Purpose To evaluate pix2pix and CycleGAN and to assess the effects of multiple combination strategies on accuracy for patch‐based synthetic computed tomography (sCT) generation for magnetic resonance (MR)‐only treatment planning in head and neck (HN) cancer patients. Materials and methods Twenty‐three deformably registered pairs of CT and mDixon FFE MR datasets from HN cancer patients treated at our institution were retrospectively analyzed to evaluate patch‐based sCT accuracy via the pix2pix and CycleGAN models. To test effects of overlapping sCT patches on estimations, we (a) trained the models for three orthogonal views to observe the effects of spatial context, (b) we increased effective set size by using per‐epoch data augmentation, and (c) we evaluated the performance of three different approaches for combining overlapping Hounsfield unit (HU) estimations for varied patch overlap parameters. Twelve of twenty‐three cases corresponded to a curated dataset previously used for atlas‐based sCT generation and were used for training with leave‐two‐out cross‐validation. Eight cases were used for independent testing and included previously unseen image features such as fused vertebrae, a small protruding bone, and tumors large enough to deform normal body contours. We analyzed the impact of MR image preprocessing including histogram standardization and intensity clipping on sCT generation accuracy. Effects of mDixon contrast (in‐phase vs water) differences were tested with three additional cases. The sCT generation accuracy was evaluated using mean absolute error (MAE) and mean error (ME) in HU between the plan CT and sCT images. Dosimetric accuracy was evaluated for all clinically relevant structures in the independent testing set and digitally reconstructed radiographs (DRRs) were evaluated with respect to the plan CT images. Results The cross‐validated MAEs for the whole‐HN region using pix2pix and CycleGAN were 66.9 ± 7.3 vs 82.3 ± 6.4 HU, respectively. On the independent testing set with additional artifacts and previously unseen image features, whole‐HN region MAEs were 94.0 ± 10.6 and 102.9 ± 14.7 HU for pix2pix and CycleGAN, respectively. For patients with different tissue contrast (water mDixon MR images), the MAEs increased to 122.1 ± 6.3 and 132.8 ± 5.5 HU for pix2pix and CycleGAN, respectively. Our results suggest that combining overlapping sCT estimations at each voxel reduced both MAE and ME compared to single‐view non‐overlapping patch results. Absolute percent mean/max dose errors were 2% or less for the PTV and all clinically relevant structures in our independent testing set, including structures with image artifacts. Quantitative DRR comparison between planning CTs and sCTs showed agreement of bony region positions to <1 mm. Conclusions The dosimetric and MAE based accuracy, along with the similarity between DRRs from sCTs, indicate that pix2pix and CycleGAN are promising methods for MR‐only treatment planning for HN cancer. Our methods investigated for overlapping patch‐based HU estimations also ...

We developed an SVM-LASSO model to predict malignancy of PNs with two CT radiomic features. We demonstrated that the model achieved an accuracy of 84.6%, which was 12.4% higher than Lung-RADS.

We proposed a framework to detect and quantify local tumor morphological changes due to chemo-radiotherapy (CRT) using Jacobian map and to extract quantitative radiomic features from the Jacobian map to predict the pathologic tumor response in locally advanced esophageal cancer patients. In 20 patients who underwent CRT, a multi-resolution BSpline deformable registration was performed to register the follow-up (post-CRT) CT to the baseline CT image. Jacobian map (J) was computed as the determinant of the gradient of the Deformation Vector Field. Jacobian map measured the ratio of local tumor volume change where J < 1 indicated tumor shrinkage and J > 1 denoted expansion. The tumor was manually delineated and corresponding anatomical landmarks were generated on the baseline and follow-up images. Intensity, texture and geometry features were then extracted from the Jacobian map of the tumor to quantify tumor morphological changes. The importance of each Jacobian feature in predicting pathologic tumor response was evaluated by both univariate and multivariate analysis. We constructed a multivariate prediction model by using a support vector machine (SVM) classifier coupled with a least absolute shrinkage and selection operator (LASSO) for feature selection. The SVM-LASSO model was evaluated using ten-times repeated 10-fold cross-validation (10×10-fold CV). After registration, the average Target Registration Error was 4.30±1.09mm (LR:1.63mm AP:1.59mm SI:3.05mm) indicating registration error was within two voxels and close to 4mm slice thickness. Visually, Jacobian map showed smoothly-varying local shrinkage and expansion regions in a tumor. Quantitatively, the average Median Jacobian was 0.80±0.10 and 1.05±0.15 for responder and non-responder tumors, respectively. These indicated that on average responder tumors had 20% median volume shrinkage while non-responder tumors had 5% median volume expansion. In univariate analysis, Minimum Jacobian (p=0.009, AUC=0.98) and Median Jacobian (p=0.004, AUC=0.95) were the most significant predictors. The SVM-LASSO model achieved the highest accuracy when these two features were selected (Sensitivity=94.4%, Specificity=91.8%, AUC=0.94). Novel features extracted from the Jacobian map quantified local tumor morphological changes using only baseline tumor contour without post-treatment tumor segmentation. The SVM-LASSO model using Median Jacobian and Minimum Jacobian achieved high accuracy in predicting pathologic tumor response. Jacobian map showed great potential for longitudinal evaluation of tumor response.

Purpose To predict the spatial and temporal trajectories of lung tumor during radiotherapy monitored under a longitudinal magnetic resonance imaging (MRI) study via a deep learning algorithm for facilitating adaptive radiotherapy (ART). Methods We monitored 10 lung cancer patients by acquiring weekly MRI‐T2w scans over a course of radiotherapy. Under an ART workflow, we developed a predictive neural network (P‐net) to predict the spatial distributions of tumors in the coming weeks utilizing images acquired earlier in the course. The three‐step P‐net consisted of a convolutional neural network to extract relevant features of the tumor and its environment, followed by a recurrence neural network constructed with gated recurrent units to analyze trajectories of tumor evolution in response to radiotherapy, and finally an attention model to weight the importance of weekly observations and produce the predictions. The performance of P‐net was measured with Dice and root mean square surface distance (RMSSD) between the algorithm‐predicted and experts‐contoured tumors under a leave‐one‐out scheme. Results Tumor shrinkage was 60% ± 27% (mean ± standard deviation) by the end of radiotherapy across nine patients. Using images from the first three weeks, P‐net predicted tumors on future weeks (4, 5, 6) with a Dice and RMSSD of (0.78 ± 0.22, 0.69 ± 0.24, 0.69 ± 0.26), and (2.1 ± 1.1 mm, 2.3 ± 0.8 mm, 2.6 ± 1.4 mm), respectively. Conclusion The proposed deep learning algorithm can capture and predict spatial and temporal patterns of tumor regression in a longitudinal imaging study. It closely follows the clinical workflow, and could facilitate the decision‐making of ART. A prospective study including more patients is warranted.

Quantification of local metabolic tumor volume (MTV) changes after Chemo-radiotherapy would allow accurate tumor response evaluation. Currently, local MTV changes in esophageal (soft-tissue) cancer are measured by registering follow-up PET to baseline PET using the same transformation obtained by deformable registration of follow-up CT to baseline CT. Such approach is suboptimal because PET and CT capture fundamentally different properties (metabolic vs. anatomy) of a tumor. In this work we combined PET and CT images into a single blended PET-CT image and registered follow-up blended PET-CT image to baseline blended PET-CT image. B-spline regularized diffeomorphic registration was used to characterize the large MTV shrinkage. Jacobian of the resulting transformation was computed to measure the local MTV changes. Radiomic features (intensity and texture) were then extracted from the Jacobian map to predict pathologic tumor response. Local MTV changes calculated using blended PET-CT registration achieved the highest correlation with ground truth segmentation (R=0.88) compared to PET-PET (R=0.80) and CT-CT (R=0.67) registrations. Moreover, using blended PET-CT registration, the multivariate prediction model achieved the highest accuracy with only one Jacobian co-occurrence texture feature (accuracy=82.3%). This novel framework can replace the conventional approach that applies CT-CT transformation to the PET data for longitudinal evaluation of tumor response.