduction: Research studies have already shown that tumor aggressiveness and response to chemical and radiation therapies are influenced by the extravascular extracellular space (VEES) of the tumor microenvironment. Assessment of VEES has been reported to be fundamental to understanding tumor response to treatment and probability of recurrence. Purpose: This pilot study investigates the association between wavelet-based radiomic features extracted from dynamic contrast-enhanced magnetic resonance images (DCE-MRI) of rat brain tumors against VEES estimated by pharmacokinetic modeling. Methods: Eight immune-compromised-RNU/RNU rats were implanted with human U251n cancer cells to form an orthotopic glioma (IACUC #1509). For each rat, two DCE-MRI studies (multi slice/echo GE, 3 slice(2mm),128x64, FOV:32x32mm2, TR/(TE1-TE2)=24ms/(2ms-4ms), flip angle=18º, 400 acquisitions, 1.55 sec interval, Magnevist was injected at acquisition no. 15) were performed (24h apart) using a 7T Varian (Agilent, 20cm bore) scanner. A single 20Gy stereotactic radiation exposure was performed before the second study. The post treatment MRIs were taken a range of 1-6.5 hrs post radiation. The time trace of relaxivity change (ΔR1) in all the voxels of the animal’s brain for all studies were calculated. Wavelet decomposition analysis was performed on the ΔR1 for each voxel and frequency-based localized approximations with 4 degrees of regularities were estimated. The VEES map was estimated from ΔR1 by the pharmacokinetic (modified Toft’s) model and a nested model selection technique. Finally, the Pearson correlation coefficients between the VEES map and corresponding wavelet coefficient maps in the tumor region were calculated. Results: The average voxel-wise Pearson correlation coefficients between the VEES maps (averaged for all animals) and their corresponding wavelet-based, radiomics coefficient maps were: r= -0.680, r= -0.802, r= -0.813, and r= -0.791 with p&lt;0.0001 for the 4 wavelet coefficients (from higher to lower frequencies), respectively. Discussion & Conclusion: This pilot study suggests that wavelet based radiomic analysis has potential to provide information pertinent to the tumor microenvironment, which correlates well with pharmacokinetic modeling. As such, this work represents an important first step toward potentially connecting radiomics with underlying biological mechanisms. Citation Format: Hassan Bagher-Ebadian, Stephen L. Brown, Olivia Valadie, Julian A. Rey, Ning Wen, Malisa Sarntinoranont, James R. Ewing, Indrin J. Chetty. Characterization of extravascular extracellular space of rat brain tumors using wavelet-based radiomics analysis of dynamic contrast enhanced MRI [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 543.

No abstract

No abstract